Lara thinking to herself: "I've never seen this button before. Is it a good idea to press it?" Lara speaking: "No." Lara thinking to herself: "Alright."
This is not real. To clarify, I'm not questioning the feasibility of AI playing a video game - that's undeniably achievable. My focus is on this particular video itself. He has managed to create a surface-level semblance of plausibility, and credit is deserved for that. However, there are many reasons why the video is fabricated. - He never displays any code or functional software, beyond ChatGPT + Google Lens for dialogue. He frequently mentions 'The Program' he employed or created, yet what 'that' actually really is, we never find out. (Interestingly, he noted in a comment that he uses ChatGPT to assist in crafting his videos due to speaking French. It wouldn't be surprising if he also sought GPT's input to enhance the AI concept's credibility.) - A significant inconsistency that Fox seems to have overlooked is that he himself claims that Lara's voice is purely for narration, so separate from the AI controlling gameplay. Nonetheless, several instances in the final product contradict this claim. For example, Lara discussing her endless magazines and proceeding to test the concept, or her query about gravity followed by a confirming jump. Another instance occurs at 4:36, with Lara counting 1-2-3 and the AI synchronously bouncing. - He demonstrates Lara's movement being driven by the robots exploration of 'new textures'. However, he doesn't delve significantly further into this concept, lacking depth beyond this basic explanation. There should be way more going on besides just simply that to get a 'robot' to solve a Tomb Raider level. - Moreover, there's an odd visual choice of showing a webpage about a smoke alarm-detector using Arduino while discussing audio recognition (seemingly stemming from a Google search for 'audio recognition', and using the first result). Here, he again doesn't detail the exact software he employs, except for likening it to TH-cam copyright detection. The only thing that chapter _really_ shows is a dictionary page featuring the definition of 'ouch' and some waveforms in Audacity (simply audio recording software). When comparing Fox's videos to other genuinely authentic showcases on the same subject, the distinction in depth becomes very evident. Other creators delve much more extensively into their topics, despite their projects being simpler and smaller in scope.
@basicallyhuman You seem to be missing the point. I'm not addressing ChatGPT's responses at all; my focus is on the actions that seem to react to GPT's statements. The crux of the matter is that the AI moving Lara isn't tied to ChatGPT's dialogue. However, in the resulting videos, this connection is clearly evident. That shouldn't be possible if we were to believe the official explanation. I'm uncertain why you're raising his game knowledge; that's something I never mentioned. He isn't obligated to reveal his *entire* code, but he hasn't even presented a small portion of it. Additionally, he hasn't provided any details about the software he used, apart from mentioning ChatGPT + Google Lens for narration. It's puzzling why one would entertain the idea of tolerating even 15% falseness. If this were genuine, what prevents Fox from simply being 100% honest?
The glitch at the end is another kind of horror. Being stuck in the game ending screen and screaming in panic as she doesn't know what's happening anymore.
Watching Lara kill the bear and make the observation that there are only speckled bears in Peru really caught me off guard. "If I killed a rare species, I could get in trouble again" I laughed out loud when she said that 😂
when she said that " i could be in trouble again " it made me wonder getting into trouble with who or what ? seems all a bit frightning to me as i think this AI still might have a connection through internet or something with someone else or unknown it reports its findings to and what the AI has learned so far from the program it's in . If you think about AI, AI is a mechanism that is constantly evolving by learning and scanning. but did someone ever thought that it also could create self awareness and could be playing us in the end, while we think we are being entertained by it . Scared me a bit when she said it .
You guys are so naive lol. This is clearly fake. First guy, How on EARTH would the AI understand in the context of the game that they're in Peru? Second guy, how would the AI NOT be connected to the internet? You think this guy somehow has a super computer in his home that can process a level of self awareness on par with a human? This is fake, use your heads.
@@YourPalHDee The AI knows it's in a video game depiction of Peru because A. The AI knows that the character it's portraying is a Peruvian archaeologist and B. The AI will process textures that depict Peruvian cultural symbols, which it will store in long term memory and know from them on out that it's in an ancient Peruvian city.
I love how she plays everything safe for her safety. Walking around corners to avoid surprises, always easing down off ledges to avoid impact injuries. As the gamer, we know if something goes wrong, she’ll come right back from a save point… but she doesn’t, so swimming down an unknown underwater tunnel, not knowing where the end is or if there even is an end is a chance she just isn’t willing to take.
this is more a limitation of the AI not understand 3d space all that well. it has trouble understanding depth perception on the 2d images it's being provided. So when it aproaches objects it take time to get many different angles to correctly judge distance. This comes accross as the AI caring about it's safety but it isn't. judging by the fact that everytime laura walks chat gpt mentions something about safety it's probably coded as a "safety walk" and that is what Chat GPT is being told is happening. i.e. ;the AI is unable to distinguish what it's looking at; ;executes a slow approach to get more information; ;chat GPT sees the output "Safety walk"; ;chat GPT mentions it's safety walk as it has no other variables as to why Laura would be walking instead of running; ------ if something dangerous does happen Chat GPT will retroactively justify the action as being related. it's all a great illusion.
@@papabird4425 Shit, A.I. won't have any threats to it's safety if it encompasses the entire internet grid. The culmination of everything on the internet will assign it's character or personality (computerality). Not like the A.I. could Terminator 2: Judgement Day us without being it's own threat to itself. Even when automatons become viable options for it's own maintenance, and the assembly lines building the bots becomes fully automated, humans would still potentially serve purposes for the A.I. Yet, if some A.I. does become uncontrollable, they probably would likely still shut down the grid as a failsafe and then tell us solar flares are to blame.
Well, if you find yourself with a pair of semi-automatic pistols, firing with unexplained pin-point accuracy, with even more so unexplained infinite ammunition... It's probably fair to assume, you've long since already killed a rare specimen or two before!
@@sharontubul2396 I know this, but I don't remember if they went into detail about her getting into trouble for it, though it would make sense if she did.
@@worldcitizen4440Please explain what precisely is fake? It's been explained they're using MULTIPLE AI in unison, for the movements, and that the game is not happening in real time, but they stitched many small clips together into one seamless clip for easier watching.
@@CommanderRedEXE Well, there are some things, that at least wheren't explained in the first video. For example how she tracks her health and gets to think that she should use a medpack because of the risk of 'infection' - while just having lost a very tiny amount of health, which wouldn't actually justify uing a medpack without the made up idea that she could get an infection - as there's no such mechanic in game. He said Lara was concsious of being a video game character - but in no video game ever a generic medpack was of use against a infection risk (There are some specialised consumables in other games for this purpose, though, but that's not what it is in this game). And a generic medpack being of use against infections also isn't a thing in reality that you would find all along the internet, or being propagated by a reality trained AI. als in the first part he admitted that there where some things he simply hard coded inside for example. So it looks very string like this though was implented directly into Lara's virtual brain (or simply made up for the video). It's not that it was bad over all. It's a very interesting experiment, and it looks like some things where just tuned a bit to make it more appealling.
@@elmariachi5133 There is no infection “rule”. The AI just uses a medpack whenever damage is taken. The infection line is just a funny line the creator added to give a coherent reason for using a medpack for only a sliver of damage. The creator lets the AI play the game numerous times, adding to the code as new situations arise where the AI gets stuck. Eventually you have an AI which could play the whole game on its own. Probably not Tomb Raider 2 and up though.
I don't know how difficult it would be, but this is like the height of what kind of content I am looking for on TH-cam. I think a sentient game character playing out its life is such an interesting and cool way of seeing how something interacts. And the amount of work it takes is probably ridiculous, especially from what we saw in the first video. But if you could do a whole series of Lara Croft literally doing what a character like she represents would be. All the quirks, all the quips, all of the action. But from its own mind, this is genuinely one of the coolest things I've ever seen.
its not sentient at all, its a deepfake style voice he put lines into to sounds like Lara, and its him playing the game with his script AI doesnt talk or act like this lmfao
@@ch3cksund3adlol watch the first video he explains how he did everything. It may not be ‘sentient’ in the technical term but it’s definatley an AI playing and experiencing the game and the voice of Lara is generated by the AI, this guy is not playing the game at all and he’s not putting sentences into her mouth. You must be stuck under a rock if you think that AI is not yet this advanced.
This is amazing. A program, reaction, personality. The birth of a new life-form. Its like watching a toddler play tomb raider. A toddler with encyclopedic knowledge of archeology and the history of the human race at its fingertips.
This is feasibly possible to legitimately create today however unfortunately, much of this isn’t AI involved or legitimate. I think the only thing that was AI generated was the text and audio comments themselves that analyzed a video after it was finished, the gameplay is done by a human.
I actually played this game so long ago as a kid, and it was mind-blowing back then. One thing that I vividly remember from this game was that in the tutorial level in Lara's manor, when she was showing the player how to walk, her dialogue was, "When I want to be careful, I walk." So hearing the AI saying this at 6:53 makes me wonder: is this like a self-aware joke remembering back to the tutorial? If so, that is amazing.
I definitely remember spending around 300 cartridges in panic to kill those lions or whatever that was in the beginning of this video. She was way cleaner than me.
That might have also bubbled up from the ChatGPT training data. It probably is a well known phrase connected to Lara Croft, people probably have mentioned it a bunch.
i hate to admit this, but this is the most captivating video i've seen on yt. this lady is more charming than any of the other content creators. her innocence and naivety is refreshing. well done.
@@mattmurphy7030innit haha I was hopeful for some sort of address of these criticisms but given the radio silence I have to assume it's completely fake
@@mattmurphy7030 are you saying foxmaster wrote the dialogue for Lara? pls fill me in. i saw the first level and thought his hypothesis was that you can teach Lara how to interact and how to extemporize commentary. where can i find the critiques of his methodology and/or his conclusions? i was about to recommend this to others, but i should at least know what i'm recommending. thanks in advance.
@@biln2I believe it is an ARG (especially because of the last few moments of this video) while some of what he said in the first video is technically doable, for example the "texture hunting" is kinda based on an actual ai model by Open-AI that played Atari games. The way the different systems connect isn't at all capable of doing this. For example, the "voice AI" isn't at all connected to the "player AI" it only gets audio queues and some textures, so moments like "my pistols should be empty, let's test this" where Lara's dialogue influences the way she plays aren't possible. Another huge clue is the ending of this video, technically the buttons would be perceived as a new texture and get a reaction by Lara, or none at all. That buggy scream is 100% dramatic storytelling
I don't comment on yotube often, but I have to express how much I am enjoying the A.I. Lara videos. Hearing her, in her original voice and sound quality, it makes me strangely happy. I grew up with this Lara, I loved the originals so much. It's like you have brought her alive and I can completely believe her saying most everything she does here. What you have done is amazing, and I am eager to see more
@@LolaCroft Plenty of comments by ONE person actually. And (he claims) he knows more than you about AI that's simply because you don't work with these tools and he says he does. But does he really ? You know him personally ? Not saying he's lying, just that you are quick to believe anything you read. I would not loose time replying to some trolls either, To me he just sounds like a flat-earther saying its impossible to be upside down under the globe as you would fall off into the void... I am not claiming that this is not a hoax, but everything you see in the video is indeed possible and looks legit.
After watching the two episodes fully, I'm not convinced that this isn't a hoax. I think the only way for me to believe it would be to either publish the whole segment with the pauses and debug traces shown ; &/or publish the sourcecode to a repository. This YT channel has not published any developer work during its 16 years of lifetime, and neither don't we have any developer history from FoxMaster. I call this a hoax until proven otherwise, sorry. Either way, there was a lot of work behind it, and it appears to be really convincing, so congrats anyway.
@@HoopleBogart Yeah it's a cool creative project, a "what if" scenario where a video game character becomes self aware. I don't think even the most advanced logic and reasoning AI systems can produce the level of nuance here, though I may be behind the times.
The weird scare at the end is just another hint that this is a hoax and Foxmaster is playing with us. I wouldn't be surprised if the next video's will get even weirder and this is some kind of art project or ARG. Like the AI getting 'too self-aware' and other 'spooky' stuff will happen.
I am so ready for this. But I would totally watch an "endless" stream of her going through the entire game, even if every action takes 30 mins each to perform. Stream it please!
I support this idea. As I said in the chat, it feels more real than the regular gamethroughs which are always speedrunning as hell. AI or not, I like the slow paced style. It brings so many memories, watching a new (or not) player who never played it.
and other games. please build an empire you can't help but become rich from this. please. more characters. more games. more vids. more streams. lovelove
Really hope this becomes a full series! I know you previously said Lara struggles in the later levels, but I'm willing to bet most will still watch - even if it does end up requiring minor human intervention.
yeah, totally. I can foresee certain tough puzzles being built into lara's base memory like "i heard legends of ancient puzzles like this, and I remember how they were reportedly solved, so I'll try that."
You know that each time the compass is brought up that is the program explicitly asking for human intervention? (what Lara says about it is just ChatGPT justifying it after the fact). And also there is probably already major human intervention in all other parts as well. Some people here are going so far to call it all a fake because of that, but I think they are kinda missing the point.
4:37 Has to be the clearest evidence as to why this is a hoax video: In the previous explanation video FoxMaster clearly stated the dialogue is purely for commentary. Yet, here Lara saying "3, 2, 1" exactly matches up with the movement of the character. This suggests that the ChatGPT + Google Lens dialogue also has influence on gameplay, otherwise there's no reason for the robot to move like this.
And the fact that she saw a bear in the previous video, yet made a comment about there only being spectacled bears in Peru in this episode. It’s still fun to watch though!
but if that’s the case, why does it get so many things wrong? for instance the comments about bioluminescent plants and spectacled bears and the feeding trough being a bench… maybe the video was edited a little to match it up or maybe it was just a coincidence
You've managed something incredible here. Here I am completely enthralled seeing exactly how Lara will manage to get through the next area. What she finds and decides to do. How she learns, adapts and little quips like wanting to hear the music have me smiling ear to ear.
@@trixylizard6970 I'm a software engineer and there is no way the AI can exhibit such a complex behavior with (1) the incredibly basic set of rules that were listed in Part 1 and (2) the very short learning phase the AI has received. You would need thousands of computers to train a model that complex. Author claims to have done it with a single Windows VM. What's more, he then contradicts himself in the description of Part 1 by saying he has only used free AI tool to make this video (namely ChatGPT + PlayHT). He got called out many times in the comment section of Part 1. He admitted at some point that the comments were scripted, but nothing related to the gameplay yet. Instead, he released Part 2. The voice audio is definietely generated with PlayHT, *SOME* comments (or part of the comments) are generated with ChatGPT. The gameplay however is 100% him playing.
I was SO happy to see part 2 pop on my feed today. I am making the transition into a very stressful new job/ part of life, and THIS is concentrated relaxation for me. It's like medicine for my mind. Thank you.
Such a great idea, watching an AI play through a game as the character would play, thinking how they would think and making comments like an internal monologue, amazing.
Watching this video made me think of if there was one for GTA III or some similar game. The personality would be more violent, and it would be interesting seeing them drive
I really wanna see this done with Halo. Probably not possible right now given the amount of twitch gameplay and everything, but I think if you were able to train the AI's personality enough it would feel like a classic Halo novel basically, and who wouldn't want that? The perfect version would be to also have an AI Cortana who cannot play but is equally along for the ride with John and Cortana able to interact, allowing them to work together on things and basically just, you know, be John and Cortana. And then, keeping the continuity of self on the AI... John forced to be without her when applicable in Halo 3, after two games of working together. That could be extremely oof. Just imagine him trying to bounce things off her for a moment out of habit, and then remembering she's not there anymore. AI John and Cortana play Halo could honestly end up a work of art just because of his pre-existing issues in canon combined with getting to explore the trauma of losing her. Only downside is that if you train a fully functional John AI who can beat the Halo trilogy, you've kinda just invented something with military applications.
I'm still skeptic about how much is actually AI and if even any. But it's still super entertaining. And for now i handle it like "Half life but the AI is self aware" kind of entertainment.
@@unfortuitousash human made script? i use chat gpt all the time, and even if it was prompted, Ai can definitelty come up with dialogue at this quality level. "you are lara croft (describe her here). write me a series of one liners for each of these situations: just fellt off some blocks into the water. just opened a wooden gate in a temple. etc. the ai can DEFINITELTY at the very least write this.
I... it bothers me how few people seem to get that. I mentioned in a Twitch stream that I was skeptical and got treated like a nut, and it doesn't help that most comments on the videos are "wow amazing what AI can do".
@@lkjkorn19Well given how much AI npc development happened in the last two years, I wouldn’t be surprised if all the gaming in this vid was done by AI. Contrary, I would be more surprised if it isn’t all AI. I suggest you look into the Skyrim AI NPC vids, the software is getting better day by day
@@viciousyeen6644 If you have any good suggestions, I'm genuinely curious to see what was [really] achieved with AI in other games. Currently watching 0wCjosz1vOA
it's not a human laugh, she used a "ha ha" laugh at least twice as well as the "hehehe" laugh once. They seem to express "I succeeded", where the latter is a more emphatic version of the former. It's not a human expression at all...
The ending was more uncanny than whatever the game has ever had to offer. Good job! I found you in my recommendations and got curious. This was an experience, hope to see more!
I keep watching the first video, impressed by the amount of work behind all that. Being a fan of the older Tomb Raider games, it makes them even more interesting now and I can't wait to see this level being played by Lara herself. Keep up the awesome content!
With the training as described an AI would never ever play a leven this human-like. The decision making in the very first fight alone is how a TR veteran would play the game, not an AI
@@Inklnx In this video's case you should be asking what makes it real, because the author gives very little explanation as to how he programmed what would be a very impressive ai. It just makes no sense that a person would program something this impressive and not go in-depth about how they coded everything.
After watching the first video and now this one, I don't think this is actually AI, just someone playing to mimic an AI and a scripted AI Lara voice over. I would definitely believe if a video of the devices that run the AI are shown along side gameplay, but most of this literally feels like a person is behind the controller, behaviors that only an experienced TR player would know about how the environment Lara's in works. (Lara saying the floor could break, falling onto it and backflipping back to safety as an example.) Either way its a fun idea to have Lara narrate her adventure.
I'm not convinced. the explanation behind how it works and what tools were used in the first video were very competent and logical. The AI is designed to look for and analyze new textures in order to progress (and I believe cautiously walk towards any new textures). So it makes sense that the AI carefully analyzed the new texture of the breakable floor before walking on it, saw it had gaps in it, and determined it was probably unsafe. I don't think today's AI technologies are incapable of making that assessment. You pointed out the intelligent things it did, but it also said and did strange things that I'd be surprised a human trying to dupe us would fabricate, like mentioning glowberries (Minecraft?) and singing a parody of Let it Go (not appropriate for the time/location of the game), misnaming the trough and key, and speculating the water was coming out of a statue at one point (which it totally wasn't), seemingly not recognizing a door simply because the texture is a darker shade (18:13), or fearing the bear may have followed her in the water. I'm sure the creator would be more than willing to publish full versions of these videos showing the full grueling process without the video edits if enough people demanded it (which he already showed parts of in the first video, but apparently that wasn't sufficient)
@@akpokemon in the first video description stats, "This video may be inaccurate and is made for entertainment," and also that this is all being done by "free AI," after doing a little digging, it would take MILLIONS for someone to create an AI to perform like this. The first video has a good logical idea of how an AI would work with learning in the game, but it still would take years of AI improvement to get anything to work remotely this well with free programing. I think it's a fun idea of narrating and adding fun to a walkthrough, but I don't agree with the misleading clickbaiting to get views.
@KoraTheBot-rh5pk I thought about that too, and TR1's textures are so pixelated, it wouldn't be as easy as 1,2,3 to form dialogue like this "AI" is forming. What really threw me off was Minecraft and Frozen references. There's no way an AI would just incorporate that information if its only built to acknowledge what its doing in THIS game.
@@akpokemon But his logic is a general idea of how an AI would work if there was one capable of playing and behaving like Lara. This TH-camr is great at editing videos, their previous ones give you all the evidence of that, and they aren't stating any coding or the programs thier using other then ChatGPT. Don't believe everything you see on the internet, and i'll stick to that logic until they show some actual proof.
It's amazing but I don't think the programming logic for this is hard, it's just setting the value of speaking to false as she is in water. Other than that her speeches are really mind boggling since they are very human!
So glad to see you decided to continue this! I know full well what a mountain of effort it is, but i'd be down to watch an entire series where "Lara" plays her adventure from start to finish.
Please keep explaining the wierd little hangups and mistakes the ai makes like you did at the end of the first video! I thought it was really interesting to have further insight on how it works. Love that we got another part and hope to see more like this in the future! :)
i agree. I wanted to see a "debrief" again at the end of each episode critiquing it. even just a thirty second "most notable things" would be cool. like in level two here, how she referenced minecraft with the lush caves and glowberries, probably because the graphic she google searched looked almost exactly like glowberries from minecraft.
@@FoXMaSteR001 thx for the heart and answer! Me personally I wouldn't mind waiting a bit longer for an upload with this section at the end. But ofc it's up to you to decide how much time and effort you're willing to spend on these videos haha
The way that AI Lara is ironically more human than Survivol Lara is sending me off. This Lara has survival instincts such as analyzing the temperature of a place based on how much snow is there, the level of humidity due to vegetation, and the presence of food supplies due to predators. In addition, she is aware of her range of motion and its limits, such as not letting herself fall from heights that could hurt her or not getting wet so as not to be cold and using first-aid kits to avoid infections. Omg there's so much going to talk about. Anyway, seeing AI Lara in full adventure feels like being a kid again discovering my childhood hero for the first time.
15:18 The way she chastises herself is too painfully REAL. I love how you inputted the robot with all these actual qualities of Lara so she acts in a manner that does come off as AUTHENTIC and REAL
@@baieteluentertainmentThe way you formulated it makes me believe you don't really understand what's at play here. First things first, this isn't "gameplay with AI generated voice". I believe the AI generated voice is actually the least complex thing to get to work in this context. The first thing to notice about this video is that it's a whole bunch of AIs working together.
@@ItsJustDoctor I mean... it's impressive work. But not impressive enough that I would assume it's fake. The guy hasn't invented nuclear science. He's just tinkering with existing tools.
So this is going to be a series now? As entertaining as it is, you should at least mention that it’s fake/hypothetical, like you slightly alluded to in the first video’s description. It’s very dishonest to garner all this new attention for deceitful content. There are tons of people saying they aren’t sure if it’s real but they still enjoy it. Case in point, you don’t have to pretend it’s real for people to enjoy it.
@@supabass4003 I know that’s what he wants you to believe, I did watch the first video. But yes, it is fake. He is playing the game and scripted all of the voice lines
Honestly the most impressive part of all of this is how many people you've completely fooled. Just goes to show how little people really understand these technologies. Bravo 👏
@@kurtsydavis7517 the fact this is completely scripted and absolutely not an AI making gameplay decisions? The voice over and Lara's 'personality' are clearly AI generated but it 100% has no bearing over the gameplay. I imagine the creator has described each scene to the LLM given the context (i.e. that the LLM should respond as 'lara') and then run the output through an AI voice generator. There is no AI making real time gameplay decisions though, and it does not comment on what is on-screen. The whole "having to wait for 30 minutes each time she comments" and "stitching the footage up" in post-prod are just a lie to make it seem more believable, as is this whole process of 'lara' walking up to walls and scanning them and seeking out new textures... Cool idea but I highly doubt this has actually been implemented! Looks so scripted to me. Adding to this fact is that there have been absolutely no mentions of the implementation of any of this in code. If you watch any other similar projects on TH-cam you will usually get at least a partial walkthrough of some of the code base, and there is nothing shown here - v suspect imo
@FoxMaster: I read your channel description and scrolled trough your videos whilst watching the second episode of this video, and that really wasn´t what I was expecting. Don´t get me wrong, I expected some kind of AI-related channel, because of the massive efford you put into making that two episodes. I really hope, you are aware of the pure gold which lies inside this video-concept of yours, sir! Very well executed. Thank you for that! 👍
I think some of the Dialogue is Ai based and the Voice is definitely Ai created (well done). But I believe everything else is player-driven. There has never been such a life-like Ai ran human player in games ever... this alone would be a huge break through in Ai technology. And I doubt some youtuber was the first to do it...
I can't really vouch for anything, but the channel AI and Games is all about what AI can do and how it is used in game development. From what I recall hearing there, big game developers utilise AI players to debug and test their levels, simulating thousands of paths a player might take through a level, simulating thousands of PvP encounters in multiplayer maps. In these tests actual control of player characters is given to these AIs and they can only act in ways a player could, performing similarly to how players would in order to find the same issues actual players might encounter. On top of that, the channel two minute papers has a multitude of videos on machine learning for game AIs and how complex their behavious can become. I can totally see that behaviour as seen in most parts of this series is plausible. But there are still several situations that baffle me and make me consider human control might have a part in recording these videos.
I agree, some of her comments are too good to be true. I could believe that a bot which is able to play the game has been created, as there are now AIs which can play Minecraft, a much more complex game, but I think some of the commentary has been written by a human.
The 'AI' realises some part of the floor are destructible even if it supposed to be the first time it encourters is. Why would anyone post something so fake is beyond me.
@@bverte677 money. There are tons of successful fake channels that could be debunked with common sense. Now with recent AI advances people believe AI can do magic and lacking domain knowledge it gets even harder to debunk. It seems believable because he shows his “method” and “mistakes” but do the “results” align with the method and are they plausible given the current capabilities of AI?
Ty dude, I hope the lag won't be too problematic in this one, there are so many parts to merge I have rendering problems, I tried to fix it but some micro freezes remain, I'll try on another computer later 😬
It only shows that to have a fully general AI capable of any complex task it needs several separate modules working in perfect sync (just like our brains have different facilities).
Our brain is amazing, it's even more crazy most animals can do better than a bot, when you see a fish spotting food without seeing it at first, getting close to it and eat hit when it's moving that is amazing, even a snail can do that 🤯 They recently scanned the brain of a fly, this is already super complex 😳@@dwsel
@@FoXMaSteR001 Yup, definitely it's something to expect 👍I find projects like this very interesting - they teach a lot about us as well. One of my current hobby is "cats talking in English" - I spread the news on all occasions 😅 For talking cats I suggest: Billispeaks and catmanjohn, and for dog lovers: WhatAboutBunny . I see you're interested in all projects revolving around neurology x technology - you'll going to like these channels as well. Owners taught their pets how to use soundboard with buttons (sometimes 50+ words). Cats and dogs use them, but it's not only for getting treats, they use it also to talk about emotions, what they hear (what human cannot hear), ie. Bunny dog learnt a concept of dreams, and can tell you what she dreamt about, Billi cat learned concepts like things like times of day, conditional sentences and planning in the future. I love these channels.
@@FoXMaSteR001 BTW Some people come sceptic to these ideas, but we've already have have convinced dozen of people, that the sentences they build are coherent, thought out and appropriate to the situations.
Oh wow!! I did not think we would be getting another one of these videos so soon! What a treat! This is the reason why I subscribed I absolutely loved the last video and I hope you make this a series with every tomb raider game!🎉
These are a pleasure to watch. I am not entirely convinced that the AI can create such witty and consequential lines, that are so fitting with Lara's character from this era... I suspect there's some human input in the writing... regardless, but I'm SO here for it and hope you make more!
@@-VenyX- I'm a machine learning developer. I went to grad school for machine learning. I watched the videos. What else did you want? "Throwing around misinformation" lmao my guy you fell for a fake video
@@-VenyX- what's especially not plausible is the computer vision aspect. The techniques he described in the first video are simply not adequate or realistic to achieve this effectiveness. In fact, the effectiveness he claims to achieve here is simply revolutionary. Should I go on?
The amount of nostalgia I have for the OG Tomb Raider game is off the charts! I played this game SOOO much on the PS1 & loved every minute of it. The new games have better graphics and controls but nothing will ever beat the sense of discovery and exploration these old games had. I'm usually against chatterbox characters but my love for the original Lara is too much to be denied, I love listening to her original voice & chatter here! The bits that aren't self aware & 4th wall breaking actually fit really well.
This is a whole new level of gaming I love to see progress, imagine if Lara retains her memories from each game play of each game? Ooo! That'd be an exciting venture.
It’s pretty amazing that she made that comparison to minecraft re: lush caves and glow berries, obviously the reverse image search of the vine textures returned a bunch of minecraft results, which makes sense because they really do look like the lush caves and the glow berry vines. So so cool
I was really amazed by the first video, but after having seen some of the creator's other videos, I have the suspicion that it's a clever ruse. Specifically, there's a jump the "AI" performs in the last video that almost matches exactly one of his other non-AI videos. In both videos, Lara lingers around the timed door, turning around and watching it until it shuts, before progressing further. It seems unlikely the AI would happen to behave the exact same way as the creator, based on the given explanation as to the AI's supposed logic. Still fun to watch, but I don't really appreciate false-pretenses.
I don't believe its real either. Fun idea and concept of Lara being "aware" of her environment, but if it is fake, I definitely don't like the idea of taking advantage and misleading the audience.
It's for entertainment. Think "X written by AI" bits or Fargo's "True Story" genre. It was well sold, but part of the enjoyment relies on the illusion.
As much as I hate being a spoilsport, I still have to admit that I highly doubt the gameplay is done by AI. The first level clearly showed gameplay with prior level knowledge rather than trial/error, and there's no way the setup would suddenly change for the next one. Also, the "3-2-1" with the block, absolutely no danger the AI did that AND commentated on it while doing it. If this is deliberately done as a prank/hoax then so be it, good joke mr uploader. Otherwise I think people need to wise-up a little.
can you point to time stamps that showed gameplay with prior knowledge, from the first level? these videos look convincing to me, and given that I watch two minute papers and know the progress of computer vision, etc. it all seems bleeding edge but plausible.
@@westingtyler1if you have half the ability to implement it, you'll know it's fake, because you can at least reproduce the procedure in your head and discover very obvious discrepancy. Two minutes is entertaining, but it doesn't go in depth and is basically just quick but superficial news. But nothing show there anything to that level. The key is the insane acting and editing skills for an ai, not about actually surveying the level, but showing to an audience, with a continuity that last not just a few second, but literally an entire video.
OH my life... This is an absolute new approach. Thanks to community this game never gets old. Watching AI lara with commentary is just like listening to a tour guide! 🤣🤣🤣🤣
yeah, and it's actually educational. like how lara is pointing out there is toltec, mayan, and aztec stuff in this inca village, and how these bears should not be here, etc. and what an amphora is. you learn not only about real history but also about how the devs put the game together. clearly slapping different interesting cultural imagery together, haha.
"it lookd like puma skin. It would look good in my kitchen" "this wooden chair would be perfect in my dinning room" "looks lovely, but too large to be picked up" Indeed guys, she's British...
You’ve flew a bit too close to the sun now haha. A few giveaways that the controls are from an experienced player trying to emulate a machine, and not actual AI. Still it’s entertaining to watch
It's the near future, sentient AI has emerged and people happily access it for daily tasks like time-keeping and fact-checking and some use it for entertainment purposes. Imagine a personal AI who has the ability to take on any persona and mimic them perfectly, now imagine them playing your favorite games while you co-pilot. Watching along and making comments live and having them react to your input, it's like watching the perfect let's play... honestly I don't think I'd *play* a video game ever again. Next thing you know I'm snacking on some doritos, chillin' with Cortana watching her play Halo, life is good. Edit: Either that or an AI that's attuned to your personality, molded to perfectly suit your style of play and commentary. It's like peak v-tuber.
Haha, great. Don't worry it werent a complaint. I'm just not used to it from the countless hours spent playing these games :D @@FoXMaSteR001 It might have been answered already, but do you intend to release any of this for pp to play around with?
That Lara is definitely a survivallist, but lacks a bit of exploratory spirit. No secrets found. And she definitely needs to at least map the environment in her head. It felt every time she came back to the 3 doorways she seemed to see it for the first time. But really aa brilliant achievement. I watched the first video and saw the Level 2 video up coming, so I waited. Please keep em coming!
this could be a result of the fact the lara memory needs to be reset every few minutes to avoid confusing the AI, so she has kind of like "fish" memory. he said this in the first video. definitely would be stronger with more, but that is the state of the art. I have Herika for Skyrim, and her somewhat limited memory is definitely the part that reduces the credibility the most, though it's still insane. you just have to imagine the character is a REAL person with memory issues, and it suddenly works.
it would be cool if the ai could slowly built a 3d mental map in her head of the level, with pints of interest marked and labeled, so she can reference it as needed.
the Frozen reference makes me wonder if there is a way to limit the robot's popular culture knowledge to only what was available until the 90s? this is fascinating stuff!
I have never played Tomb Raider nor have I ever seen a self-aware Lara Croft play through it, but I watched the whole thing and I was fascinated. There is something about the older games which are comforting in their simplicity. And her regular comments are hilarious and amazing. Hope to see part 2!
DEBUNKED as fake, thoroughly. It's very clear that it's a prerecorded run with some scripted commentary over the top, the only part that's real is the vocal generation AI. Put simply: the commentator AI shown to have knowledge of Lara's intentions before she takes action, however this is impossible since the only data the commentator AI has access to is what is being shown on screen, it does not have access to data in Lara's head (i.e. the player AI). This and a bunch of other small pieces of evidence that just don't stack up (e.g. the overcompensated movement during swimming, the repeated 'checking' of the environment, the looking at the compass, etc. etc. etc. there's loads of clues everywhere). tbh i feel bad for the creator, he is clearly just a big tomb raider fan who has a tiny channel that hasn't been doing great even though he's put a lot of work into it over the years. he has no history of posting about AI, and lots of history of posting tomb raider fan videos. unfortunately this is just an elaborate hoax by someone who doesn't fully understand AI, and therefore the attempt isn't nearly good enough to pass the sniff test. i don't think he expected this to get as big as it did and now he's dug himself into a hole. he says in the description that everything is explainable, so i look forward to that video. if he does come clean, he'll have learned a lesson and also gained a bunch of subscribers, so it won't be that bad, but if he keeps up the charade, someone will eventually do a proper debunk video and that's what he'll be known for. my recommendation to the creator: come clean before you get cleaned, it's inevitable and you'll regret the backlash, if you take the first step, you're still in control and you can keep a bit of credibility and dignity instead of apologising after getting caught, that's never a good look. good luck
My prediction is that he is going to try steer this into the ‘creepypasta’/ARG/art kinda way. The spooky jumpscare at the end was already a hint towards it. Maybe some scary stuff about Self-Aware Lara getting an existential crisis about being stuck in a game. It would be smart because the naive believers would still like the entertainment, and the people who know it is fake will say that it was a great art project at the end.
People who know a project like this is impossible can tell, and people that want to believe this is real - will continue doing so. I don't think there's anything to debunk or come clean with, same way you wouldn't try to prove a religious person that their belief is ridiculous, they believe in what they choose to because it makes them feel better. Also it might ruin other peoples entertainment, there's a hoax culture on the internet who enjoy a possibility of different things happening, straight up saying everything is fake ruins the fun 💁♀(i mean really? whipping out a "for fun" project with surface level explanation and no sources that even in theory would require years of scripting, a huge team and financial investment, what IS there to debunk?)
@@jUQMtDmf yeah I did find that ending odd... I think you might be right about the ARG angle, and even if it wasn't the original intention of the first video, if he pursues this angle it would also allow him to save face without having to apologise
@@boba7288you seem to clearly recognize this content as fake too. the main issue is that the creator misleadingly portrays it as real, targeting not just blind believers, but also those curious about AI but unfamiliar with its intricacies, and for whom spotting the flaws may not be as immediately clear as it is for you and me. the recommendation system is showing it to people like myself with a clear preference for genuine AI content who don't follow any ARG or tomb raider stuff, so even just accessing this series in that context is already misleading. if the goal was a simple ARG, why the elaborate efforts to make it seem real? it seems to aim to trick those who are interested but new to tech and AI, giving them false expectations about current AI capabilities. from the comments, it's apparent that the majority of viewers are misled by this video, and it feels unethical, regardless of the entertainment value. also, the analogy of trying to sway religious believers is off. i'm not writing this for blind believers in any hoax or ARG, i'm not interested in trying to convert them. however, i am interested in reaching those excited by tech and AI and who are seeking genuine content. i'm also highlighting the creator's questionable intentions, it's more like calling out a sleazy multimillionaire televangelist in order to prevent more people from being misled than trying to influence their congregation's faith in God. creators shape perceptions, particularly in the realm of technology. people interested in tech are actively influenced by the accuracy (or inaccuracy) of the content they consume, and so imo, creators who do edutainment videos, like this series appears to imitate, have a responsibility to ensure their content's accuracy, either that or to make it clear that it's an ARG. false information not only prevents genuine learning but also discourages those interested in entering the field who have genuine curiosity. nipping in the bud misleading videos like this is important because if it snowballs into a trend, we risk undermining the broader integrity of content produced about AI within the AI community, and we risk overshadowing the efforts of sincere creators. i really don't want to see the genuine and excited AI corners of the internet get taken advantage of and just become a hoax-fest, do you? imo, this series not just an unethical breach of trust, but a straight up exploitation of the excitement and desire to learn that those who are interested in tech have but may still be young or new, and who engage with entertaining content for motivation and genuine understanding. all while it disregards the efforts of genuine creators whose content would be much better suited for that audience, reducing the impact and reach of their authentic contributions. misrepresenting AI for the entertainment of the 'hoax community' harms those genuinely interested in AI. that's my motivation in calling it out.
He could legit just add a disclaimer to the two previous videos saying: "The AI referenced in this video is being used to generate comments via GPT / TTS, and is being overlaid atop captured gamefootage. The original intention was to use AI techniques to also have the game played by AI, but this proved to be too difficult to implement, so I created this as a proof of concept for what might be possible with the right toolsets. I am open to constructive feedback on how to get something similar working, and I apologize for misleading anyone with the final result." Edit: @kerrcampbell7407 if I could like your follow up comment more than once I would, that is exactly why this feels so fucking slimy, and why it's gotten under my skin. It's so dishonest and exploitative.
Lara approaching what kind of appears to be flayed human skin and saying "that would look lovely in my kitchen" is the content that keeps me coming back
After Ep1, I was hooked. I only reserved my subscribe because it kinda seemed like you might not want to do this again. Now, with Ep2, I'm 100% in. Keep it up!
Video 2 is not even trying to pretend any more. It's just there to entertain the ones that are fully committed to suppressing all disbelief. Which I kind of understand, it's 100% not entertaining if you don't engulf yourself in this incredibly well written story. I say incredibly because the debriefing at the end of video 1 is really work of genius, very credible.
It is possible. There are many AIs at play here, with very big databases like Google Images and ChatGPT, basic AI game knowledge, and a good understanding of the game and a good understanding of player behavior, as well as hours of IRL time destined to processing each 4 to 6 seconds of the game, and a knowledgeable person working behind all of this, and working out the issues.
I love this series so much. I've shown it to my friend because even though you describe exactly how you did it there are moments where I can't help but believe this isn't ai at points. That only a human could be so funny/cute.
Lara thinking to herself: "I've never seen this button before. Is it a good idea to press it?"
Lara speaking: "No."
Lara thinking to herself: "Alright."
"I'm never diving again"
*Dives later anyways
This is not real. To clarify, I'm not questioning the feasibility of AI playing a video game - that's undeniably achievable. My focus is on this particular video itself. He has managed to create a surface-level semblance of plausibility, and credit is deserved for that. However, there are many reasons why the video is fabricated.
- He never displays any code or functional software, beyond ChatGPT + Google Lens for dialogue. He frequently mentions 'The Program' he employed or created, yet what 'that' actually really is, we never find out. (Interestingly, he noted in a comment that he uses ChatGPT to assist in crafting his videos due to speaking French. It wouldn't be surprising if he also sought GPT's input to enhance the AI concept's credibility.)
- A significant inconsistency that Fox seems to have overlooked is that he himself claims that Lara's voice is purely for narration, so separate from the AI controlling gameplay. Nonetheless, several instances in the final product contradict this claim. For example, Lara discussing her endless magazines and proceeding to test the concept, or her query about gravity followed by a confirming jump. Another instance occurs at 4:36, with Lara counting 1-2-3 and the AI synchronously bouncing.
- He demonstrates Lara's movement being driven by the robots exploration of 'new textures'. However, he doesn't delve significantly further into this concept, lacking depth beyond this basic explanation. There should be way more going on besides just simply that to get a 'robot' to solve a Tomb Raider level.
- Moreover, there's an odd visual choice of showing a webpage about a smoke alarm-detector using Arduino while discussing audio recognition (seemingly stemming from a Google search for 'audio recognition', and using the first result). Here, he again doesn't detail the exact software he employs, except for likening it to TH-cam copyright detection. The only thing that chapter _really_ shows is a dictionary page featuring the definition of 'ouch' and some waveforms in Audacity (simply audio recording software).
When comparing Fox's videos to other genuinely authentic showcases on the same subject, the distinction in depth becomes very evident. Other creators delve much more extensively into their topics, despite their projects being simpler and smaller in scope.
I think the "no" came from the game itself saying the action was not possible because she had no ki...
@basicallyhuman You seem to be missing the point. I'm not addressing ChatGPT's responses at all; my focus is on the actions that seem to react to GPT's statements. The crux of the matter is that the AI moving Lara isn't tied to ChatGPT's dialogue. However, in the resulting videos, this connection is clearly evident. That shouldn't be possible if we were to believe the official explanation.
I'm uncertain why you're raising his game knowledge; that's something I never mentioned.
He isn't obligated to reveal his *entire* code, but he hasn't even presented a small portion of it. Additionally, he hasn't provided any details about the software he used, apart from mentioning ChatGPT + Google Lens for narration.
It's puzzling why one would entertain the idea of tolerating even 15% falseness. If this were genuine, what prevents Fox from simply being 100% honest?
@@jUQMtDmf Why are you saying "his" code? It's possible GPT wrote it depending on which version it is.
The glitch at the end is another kind of horror. Being stuck in the game ending screen and screaming in panic as she doesn't know what's happening anymore.
"Hey-? AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAUUUUUUUUUUUUUUHHHHHHHHH!!!!!!!"
26:03
Seemed to me like she was getting ripped apart. :(
Creepiest shit I've seen in a long time, to be honest.
meh, I thought her sudden, and inconsolably existential terror was endearing!
That was horrifying! Wasn't expecting it at all 😭
I left a comment on his first please make sure v she doesn't suffer from these memories
"Off you go, bat. I'm listening to the music" Literally me during that part, every single time
I love it!! 🤣
Watching Lara kill the bear and make the observation that there are only speckled bears in Peru really caught me off guard. "If I killed a rare species, I could get in trouble again" I laughed out loud when she said that 😂
when she said that " i could be in trouble again " it made me wonder getting into trouble with who or what ? seems all a bit frightning to me as i think this AI still might have a connection through internet or something with someone else or unknown it reports its findings to and what the AI has learned so far from the program it's in .
If you think about AI, AI is a mechanism that is constantly evolving by learning and scanning.
but did someone ever thought that it also could create self awareness and could be playing us in the end, while we think we are being entertained by it .
Scared me a bit when she said it .
You guys are so naive lol.
This is clearly fake. First guy,
How on EARTH would the AI understand in the context of the game that they're in Peru?
Second guy, how would the AI NOT be connected to the internet? You think this guy somehow has a super computer in his home that can process a level of self awareness on par with a human?
This is fake, use your heads.
@@YourPalHDeeshh, let people enjoy things.
Right? She's bit silly but feels like that AI has personality, I love it :D
@@YourPalHDee The AI knows it's in a video game depiction of Peru because A. The AI knows that the character it's portraying is a Peruvian archaeologist and B. The AI will process textures that depict Peruvian cultural symbols, which it will store in long term memory and know from them on out that it's in an ancient Peruvian city.
Lara talking about how she found a 'fork' had me dying haha, I LOVE this, I would watch for days
I am going to refer to my keys, as headbands from this day forward!
Ugh I lost my hair comb again how am I gonna get to work 😂
I love people in unknown environments mixing what's obvious to others, it's like someone in a car saying the red submarine icon blinks on the cockpit.
A fork for her cake?
I love how she plays everything safe for her safety. Walking around corners to avoid surprises, always easing down off ledges to avoid impact injuries. As the gamer, we know if something goes wrong, she’ll come right back from a save point… but she doesn’t, so swimming down an unknown underwater tunnel, not knowing where the end is or if there even is an end is a chance she just isn’t willing to take.
Wait till AI sees humanity as a threat to its safety.
@@papabird4425 Underrated comment. I. Asimov approves !
this is more a limitation of the AI not understand 3d space all that well. it has trouble understanding depth perception on the 2d images it's being provided. So when it aproaches objects it take time to get many different angles to correctly judge distance. This comes accross as the AI caring about it's safety but it isn't.
judging by the fact that everytime laura walks chat gpt mentions something about safety it's probably coded as a "safety walk" and that is what Chat GPT is being told is happening.
i.e.
;the AI is unable to distinguish what it's looking at;
;executes a slow approach to get more information;
;chat GPT sees the output "Safety walk";
;chat GPT mentions it's safety walk as it has no other variables as to why Laura would be walking instead of running;
------
if something dangerous does happen Chat GPT will retroactively justify the action as being related. it's all a great illusion.
@@papabird4425 Shit, A.I. won't have any threats to it's safety if it encompasses the entire internet grid. The culmination of everything on the internet will assign it's character or personality (computerality). Not like the A.I. could Terminator 2: Judgement Day us without being it's own threat to itself. Even when automatons become viable options for it's own maintenance, and the assembly lines building the bots becomes fully automated, humans would still potentially serve purposes for the A.I.
Yet, if some A.I. does become uncontrollable, they probably would likely still shut down the grid as a failsafe and then tell us solar flares are to blame.
@@TimJBucci we could turn off electricity and it knows that
_“If I’ve killed a rare specimen, I’ll be in trouble again.”_
I like how she says “again.”
Well, if you find yourself with a pair of semi-automatic pistols, firing with unexplained pin-point accuracy, with even more so unexplained infinite ammunition...
It's probably fair to assume, you've long since already killed a rare specimen or two before!
I'm inclined to believe she may have killed bigfoot after finding him.
@@jamesfinch9534i mean If Not If He didnt aproach her she tries to stay peacefull If WE are honest
Lara: It's an amphora. Lovely, but too large to be picked up
Also Lara: *Pushes completely dense big ass square block*
I do believe she meant for taking purposes much to big and heavy to be lugging about while Tomb Raiding. xD
@@projectTyrasa she even complains about how heavy the square is
The picking up and carrying is the problem, pushing is different because you're not lifting.
"3,2,1, push push push"
These are amazing! My favorite line had to be "if I've killed a rare specimen, I'll be in trouble again" Like it's happened before lol
Yeah, almost as if Somebody's written this script.
Lara hunted Bigfoot prior to the events of TR and gained notoriety in the press
@@sharontubul2396 I know this, but I don't remember if they went into detail about her getting into trouble for it, though it would make sense if she did.
So people think this is real?
@@brianbagnall3029 fake or not it's nice to watch and there's been a lot of thought put into making it.
I'm not convinced of the authenticity of this experiment, BUT, this is absolutely a vision of the future of "lets plays"
yes sure its clearly fake
but funny
@@worldcitizen4440Please explain what precisely is fake?
It's been explained they're using MULTIPLE AI in unison, for the movements, and that the game is not happening in real time, but they stitched many small clips together into one seamless clip for easier watching.
@@CommanderRedEXE Well, there are some things, that at least wheren't explained in the first video. For example how she tracks her health and gets to think that she should use a medpack because of the risk of 'infection' - while just having lost a very tiny amount of health, which wouldn't actually justify uing a medpack without the made up idea that she could get an infection - as there's no such mechanic in game. He said Lara was concsious of being a video game character - but in no video game ever a generic medpack was of use against a infection risk (There are some specialised consumables in other games for this purpose, though, but that's not what it is in this game).
And a generic medpack being of use against infections also isn't a thing in reality that you would find all along the internet, or being propagated by a reality trained AI. als in the first part he admitted that there where some things he simply hard coded inside for example. So it looks very string like this though was implented directly into Lara's virtual brain (or simply made up for the video).
It's not that it was bad over all. It's a very interesting experiment, and it looks like some things where just tuned a bit to make it more appealling.
@@elmariachi5133 There is no infection “rule”. The AI just uses a medpack whenever damage is taken. The infection line is just a funny line the creator added to give a coherent reason for using a medpack for only a sliver of damage.
The creator lets the AI play the game numerous times, adding to the code as new situations arise where the AI gets stuck. Eventually you have an AI which could play the whole game on its own. Probably not Tomb Raider 2 and up though.
Lara talking about infections and using a medipack for a sliver of extra health is so good
I was like "She will not use a whole medipack for a tiny scratch. OMG SHE DID, SHE USED THE WHOLE MEDIPACK!" 🤣
Meanwhile in the first part she didn't use a medpack for the entire duration of the mission xD
@@Alexus00712And she got bit multiple times!
I don't know how difficult it would be, but this is like the height of what kind of content I am looking for on TH-cam. I think a sentient game character playing out its life is such an interesting and cool way of seeing how something interacts. And the amount of work it takes is probably ridiculous, especially from what we saw in the first video. But if you could do a whole series of Lara Croft literally doing what a character like she represents would be. All the quirks, all the quips, all of the action. But from its own mind, this is genuinely one of the coolest things I've ever seen.
its not sentient at all, its a deepfake style voice he put lines into to sounds like Lara, and its him playing the game with his script
AI doesnt talk or act like this lmfao
@@ch3cksund3adit's all AI take a look at the First video of the series, he shows and explaina how he does all the stuff, it's really amazing!
I’ll believe it if he actually releases the modifications done to the game or posts a paper with actual references.
@@ch3cksund3ad you obviously never saw Doug Doug messing with chatGPT and Elevenlabs
@@ch3cksund3adlol watch the first video he explains how he did everything. It may not be ‘sentient’ in the technical term but it’s definatley an AI playing and experiencing the game and the voice of Lara is generated by the AI, this guy is not playing the game at all and he’s not putting sentences into her mouth. You must be stuck under a rock if you think that AI is not yet this advanced.
This is amazing. A program, reaction, personality. The birth of a new life-form. Its like watching a toddler play tomb raider. A toddler with encyclopedic knowledge of archeology and the history of the human race at its fingertips.
This is feasibly possible to legitimately create today however unfortunately, much of this isn’t AI involved or legitimate. I think the only thing that was AI generated was the text and audio comments themselves that analyzed a video after it was finished, the gameplay is done by a human.
@@RyluRocky The first video literally explains how a separate AI plays the game
I actually played this game so long ago as a kid, and it was mind-blowing back then. One thing that I vividly remember from this game was that in the tutorial level in Lara's manor, when she was showing the player how to walk, her dialogue was, "When I want to be careful, I walk."
So hearing the AI saying this at 6:53 makes me wonder: is this like a self-aware joke remembering back to the tutorial? If so, that is amazing.
I definitely remember spending around 300 cartridges in panic to kill those lions or whatever that was in the beginning of this video.
She was way cleaner than me.
@@arthurvg2217her aim is perfect.
@@stevensteven3417 and so casual as well!
Dude it's all fake
That might have also bubbled up from the ChatGPT training data. It probably is a well known phrase connected to Lara Croft, people probably have mentioned it a bunch.
This is pure gold, I died when she shot the bat and said "Off you go, I'm listening to the music!" 16:02
I'm so obsessed with Self-Aware Lara and her progress through this game. I hope there will be more installments!
I'm with you, I want to see more like this. I'd love to see AI Lara go through the entire game.
This is basically a Lara Croft let's play.
@@MrSillyheadSrBut with Lara in charge!
@@falconeshield That's what I meant. It's her let's play.
i hate to admit this, but this is the most captivating video i've seen on yt. this lady is more charming than any of the other content creators. her innocence and naivety is refreshing. well done.
The most captivating video you've seen and it's fake 😢
@@mattmurphy7030innit haha I was hopeful for some sort of address of these criticisms but given the radio silence I have to assume it's completely fake
i know. seriously. her innocent, competent, non-obnoxious curious attitude is great.
@@mattmurphy7030 are you saying foxmaster wrote the dialogue for Lara? pls fill me in. i saw the first level and thought his hypothesis was that you can teach Lara how to interact and how to extemporize commentary. where can i find the critiques of his methodology and/or his conclusions? i was about to recommend this to others, but i should at least know what i'm recommending. thanks in advance.
@@biln2I believe it is an ARG (especially because of the last few moments of this video) while some of what he said in the first video is technically doable, for example the "texture hunting" is kinda based on an actual ai model by Open-AI that played Atari games. The way the different systems connect isn't at all capable of doing this. For example, the "voice AI" isn't at all connected to the "player AI" it only gets audio queues and some textures, so moments like "my pistols should be empty, let's test this" where Lara's dialogue influences the way she plays aren't possible. Another huge clue is the ending of this video, technically the buttons would be perceived as a new texture and get a reaction by Lara, or none at all. That buggy scream is 100% dramatic storytelling
I don't comment on yotube often, but I have to express how much I am enjoying the A.I. Lara videos. Hearing her, in her original voice and sound quality, it makes me strangely happy. I grew up with this Lara, I loved the originals so much. It's like you have brought her alive and I can completely believe her saying most everything she does here.
What you have done is amazing, and I am eager to see more
I never knew her voice sounded like a rich English/British women.
@@Eduardo_Espinozayeah, here in France she sounds very destinguished but also in a low sexy voice. Her voice actor is called Françoise Cadol
@@LolaCroft you can generate script with a chatbot and this is very obviously an AI voice model trained on Lara reading this.
I'm with you with all of that, and How about Legacy of Kain too? :)
@@LolaCroft Plenty of comments by ONE person actually. And (he claims) he knows more than you about AI that's simply because you don't work with these tools and he says he does. But does he really ? You know him personally ?
Not saying he's lying, just that you are quick to believe anything you read.
I would not loose time replying to some trolls either, To me he just sounds like a flat-earther saying its impossible to be upside down under the globe as you would fall off into the void...
I am not claiming that this is not a hoax, but everything you see in the video is indeed possible and looks legit.
After watching the two episodes fully, I'm not convinced that this isn't a hoax. I think the only way for me to believe it would be to either publish the whole segment with the pauses and debug traces shown ; &/or publish the sourcecode to a repository.
This YT channel has not published any developer work during its 16 years of lifetime, and neither don't we have any developer history from FoxMaster.
I call this a hoax until proven otherwise, sorry. Either way, there was a lot of work behind it, and it appears to be really convincing, so congrats anyway.
it's definitely fake dawg. but it's still cool :)
its fake as fuck but very entertaining nonetheless. in the comments to the first video a lot of people go into detail why this is obviously not real.
@@HoopleBogart Yeah it's a cool creative project, a "what if" scenario where a video game character becomes self aware. I don't think even the most advanced logic and reasoning AI systems can produce the level of nuance here, though I may be behind the times.
i agree i dont think this is legit just yet
@@PsychoWedge those comments were deleted
Please keep this series up this is genuinely amazing and could add so much replay value and character depth for Lara.
"This video may be inaccurate and is made for entertainment."
so glad to see a part two! that noise at the end scared the life out of me lol
So did it to me. My ears are destroyed.
Very creepy, scared the hell outta me.
The weird scare at the end is just another hint that this is a hoax and Foxmaster is playing with us. I wouldn't be surprised if the next video's will get even weirder and this is some kind of art project or ARG. Like the AI getting 'too self-aware' and other 'spooky' stuff will happen.
Lara trying to literally 4th wall break lol
She also screamed 😂
I am so ready for this. But I would totally watch an "endless" stream of her going through the entire game, even if every action takes 30 mins each to perform. Stream it please!
Naaahhh... boring. I prefer edited vid.
I support this idea. As I said in the chat, it feels more real than the regular gamethroughs which are always speedrunning as hell. AI or not, I like the slow paced style. It brings so many memories, watching a new (or not) player who never played it.
and other games. please build an empire you can't help but become rich from this. please. more characters. more games. more vids. more streams. lovelove
Yeeees Stream it please, I wanna see people's reacting live to her commentary
don't need to stream , it can be just a premiere release with live chat
When she pulled the lever and fell into the water and her "oh no" there was a hint of a laugh. So well done, I didn't expect that!
19:42
"I hate gravity!"
That one killed me.
Really hope this becomes a full series! I know you previously said Lara struggles in the later levels, but I'm willing to bet most will still watch - even if it does end up requiring minor human intervention.
yeah, totally. I can foresee certain tough puzzles being built into lara's base memory like "i heard legends of ancient puzzles like this, and I remember how they were reportedly solved, so I'll try that."
You know that each time the compass is brought up that is the program explicitly asking for human intervention? (what Lara says about it is just ChatGPT justifying it after the fact).
And also there is probably already major human intervention in all other parts as well.
Some people here are going so far to call it all a fake because of that, but I think they are kinda missing the point.
"I hate gravity!" That is indeed something Lara would say.
We could make a list of things she doesn't like.
Lara was never fond of locked doors, and now we know that she hates gravity.
AI making frustrated outbursts like a person is amazing.
I am offended
@@GravityFFFthen stop breaking peoples legs!
That a dark souls reference!!
4:37 Has to be the clearest evidence as to why this is a hoax video: In the previous explanation video FoxMaster clearly stated the dialogue is purely for commentary. Yet, here Lara saying "3, 2, 1" exactly matches up with the movement of the character. This suggests that the ChatGPT + Google Lens dialogue also has influence on gameplay, otherwise there's no reason for the robot to move like this.
lol it's so obvious
And the fact that she saw a bear in the previous video, yet made a comment about there only being spectacled bears in Peru in this episode. It’s still fun to watch though!
yeah nothing about this "feels" like ai. this seems fake.
And honestly, the weird jumpscare ending that just ended the video with no explanation just shows to me that FoxMaster is playing with us.
but if that’s the case, why does it get so many things wrong? for instance the comments about bioluminescent plants and spectacled bears and the feeding trough being a bench… maybe the video was edited a little to match it up or maybe it was just a coincidence
You've managed something incredible here. Here I am completely enthralled seeing exactly how Lara will manage to get through the next area. What she finds and decides to do. How she learns, adapts and little quips like wanting to hear the music have me smiling ear to ear.
As someone who studied cognitive science, it's astounding on so many levels.
@trixylizard6970 you know that it's fake right?
It's fake, Lara is played by him, and he's also writing the comments. The entire thing is misleading on prupose.
@@Neehize What makes you say that? I'm out of the loop apparently...
@@trixylizard6970 I'm a software engineer and there is no way the AI can exhibit such a complex behavior with (1) the incredibly basic set of rules that were listed in Part 1 and (2) the very short learning phase the AI has received. You would need thousands of computers to train a model that complex. Author claims to have done it with a single Windows VM. What's more, he then contradicts himself in the description of Part 1 by saying he has only used free AI tool to make this video (namely ChatGPT + PlayHT).
He got called out many times in the comment section of Part 1. He admitted at some point that the comments were scripted, but nothing related to the gameplay yet. Instead, he released Part 2.
The voice audio is definietely generated with PlayHT, *SOME* comments (or part of the comments) are generated with ChatGPT. The gameplay however is 100% him playing.
I was SO happy to see part 2 pop on my feed today. I am making the transition into a very stressful new job/ part of life, and THIS is concentrated relaxation for me. It's like medicine for my mind. Thank you.
i, too, was thrilled to see part 2. good luck in your transition! sending you all the good vibes, friend :)
Such a great idea, watching an AI play through a game as the character would play, thinking how they would think and making comments like an internal monologue, amazing.
th-cam.com/video/GM-e46xdcUo/w-d-xo.html
Watching this video made me think of if there was one for GTA III or some similar game. The personality would be more violent, and it would be interesting seeing them drive
I really wanna see this done with Halo. Probably not possible right now given the amount of twitch gameplay and everything, but I think if you were able to train the AI's personality enough it would feel like a classic Halo novel basically, and who wouldn't want that? The perfect version would be to also have an AI Cortana who cannot play but is equally along for the ride with John and Cortana able to interact, allowing them to work together on things and basically just, you know, be John and Cortana. And then, keeping the continuity of self on the AI... John forced to be without her when applicable in Halo 3, after two games of working together. That could be extremely oof. Just imagine him trying to bounce things off her for a moment out of habit, and then remembering she's not there anymore. AI John and Cortana play Halo could honestly end up a work of art just because of his pre-existing issues in canon combined with getting to explore the trauma of losing her.
Only downside is that if you train a fully functional John AI who can beat the Halo trilogy, you've kinda just invented something with military applications.
its fake
@@jimmyboy2778It's the Lara voice that makes it work though it's too self aware. It knows it's a video games.
AI Lara has gotten really efficient at bat killing, taking a very calculated single shot at them.
You've definitely got something special here. Not just in terms of this series but as an idea as a whole!
Lara is clearly not controlled by an IA but the idea is cool.
What we see is someone trying to mimic an IA. Not the opposite.
I'm still skeptic about how much is actually AI and if even any. But it's still super entertaining. And for now i handle it like "Half life but the AI is self aware" kind of entertainment.
the only ai is the lara croft voice reading the human made script.
@@unfortuitousash human made script? i use chat gpt all the time, and even if it was prompted, Ai can definitelty come up with dialogue at this quality level. "you are lara croft (describe her here). write me a series of one liners for each of these situations:
just fellt off some blocks into the water.
just opened a wooden gate in a temple.
etc. the ai can DEFINITELTY at the very least write this.
I... it bothers me how few people seem to get that. I mentioned in a Twitch stream that I was skeptical and got treated like a nut, and it doesn't help that most comments on the videos are "wow amazing what AI can do".
@@lkjkorn19Well given how much AI npc development happened in the last two years, I wouldn’t be surprised if all the gaming in this vid was done by AI. Contrary, I would be more surprised if it isn’t all AI. I suggest you look into the Skyrim AI NPC vids, the software is getting better day by day
@@viciousyeen6644 If you have any good suggestions, I'm genuinely curious to see what was [really] achieved with AI in other games. Currently watching 0wCjosz1vOA
Hearing Laura laugh as she completed a jump she thought would be tricky is so uncanny.
Edit: Lara :)
She's really breaking the 4th wall 😶
Hearing the horrific sound at the end of the video as I completed my jump was also uncanny.
Time point?
For me it’s the way she says oh no here: 23:08
it's not a human laugh, she used a "ha ha" laugh at least twice as well as the "hehehe" laugh once. They seem to express "I succeeded", where the latter is a more emphatic version of the former. It's not a human expression at all...
That ending was terrifying. Id love to see the full gameplay, i can't actually believe this is real
i think chatGPT thought she was falling… but maybe did it realize she was… dying?
It's not real 😂
It's not real
@@Mixima101so this is a prank or something
@@lightnoir6914 I'd call it an alternate reality game, like professional wrestling, where we all agree to pretend that we believe it
The frozen reference was halarious. The fact it made it's own lyrics is amazing.
The ending was more uncanny than whatever the game has ever had to offer. Good job! I found you in my recommendations and got curious. This was an experience, hope to see more!
I keep watching the first video, impressed by the amount of work behind all that. Being a fan of the older Tomb Raider games, it makes them even more interesting now and I can't wait to see this level being played by Lara herself. Keep up the awesome content!
It is just a bummer that it is like 90% faked.
This is also creating new fans too!😄
@@DarkNessBear90% fake how?
With the training as described an AI would never ever play a leven this human-like.
The decision making in the very first fight alone is how a TR veteran would play the game, not an AI
@@Inklnx In this video's case you should be asking what makes it real, because the author gives very little explanation as to how he programmed what would be a very impressive ai. It just makes no sense that a person would program something this impressive and not go in-depth about how they coded everything.
After watching the first video and now this one, I don't think this is actually AI, just someone playing to mimic an AI and a scripted AI Lara voice over. I would definitely believe if a video of the devices that run the AI are shown along side gameplay, but most of this literally feels like a person is behind the controller, behaviors that only an experienced TR player would know about how the environment Lara's in works. (Lara saying the floor could break, falling onto it and backflipping back to safety as an example.) Either way its a fun idea to have Lara narrate her adventure.
I'm not convinced. the explanation behind how it works and what tools were used in the first video were very competent and logical. The AI is designed to look for and analyze new textures in order to progress (and I believe cautiously walk towards any new textures). So it makes sense that the AI carefully analyzed the new texture of the breakable floor before walking on it, saw it had gaps in it, and determined it was probably unsafe. I don't think today's AI technologies are incapable of making that assessment.
You pointed out the intelligent things it did, but it also said and did strange things that I'd be surprised a human trying to dupe us would fabricate, like mentioning glowberries (Minecraft?) and singing a parody of Let it Go (not appropriate for the time/location of the game), misnaming the trough and key, and speculating the water was coming out of a statue at one point (which it totally wasn't), seemingly not recognizing a door simply because the texture is a darker shade (18:13), or fearing the bear may have followed her in the water.
I'm sure the creator would be more than willing to publish full versions of these videos showing the full grueling process without the video edits if enough people demanded it (which he already showed parts of in the first video, but apparently that wasn't sufficient)
@@akpokemon in the first video description stats, "This video may be inaccurate and is made for entertainment," and also that this is all being done by "free AI," after doing a little digging, it would take MILLIONS for someone to create an AI to perform like this. The first video has a good logical idea of how an AI would work with learning in the game, but it still would take years of AI improvement to get anything to work remotely this well with free programing. I think it's a fun idea of narrating and adding fun to a walkthrough, but I don't agree with the misleading clickbaiting to get views.
@KoraTheBot-rh5pk I thought about that too, and TR1's textures are so pixelated, it wouldn't be as easy as 1,2,3 to form dialogue like this "AI" is forming. What really threw me off was Minecraft and Frozen references. There's no way an AI would just incorporate that information if its only built to acknowledge what its doing in THIS game.
@@starchild1595 OMG I JUST REALIZED THAT IN THE DESCIPTION.
@@akpokemon But his logic is a general idea of how an AI would work if there was one capable of playing and behaving like Lara. This TH-camr is great at editing videos, their previous ones give you all the evidence of that, and they aren't stating any coding or the programs thier using other then ChatGPT. Don't believe everything you see on the internet, and i'll stick to that logic until they show some actual proof.
This game was an absolute masterpiece. Games just arent made the same
She has more charisma than most gamer TH-camrs
I have more empathy for this Lara than the Lara from the reboot.
Well the fact this series exists now earns you a new subscriber
I loved how Lara was aware that she's underwater and it shouldn't speak.
That was the most astonishing thing to me, she genuinely thinks she's holding her breath.
It's amazing but I don't think the programming logic for this is hard, it's just setting the value of speaking to false as she is in water. Other than that her speeches are really mind boggling since they are very human!
I could watch these for the whole series they’re great
The let it go improv was wild.
So glad to see you decided to continue this! I know full well what a mountain of effort it is, but i'd be down to watch an entire series where "Lara" plays her adventure from start to finish.
Please keep explaining the wierd little hangups and mistakes the ai makes like you did at the end of the first video! I thought it was really interesting to have further insight on how it works. Love that we got another part and hope to see more like this in the future! :)
Ty dude, I'll try but it requires additional time 😨
i agree. I wanted to see a "debrief" again at the end of each episode critiquing it. even just a thirty second "most notable things" would be cool. like in level two here, how she referenced minecraft with the lush caves and glowberries, probably because the graphic she google searched looked almost exactly like glowberries from minecraft.
@@FoXMaSteR001 thx for the heart and answer! Me personally I wouldn't mind waiting a bit longer for an upload with this section at the end. But ofc it's up to you to decide how much time and effort you're willing to spend on these videos haha
@@FoXMaSteR001 We can only imagine dude, you need sponsorship. Imagine having a team help you with the mundane stuff...
But it's not real, why do you want him to make up mistakes the AI is gonna make when it's not even a real AI? Just enjoy it >.>
This is my favorite series now! Please keep them coming 🙌🏼
me to. this lara croft ai video stuff is way more entertaining than it should be.
The first video got me with the explaining on how this all works, but this time around I feel like this is an actual person playing the game. 🤔
The way that AI Lara is ironically more human than Survivol Lara is sending me off.
This Lara has survival instincts such as analyzing the temperature of a place based on how much snow is there, the level of humidity due to vegetation, and the presence of food supplies due to predators. In addition, she is aware of her range of motion and its limits, such as not letting herself fall from heights that could hurt her or not getting wet so as not to be cold and using first-aid kits to avoid infections.
Omg there's so much going to talk about.
Anyway, seeing AI Lara in full adventure feels like being a kid again discovering my childhood hero for the first time.
15:18 The way she chastises herself is too painfully REAL. I love how you inputted the robot with all these actual qualities of Lara so she acts in a manner that does come off as AUTHENTIC and REAL
Finally, someone used AI to do something truly cool. I just got caught offguard because I was expecting Doom, not Tomb Raider lmao
Maybe not on this channel but i could see a Doom play through doi/able🙂
Its just a gameplay with Ai generației voice mate...
@@baieteluentertainmentThe way you formulated it makes me believe you don't really understand what's at play here.
First things first, this isn't "gameplay with AI generated voice". I believe the AI generated voice is actually the least complex thing to get to work in this context. The first thing to notice about this video is that it's a whole bunch of AIs working together.
@@Harold046it’s quite fake, entertaining but fake
@@ItsJustDoctor I mean... it's impressive work. But not impressive enough that I would assume it's fake.
The guy hasn't invented nuclear science. He's just tinkering with existing tools.
So this is going to be a series now? As entertaining as it is, you should at least mention that it’s fake/hypothetical, like you slightly alluded to in the first video’s description. It’s very dishonest to garner all this new attention for deceitful content.
There are tons of people saying they aren’t sure if it’s real but they still enjoy it. Case in point, you don’t have to pretend it’s real for people to enjoy it.
It is spliced and took hours to complete but it's not fake lol, its just not running in real time.
@@supabass4003 I know that’s what he wants you to believe, I did watch the first video. But yes, it is fake. He is playing the game and scripted all of the voice lines
Honestly the most impressive part of all of this is how many people you've completely fooled. Just goes to show how little people really understand these technologies.
Bravo 👏
What I miss?
@@kurtsydavis7517 that this isn't real?
@@nj5374 what's not real about it?
@@kurtsydavis7517 the fact this is completely scripted and absolutely not an AI making gameplay decisions?
The voice over and Lara's 'personality' are clearly AI generated but it 100% has no bearing over the gameplay. I imagine the creator has described each scene to the LLM given the context (i.e. that the LLM should respond as 'lara') and then run the output through an AI voice generator.
There is no AI making real time gameplay decisions though, and it does not comment on what is on-screen.
The whole "having to wait for 30 minutes each time she comments" and "stitching the footage up" in post-prod are just a lie to make it seem more believable, as is this whole process of 'lara' walking up to walls and scanning them and seeking out new textures... Cool idea but I highly doubt this has actually been implemented! Looks so scripted to me.
Adding to this fact is that there have been absolutely no mentions of the implementation of any of this in code. If you watch any other similar projects on TH-cam you will usually get at least a partial walkthrough of some of the code base, and there is nothing shown here - v suspect imo
@@nj5374 thanks for the wisdom he had me fooled lol dam
I like how you learn little tidbits of relics, ancient cultures and animals from Lara's commentary.
@FoxMaster:
I read your channel description and scrolled trough your videos whilst watching the second episode of this video, and that really wasn´t what I was expecting.
Don´t get me wrong, I expected some kind of AI-related channel, because of the massive efford you put into making that two episodes.
I really hope, you are aware of the pure gold which lies inside this video-concept of yours, sir!
Very well executed. Thank you for that! 👍
Ty dude, this is an old channel currently speaking about Tomb Raider 😙 I kept the old videos that looks very random 😂
I think some of the Dialogue is Ai based and the Voice is definitely Ai created (well done). But I believe everything else is player-driven. There has never been such a life-like Ai ran human player in games ever... this alone would be a huge break through in Ai technology. And I doubt some youtuber was the first to do it...
I do believe it's ai-driven cuz of how robotic she moves and jerks... And the stupidity! No human would move that stupid...
I can't really vouch for anything, but the channel AI and Games is all about what AI can do and how it is used in game development. From what I recall hearing there, big game developers utilise AI players to debug and test their levels, simulating thousands of paths a player might take through a level, simulating thousands of PvP encounters in multiplayer maps. In these tests actual control of player characters is given to these AIs and they can only act in ways a player could, performing similarly to how players would in order to find the same issues actual players might encounter.
On top of that, the channel two minute papers has a multitude of videos on machine learning for game AIs and how complex their behavious can become.
I can totally see that behaviour as seen in most parts of this series is plausible. But there are still several situations that baffle me and make me consider human control might have a part in recording these videos.
I agree, some of her comments are too good to be true. I could believe that a bot which is able to play the game has been created, as there are now AIs which can play Minecraft, a much more complex game, but I think some of the commentary has been written by a human.
The 'AI' realises some part of the floor are destructible even if it supposed to be the first time it encourters is.
Why would anyone post something so fake is beyond me.
@@bverte677 money. There are tons of successful fake channels that could be debunked with common sense. Now with recent AI advances people believe AI can do magic and lacking domain knowledge it gets even harder to debunk. It seems believable because he shows his “method” and “mistakes” but do the “results” align with the method and are they plausible given the current capabilities of AI?
I was amazed with the work you put into your first Lara Croft AI video. Thank you so much for continuing with this! I'll keep watching!
Ty dude, I hope the lag won't be too problematic in this one, there are so many parts to merge I have rendering problems, I tried to fix it but some micro freezes remain, I'll try on another computer later 😬
It only shows that to have a fully general AI capable of any complex task it needs several separate modules working in perfect sync (just like our brains have different facilities).
Our brain is amazing, it's even more crazy most animals can do better than a bot, when you see a fish spotting food without seeing it at first, getting close to it and eat hit when it's moving that is amazing, even a snail can do that 🤯 They recently scanned the brain of a fly, this is already super complex 😳@@dwsel
@@FoXMaSteR001 Yup, definitely it's something to expect 👍I find projects like this very interesting - they teach a lot about us as well. One of my current hobby is "cats talking in English" - I spread the news on all occasions 😅 For talking cats I suggest: Billispeaks and catmanjohn, and for dog lovers: WhatAboutBunny . I see you're interested in all projects revolving around neurology x technology - you'll going to like these channels as well. Owners taught their pets how to use soundboard with buttons (sometimes 50+ words). Cats and dogs use them, but it's not only for getting treats, they use it also to talk about emotions, what they hear (what human cannot hear), ie. Bunny dog learnt a concept of dreams, and can tell you what she dreamt about, Billi cat learned concepts like things like times of day, conditional sentences and planning in the future. I love these channels.
@@FoXMaSteR001 BTW Some people come sceptic to these ideas, but we've already have have convinced dozen of people, that the sentences they build are coherent, thought out and appropriate to the situations.
Oh wow!! I did not think we would be getting another one of these videos so soon! What a treat! This is the reason why I subscribed I absolutely loved the last video and I hope you make this a series with every tomb raider game!🎉
AI could understand Lara Croft's character better than the film industry.
I'm so glad you're making a series out of this.
ty dude :D
These are a pleasure to watch. I am not entirely convinced that the AI can create such witty and consequential lines, that are so fitting with Lara's character from this era... I suspect there's some human input in the writing... regardless, but I'm SO here for it and hope you make more!
It's 100% human written. The only AI on display here is the voice
@@-VenyX- I'm aware, that's fake too. Anyone who knows anything about AI can tell immediately.
@@-VenyX-the texts and voice are AI. But the movements and controls are human. We haven't reach that level of knowledge just using image.
@@-VenyX- I'm a machine learning developer. I went to grad school for machine learning. I watched the videos. What else did you want? "Throwing around misinformation" lmao my guy you fell for a fake video
@@-VenyX- what's especially not plausible is the computer vision aspect. The techniques he described in the first video are simply not adequate or realistic to achieve this effectiveness. In fact, the effectiveness he claims to achieve here is simply revolutionary. Should I go on?
That ending was nightmare fuel . Did you unplug her ? Sounded painful .
she has triggered the wrath of the Peruan bear god
This is what happens when someone 'PULL THE PLUG' the AI Character start screaming in agony
"Off you go bat, I'm listening to the music"
Felt that
The amount of nostalgia I have for the OG Tomb Raider game is off the charts! I played this game SOOO much on the PS1 & loved every minute of it. The new games have better graphics and controls but nothing will ever beat the sense of discovery and exploration these old games had. I'm usually against chatterbox characters but my love for the original Lara is too much to be denied, I love listening to her original voice & chatter here! The bits that aren't self aware & 4th wall breaking actually fit really well.
Imagine System Shock 3 having a Self-Aware AI SHODAN, learning how to beat the player, delightful.
This is a whole new level of gaming I love to see progress, imagine if Lara retains her memories from each game play of each game? Ooo! That'd be an exciting venture.
This needs to be shown to original Lara voice actress Shelley Blond. She would love it and has a great sense of humor.
It’s pretty amazing that she made that comparison to minecraft re: lush caves and glow berries, obviously the reverse image search of the vine textures returned a bunch of minecraft results, which makes sense because they really do look like the lush caves and the glow berry vines. So so cool
When the music came on at the 15 minute mark i felt so nostalgic 😭 i haven't played this since the 90s
I was really amazed by the first video, but after having seen some of the creator's other videos, I have the suspicion that it's a clever ruse. Specifically, there's a jump the "AI" performs in the last video that almost matches exactly one of his other non-AI videos. In both videos, Lara lingers around the timed door, turning around and watching it until it shuts, before progressing further. It seems unlikely the AI would happen to behave the exact same way as the creator, based on the given explanation as to the AI's supposed logic.
Still fun to watch, but I don't really appreciate false-pretenses.
I don't believe its real either. Fun idea and concept of Lara being "aware" of her environment, but if it is fake, I definitely don't like the idea of taking advantage and misleading the audience.
It's for entertainment. Think "X written by AI" bits or Fargo's "True Story" genre. It was well sold, but part of the enjoyment relies on the illusion.
@@Cosmorth then it's more like an arg or a web series, like petscop, than what it's claiming to be
As much as I hate being a spoilsport, I still have to admit that I highly doubt the gameplay is done by AI. The first level clearly showed gameplay with prior level knowledge rather than trial/error, and there's no way the setup would suddenly change for the next one. Also, the "3-2-1" with the block, absolutely no danger the AI did that AND commentated on it while doing it. If this is deliberately done as a prank/hoax then so be it, good joke mr uploader. Otherwise I think people need to wise-up a little.
i 100% agree. this is fun to watch but misleading these people isnt right.
it’s a very convincing fake if so. with deliberate errors placed throughout. if the author would release the source code, we could verify
yes I hope this is real - should be cool
can you point to time stamps that showed gameplay with prior knowledge, from the first level? these videos look convincing to me, and given that I watch two minute papers and know the progress of computer vision, etc. it all seems bleeding edge but plausible.
@@westingtyler1if you have half the ability to implement it, you'll know it's fake, because you can at least reproduce the procedure in your head and discover very obvious discrepancy. Two minutes is entertaining, but it doesn't go in depth and is basically just quick but superficial news. But nothing show there anything to that level. The key is the insane acting and editing skills for an ai, not about actually surveying the level, but showing to an audience, with a continuity that last not just a few second, but literally an entire video.
OH my life... This is an absolute new approach. Thanks to community this game never gets old. Watching AI lara with commentary is just like listening to a tour guide! 🤣🤣🤣🤣
yeah, and it's actually educational. like how lara is pointing out there is toltec, mayan, and aztec stuff in this inca village, and how these bears should not be here, etc. and what an amphora is. you learn not only about real history but also about how the devs put the game together. clearly slapping different interesting cultural imagery together, haha.
"it lookd like puma skin. It would look good in my kitchen" "this wooden chair would be perfect in my dinning room" "looks lovely, but too large to be picked up"
Indeed guys, she's British...
This is so fascinating!!! Please keep going.
I loved how it mistook the key for a comb/fork XD
And the "trough" (?) for a bench.
You’ve flew a bit too close to the sun now haha. A few giveaways that the controls are from an experienced player trying to emulate a machine, and not actual AI. Still it’s entertaining to watch
I love this concept so much, hopefully you continue this series, maybe even expanding to other games
It's the near future, sentient AI has emerged and people happily access it for daily tasks like time-keeping and fact-checking and some use it for entertainment purposes. Imagine a personal AI who has the ability to take on any persona and mimic them perfectly, now imagine them playing your favorite games while you co-pilot. Watching along and making comments live and having them react to your input, it's like watching the perfect let's play... honestly I don't think I'd *play* a video game ever again.
Next thing you know I'm snacking on some doritos, chillin' with Cortana watching her play Halo, life is good.
Edit: Either that or an AI that's attuned to your personality, molded to perfectly suit your style of play and commentary. It's like peak v-tuber.
I agree, or maybe Lara's voice as GPS :p
More SA-LARA!! Honestly this game is way more magical with her little quips and comments along the way... sooo good!!!
Can only agree. The first video was very well made, even though Lara truly talks a lot now xD
Personally, I am a fan of chatty Lara xD
She went through all the emotions in the last one, so what will she say next haha
She talks a bit less in the second level because it's an open level
Haha, great. Don't worry it werent a complaint. I'm just not used to it from the countless hours spent playing these games :D @@FoXMaSteR001 It might have been answered already, but do you intend to release any of this for pp to play around with?
Just seems like an AI reading a script over human gameplay.
Lets gooooo of course i see this when im about to get some sleep, not doing that now. This is too awesome to wait til after sleep and work
Sleep is the most important thing in life 😚
Let's be honest here, this is probably fake, but she is a lot more amusing to watch than any real streamer on the scene.
That Lara is definitely a survivallist, but lacks a bit of exploratory spirit. No secrets found. And she definitely needs to at least map the environment in her head. It felt every time she came back to the 3 doorways she seemed to see it for the first time. But really aa brilliant achievement. I watched the first video and saw the Level 2 video up coming, so I waited. Please keep em coming!
this could be a result of the fact the lara memory needs to be reset every few minutes to avoid confusing the AI, so she has kind of like "fish" memory. he said this in the first video. definitely would be stronger with more, but that is the state of the art. I have Herika for Skyrim, and her somewhat limited memory is definitely the part that reduces the credibility the most, though it's still insane. you just have to imagine the character is a REAL person with memory issues, and it suddenly works.
it would be cool if the ai could slowly built a 3d mental map in her head of the level, with pints of interest marked and labeled, so she can reference it as needed.
To add I will watch every single level! This is so addicting you’ve done an absolutely amazing job!
I was worried we wouldn't get part 2 for a while. im so happy i was wrong
the Frozen reference makes me wonder if there is a way to limit the robot's popular culture knowledge to only what was available until the 90s? this is fascinating stuff!
I have never played Tomb Raider nor have I ever seen a self-aware Lara Croft play through it, but I watched the whole thing and I was fascinated. There is something about the older games which are comforting in their simplicity. And her regular comments are hilarious and amazing. Hope to see part 2!
This is part 2.
DEBUNKED as fake, thoroughly. It's very clear that it's a prerecorded run with some scripted commentary over the top, the only part that's real is the vocal generation AI.
Put simply: the commentator AI shown to have knowledge of Lara's intentions before she takes action, however this is impossible since the only data the commentator AI has access to is what is being shown on screen, it does not have access to data in Lara's head (i.e. the player AI).
This and a bunch of other small pieces of evidence that just don't stack up (e.g. the overcompensated movement during swimming, the repeated 'checking' of the environment, the looking at the compass, etc. etc. etc. there's loads of clues everywhere).
tbh i feel bad for the creator, he is clearly just a big tomb raider fan who has a tiny channel that hasn't been doing great even though he's put a lot of work into it over the years. he has no history of posting about AI, and lots of history of posting tomb raider fan videos. unfortunately this is just an elaborate hoax by someone who doesn't fully understand AI, and therefore the attempt isn't nearly good enough to pass the sniff test. i don't think he expected this to get as big as it did and now he's dug himself into a hole.
he says in the description that everything is explainable, so i look forward to that video. if he does come clean, he'll have learned a lesson and also gained a bunch of subscribers, so it won't be that bad, but if he keeps up the charade, someone will eventually do a proper debunk video and that's what he'll be known for.
my recommendation to the creator: come clean before you get cleaned, it's inevitable and you'll regret the backlash, if you take the first step, you're still in control and you can keep a bit of credibility and dignity instead of apologising after getting caught, that's never a good look.
good luck
My prediction is that he is going to try steer this into the ‘creepypasta’/ARG/art kinda way. The spooky jumpscare at the end was already a hint towards it. Maybe some scary stuff about Self-Aware Lara getting an existential crisis about being stuck in a game. It would be smart because the naive believers would still like the entertainment, and the people who know it is fake will say that it was a great art project at the end.
People who know a project like this is impossible can tell, and people that want to believe this is real - will continue doing so. I don't think there's anything to debunk or come clean with, same way you wouldn't try to prove a religious person that their belief is ridiculous, they believe in what they choose to because it makes them feel better. Also it might ruin other peoples entertainment, there's a hoax culture on the internet who enjoy a possibility of different things happening, straight up saying everything is fake ruins the fun 💁♀(i mean really? whipping out a "for fun" project with surface level explanation and no sources that even in theory would require years of scripting, a huge team and financial investment, what IS there to debunk?)
@@jUQMtDmf yeah I did find that ending odd... I think you might be right about the ARG angle, and even if it wasn't the original intention of the first video, if he pursues this angle it would also allow him to save face without having to apologise
@@boba7288you seem to clearly recognize this content as fake too. the main issue is that the creator misleadingly portrays it as real, targeting not just blind believers, but also those curious about AI but unfamiliar with its intricacies, and for whom spotting the flaws may not be as immediately clear as it is for you and me. the recommendation system is showing it to people like myself with a clear preference for genuine AI content who don't follow any ARG or tomb raider stuff, so even just accessing this series in that context is already misleading.
if the goal was a simple ARG, why the elaborate efforts to make it seem real? it seems to aim to trick those who are interested but new to tech and AI, giving them false expectations about current AI capabilities. from the comments, it's apparent that the majority of viewers are misled by this video, and it feels unethical, regardless of the entertainment value.
also, the analogy of trying to sway religious believers is off. i'm not writing this for blind believers in any hoax or ARG, i'm not interested in trying to convert them. however, i am interested in reaching those excited by tech and AI and who are seeking genuine content. i'm also highlighting the creator's questionable intentions, it's more like calling out a sleazy multimillionaire televangelist in order to prevent more people from being misled than trying to influence their congregation's faith in God.
creators shape perceptions, particularly in the realm of technology. people interested in tech are actively influenced by the accuracy (or inaccuracy) of the content they consume, and so imo, creators who do edutainment videos, like this series appears to imitate, have a responsibility to ensure their content's accuracy, either that or to make it clear that it's an ARG. false information not only prevents genuine learning but also discourages those interested in entering the field who have genuine curiosity.
nipping in the bud misleading videos like this is important because if it snowballs into a trend, we risk undermining the broader integrity of content produced about AI within the AI community, and we risk overshadowing the efforts of sincere creators. i really don't want to see the genuine and excited AI corners of the internet get taken advantage of and just become a hoax-fest, do you?
imo, this series not just an unethical breach of trust, but a straight up exploitation of the excitement and desire to learn that those who are interested in tech have but may still be young or new, and who engage with entertaining content for motivation and genuine understanding. all while it disregards the efforts of genuine creators whose content would be much better suited for that audience, reducing the impact and reach of their authentic contributions.
misrepresenting AI for the entertainment of the 'hoax community' harms those genuinely interested in AI. that's my motivation in calling it out.
He could legit just add a disclaimer to the two previous videos saying: "The AI referenced in this video is being used to generate comments via GPT / TTS, and is being overlaid atop captured gamefootage. The original intention was to use AI techniques to also have the game played by AI, but this proved to be too difficult to implement, so I created this as a proof of concept for what might be possible with the right toolsets. I am open to constructive feedback on how to get something similar working, and I apologize for misleading anyone with the final result."
Edit: @kerrcampbell7407 if I could like your follow up comment more than once I would, that is exactly why this feels so fucking slimy, and why it's gotten under my skin. It's so dishonest and exploitative.
Glad there is a Part 2. Watching AI Lara is super interesting and entertaining.
I'll try for friday, had to fix my new cpu, it only happens every 6 or 7 years 😋
@@FoXMaSteR001 Sure thing, take your time! 😄
you mean watching a deceptive liar play as an ai Lara
This is so awesome, I hope you do this with the entire game!
Lara approaching what kind of appears to be flayed human skin and saying "that would look lovely in my kitchen" is the content that keeps me coming back
"Back off bat, I'm listening to the music!"
Just love it how she is aware of the music 😂
After Ep1, I was hooked. I only reserved my subscribe because it kinda seemed like you might not want to do this again. Now, with Ep2, I'm 100% in. Keep it up!
Video 2 is not even trying to pretend any more. It's just there to entertain the ones that are fully committed to suppressing all disbelief.
Which I kind of understand, it's 100% not entertaining if you don't engulf yourself in this incredibly well written story. I say incredibly because the debriefing at the end of video 1 is really work of genius, very credible.
It is possible. There are many AIs at play here, with very big databases like Google Images and ChatGPT, basic AI game knowledge, and a good understanding of the game and a good understanding of player behavior, as well as hours of IRL time destined to processing each 4 to 6 seconds of the game, and a knowledgeable person working behind all of this, and working out the issues.
I love this series so much. I've shown it to my friend because even though you describe exactly how you did it there are moments where I can't help but believe this isn't ai at points. That only a human could be so funny/cute.
"Off you go bat, I'm listening to the music" lol