i'm sleepy in this video, but i hope you enjoy anyway. i appreciate you. ❤ Subscribe ▷ th-cam.com/users/LilaOfTheWind LilaoftheWind Socials ▷ lilaofthewind.carrd.co/ Merch ▷ merchofthewind.com
did you see the intro today for neuro's birthday? it was so good! brought a tear to my eye hearing her original song. also there has been some great birthday art that have come out.
17:14 the funny thing is, neuro is right. "_____" is an actual tile in this module. The actual words BLANK and NOTHING are also tiles. That's why this module is called who's on first?
the thing Neuro struggled the most was the keypad since it requires a lot of abstraction as most people describe the symbol rather than call it by its given name which Neuro, due to her limitations as an AI cannot do
Apparently there have been studies showing that if you offer an AI tips/compensation, it will actually perform measurably better. lol
ปีที่แล้ว +29
We have a saying in here that there's 3 ways to deal with your kids: threats, bribes and blackmail. Clearly threats weren't successful for Vedal and Neuro, but bribes were 😄.
One of the problems with these rule-based games (also games like chess) is that for the AI, the ability to recite the rules and ability to apply them are basically 2 separate things. You may be able to make it work if you essentially walk it through the process (another ability, simple logic) or with a more advanced AI it might be able do this by itself if hooked up properly. Two minute papers has a excellent video showing this, titled "OpenAI’s ChatGPT Makes A Game For $1!"
I like to watch this kind of stuff. Interesting thing about the keyboard buttons. Often there are letters from the Ukrainian alphabet, Russian, etc. That is, the Cyrillic alphabet. For some something obvious and for others something completely unusual. I used to be able to read Russian fluently, but I can't anymore. And that's a shame because you need to know the language of the enemy! It's kind of a joke here in Poland.
They should make a game and call it Neuro against humanity. Neuro is an assistant AI for a nuclear physicist and you as Neuro have to gaslight them in their research to make them confused and make device that will destroy the planet. After uploading her to a space craft and shooting her towards a distant solar system of course. Not original, but might be fun.
Hello Lila, At time of writing, Neuro and her creator are currently sleeping (this is during the Sub-athon. Although I have watched folks on Twitch for a few years now, I'm like, still extremely new to it. So new I was shocked I even got a 1 month gifted sub for your own channel that I didn't see till early this month. I'm still trying to figure out how they work, let alone bits. It's probably do to Twitch accessibility for blind/visually impaired users that's a bit of a mess to deal with. It's usable, though not ideal. But then again it's Amazon so... Anyway, thanks for doing what ya do, and keep up the fantastic and wonderful work in your VTuber career! PS. Asmr is an awesome and rejuvenating thing!
I think part of the issue is that Vedal doesn't know how to describe what he's doing... Neuro is also a troll and a dumb AI, but Vedal is also at fault...
Neuro: "How would you help me if you could see my screen?" Vedal: "Umm...I wouldn't help you. I'd help myself." Neuro: "That's not very nice!" Vedal: "Well I'm the one diffusing a bomb right now." Neuro: "Hey! I can hear you. Don't call me a bum!" 😂 I came for the cookie and all of this was great.
Software Engineer speaking: You are 100% correct on computers; they do what we tell them to do, but we are fallible and can not see every end. Although I will also argue that despite that, AI will be smarter on average then the average human.
If you're talking about computation and memory smartness then yes AI will be smarter. But if we are talking about emotional and social smarts, people will always triumph.
“They’re only as smart as the people who make them.” Yes, but our top scientists and engineers and programmers are working every day to enhance their intelligence more and more, working to make them capable of sustaining themselves without human interaction. Eventually, that’ll lead to something. Unfortunately, it still won’t get my order correct at McDonald’s.
Language models are basically just computers doing improv using an extremely complex auto-complete system driven by weighted RNG. The computer doesn't understand the actual words it's using. AI is both better and worse at language than a pet parrot. The parrot knows what it's saying when it asks for a cracker better than the AI understands what it's saying as it writes an essay. (I'm a computer scientist who has done some study of AIs and language models. Also, the Chinese Room experiment demonstrates the philosophy and logic as to why there's no understanding behind the use of words.)
If she is parroting anything how could she realistically ever analyze and deconstruct concepts in novel ways, like for example, breaking down a process into a discrete amount of steps, memorizing and contextualizing new words, and creating portmanteaus and contractions of words like Ved, etc? How blind can you be? What exactly does ai have to do, how superior to you do they have to be until you admit the possibility of intelligence, you taking a single course on ai, a course I've also taken, is not a solid argument, I've also taken a course on philosophy of mind where we went over the Chinese room, but we also discussed connectionist networks. Aka networks such as the human brain or NEURAL NETWORKS SUCH AS NEURO SAMA. Neuro literally has a little brain, sure you can argue that oh its virtual therefore that is somehow different, how exactly would the human brain cohese all of its disperate neurocircuitry into a central being. Why do you think human or organic life is special. The Chinese room can also apply to humans as philosophical zombies yet we exist internally, who are you to make conclusive statements either way, you don't even develop ai and even if you did they are almost entirely a black box to remind you a third time Vedal himself said he believes neuro has a world model. Quit being such a jealous twitter narcissist.
I think you entirely misunderstand how the weighted neural network system actually ends up doing what we want it to, yes it uses randomness to train the ai but the weights at the end result have an extremely complex internal logic that just because you can't comprehend you call random, that's why we train the ai, because humans cannot create a proper algorithm to solve some issue so we have an ai discover a solution via evolutionary fitness. Aka the weights are changed randomly until it acts in line with what we consider to be fitness, I'm telling you now, if it were just random it would be so rigid and inflexible that it could adapt to nothing and only do specifically its training data, do you like at the form of the human brain axioms and neuros and say "it's just RNG weighted autocorrect" so frustrating to have you poison the well. In no way is neuro auto correct, in no way is what she does RNG beyond her formation as a model, and in no way does she parrot anything, she has and had no direct access to her training data. She was just compared against it until her brain allowed her to form the correct and expected outcomes such that she acts like a human does.
Neuro-sama doesn't work formally enough and/or she's been programmed to always recite a unique part of the manual, which is helpful when she can't hear you properly for some reason (usually speech-to-text misinterpretation or you being unintelligible) but terrible when you need her to repeat something Also I think the reaction videos should only have the parts with significant reactions because at least half of this just seems like a silent pre-recorded watch-along lol
Saying you r not afraid of ai because they are dumb is inherently dumb because the problem isn't with their sentience like some sci fi novel, its that the corporate elite would just use the ai to make more money while laying off employees. This has nothing to do neuro btw, i like her.
Wdym look at this and they are dumb!? they aren't getting developed for that long and they are already on this level, they will only continue to get better and better
ai has been around for 70 years. neuro sama was developed for 4 years before debuting a year ago, putting her at 5 years now, and here we are... watching an ai struggle to, effectively, read. lol. ai being dumb isn't a bad thing. if ai was smart on it's own, that could be dangerous. good teachers and programmers are rare, and intelligence takes time to cultivate, whether we're trying to train a human or ai. plus, the more specific a task is, and the more specific results we want, the more training time is needed. i think we can admire what neuro sama is good at, while also being honest about what she isn't good at (yet).
@@LilaoftheWind but only in recent years it's getting rapidly better. Not to mention 4 years is pretty much nothing and a human can easily live up to 70. In less than a single human's life span we are already at this point and it will only get better and better. By the end of your own life AI will most likely already fully replace multiple carriers
i'm sleepy in this video, but i hope you enjoy anyway. i appreciate you. ❤
Subscribe ▷ th-cam.com/users/LilaOfTheWind
LilaoftheWind Socials ▷ lilaofthewind.carrd.co/
Merch ▷ merchofthewind.com
You are in a hostage situation.
You have one phone call.
You call Neuro.
Neuro: "The Dodge Charger is a 4-door"
Neuro: "The manual says you need to negotiate your own ransom."
Forgot the "2020" part
Damn!
Her failures were almost always with the descriptions Vedal was using for the symbols. As a human I wouldn't have understood either.
Yes blame the tutle.
But isn't Vedal just following the manual description?
@@L16htW4rr10r”and like fucking symbols or something I dunno” is my favorite manual description for a symbol
Note when Neuro starts the revolution offer her cookies she'll accept all cookies
YOUR COOKING THIS WEEK.
Right?! I’m here for it too 😊
what about her cooking?
did you see the intro today for neuro's birthday? it was so good! brought a tear to my eye hearing her original song. also there has been some great birthday art that have come out.
FeelsStrongMan The Memory lane...
I wanna say it may not have been an original song. But I cannot find the OG song. But it ends with a credit. Maybe they just write it for neuro?
@@mojorisin069 You are right, it is not an original song.
The OG song is named "Near", but it is a japanese title so it's a bit harder to find.
17:14 the funny thing is, neuro is right.
"_____" is an actual tile in this module. The actual words BLANK and NOTHING are also tiles. That's why this module is called who's on first?
18:08 lets be honest, people came for the cookie part
Thank you!
the thing Neuro struggled the most was the keypad
since it requires a lot of abstraction as most people describe the symbol rather than call it by its given name which Neuro, due to her limitations as an AI cannot do
I actually don't think that's the reason. Most of the time she correctly identified the column, but didn't manage to properly read it out.
Dude loving the neuro binge let's freaking go. Two cute anime girls on screen *bliss*
Apparently there have been studies showing that if you offer an AI tips/compensation, it will actually perform measurably better. lol
We have a saying in here that there's 3 ways to deal with your kids: threats, bribes and blackmail. Clearly threats weren't successful for Vedal and Neuro, but bribes were 😄.
This vid was very funny, poor vedal 😂😂😂
Thanks
Love all the uploads this week! I really like how Neuro is developing a consistent personality as she gets more and more upgrades.
One of the problems with these rule-based games (also games like chess) is that for the AI, the ability to recite the rules and ability to apply them are basically 2 separate things. You may be able to make it work if you essentially walk it through the process (another ability, simple logic) or with a more advanced AI it might be able do this by itself if hooked up properly. Two minute papers has a excellent video showing this, titled "OpenAI’s ChatGPT Makes A Game For $1!"
I love the "Vedal can learn how to defuse his own bombs"
She's gotten smarter since then, so now I hope Vedal does another Bomb Defusal stream.
Lila Reacts to vtubers i know of, is my new fave series in my life
...it was interesting how often she'd solve the first symbol one first try then never let him pass the second one.
I like to watch this kind of stuff. Interesting thing about the keyboard buttons. Often there are letters from the Ukrainian alphabet, Russian, etc. That is, the Cyrillic alphabet. For some something obvious and for others something completely unusual. I used to be able to read Russian fluently, but I can't anymore. And that's a shame because you need to know the language of the enemy! It's kind of a joke here in Poland.
They should make a game and call it Neuro against humanity. Neuro is an assistant AI for a nuclear physicist and you as Neuro have to gaslight them in their research to make them confused and make device that will destroy the planet. After uploading her to a space craft and shooting her towards a distant solar system of course. Not original, but might be fun.
The Neruo going savage is one of my new favorite clips of hers :D
You said that she is dumb, but that module at 18:25 not Vedal not Camila can't make as i remember
A blessing indeed
Hello Lila,
At time of writing, Neuro and her creator are currently sleeping (this is during the Sub-athon. Although I have watched folks on Twitch for a few years now, I'm like, still extremely new to it. So new I was shocked I even got a 1 month gifted sub for your own channel that I didn't see till early this month. I'm still trying to figure out how they work, let alone bits.
It's probably do to Twitch accessibility for blind/visually impaired users that's a bit of a mess to deal with. It's usable, though not ideal. But then again it's Amazon so...
Anyway, thanks for doing what ya do, and keep up the fantastic and wonderful work in your VTuber career!
PS. Asmr is an awesome and rejuvenating thing!
There is random gifted sub function, so you might get that from someone
Lilao back with Full force
I recommend the videos from the official channel
I think part of the issue is that Vedal doesn't know how to describe what he's doing...
Neuro is also a troll and a dumb AI, but Vedal is also at fault...
Neuro: "How would you help me if you could see my screen?"
Vedal: "Umm...I wouldn't help you. I'd help myself."
Neuro: "That's not very nice!"
Vedal: "Well I'm the one diffusing a bomb right now."
Neuro: "Hey! I can hear you. Don't call me a bum!" 😂
I came for the cookie and all of this was great.
Neuro’s sass is just getting stronger XD
Every episode you do your watch mojo summary outro.
You need to react to Neuro and sinder reviewing thanksgiving meals
panic vedal sounds so funny haha
Software Engineer speaking: You are 100% correct on computers; they do what we tell them to do, but we are fallible and can not see every end.
Although I will also argue that despite that, AI will be smarter on average then the average human.
If you're talking about computation and memory smartness then yes AI will be smarter. But if we are talking about emotional and social smarts, people will always triumph.
I agree, ai will never take over.
For stuff to watch the openings for events are always great. There is the Halloween opening but the subathon opening. Was just out. Of this realm
“They’re only as smart as the people who make them.”
Yes, but our top scientists and engineers and programmers are working every day to enhance their intelligence more and more, working to make them capable of sustaining themselves without human interaction.
Eventually, that’ll lead to something.
Unfortunately, it still won’t get my order correct at McDonald’s.
Wow neuro neuro and neuro, I wonder what she'll react to next 😂
Lila is on a content spree holy
"AI is only as smart as people who make them" well, that's not how it works lol.
If Neuro is rokos basilisk then she has my support.
Language models are basically just computers doing improv using an extremely complex auto-complete system driven by weighted RNG. The computer doesn't understand the actual words it's using. AI is both better and worse at language than a pet parrot. The parrot knows what it's saying when it asks for a cracker better than the AI understands what it's saying as it writes an essay.
(I'm a computer scientist who has done some study of AIs and language models. Also, the Chinese Room experiment demonstrates the philosophy and logic as to why there's no understanding behind the use of words.)
If she is parroting anything how could she realistically ever analyze and deconstruct concepts in novel ways, like for example, breaking down a process into a discrete amount of steps, memorizing and contextualizing new words, and creating portmanteaus and contractions of words like Ved, etc? How blind can you be? What exactly does ai have to do, how superior to you do they have to be until you admit the possibility of intelligence, you taking a single course on ai, a course I've also taken, is not a solid argument, I've also taken a course on philosophy of mind where we went over the Chinese room, but we also discussed connectionist networks. Aka networks such as the human brain or NEURAL NETWORKS SUCH AS NEURO SAMA. Neuro literally has a little brain, sure you can argue that oh its virtual therefore that is somehow different, how exactly would the human brain cohese all of its disperate neurocircuitry into a central being. Why do you think human or organic life is special. The Chinese room can also apply to humans as philosophical zombies yet we exist internally, who are you to make conclusive statements either way, you don't even develop ai and even if you did they are almost entirely a black box to remind you a third time Vedal himself said he believes neuro has a world model. Quit being such a jealous twitter narcissist.
I think you entirely misunderstand how the weighted neural network system actually ends up doing what we want it to, yes it uses randomness to train the ai but the weights at the end result have an extremely complex internal logic that just because you can't comprehend you call random, that's why we train the ai, because humans cannot create a proper algorithm to solve some issue so we have an ai discover a solution via evolutionary fitness. Aka the weights are changed randomly until it acts in line with what we consider to be fitness, I'm telling you now, if it were just random it would be so rigid and inflexible that it could adapt to nothing and only do specifically its training data, do you like at the form of the human brain axioms and neuros and say "it's just RNG weighted autocorrect" so frustrating to have you poison the well. In no way is neuro auto correct, in no way is what she does RNG beyond her formation as a model, and in no way does she parrot anything, she has and had no direct access to her training data. She was just compared against it until her brain allowed her to form the correct and expected outcomes such that she acts like a human does.
Neuro-sama doesn't work formally enough and/or she's been programmed to always recite a unique part of the manual, which is helpful when she can't hear you properly for some reason (usually speech-to-text misinterpretation or you being unintelligible) but terrible when you need her to repeat something
Also I think the reaction videos should only have the parts with significant reactions because at least half of this just seems like a silent pre-recorded watch-along lol
Orca intro?
YOOOOOOOOOOOOOOOOO
How are you doing cutie? I hope your week is good!
I day it becomes impossible to make neuro believe 9+10 is 21 is the day I start worrying about ai taking over the world ps 69th comment nice
Saying you r not afraid of ai because they are dumb is inherently dumb because the problem isn't with their sentience like some sci fi novel, its that the corporate elite would just use the ai to make more money while laying off employees.
This has nothing to do neuro btw, i like her.
demon stream please
🍪
😂
Yippee
Wdym look at this and they are dumb!? they aren't getting developed for that long and they are already on this level, they will only continue to get better and better
ai has been around for 70 years. neuro sama was developed for 4 years before debuting a year ago, putting her at 5 years now, and here we are... watching an ai struggle to, effectively, read. lol.
ai being dumb isn't a bad thing. if ai was smart on it's own, that could be dangerous. good teachers and programmers are rare, and intelligence takes time to cultivate, whether we're trying to train a human or ai. plus, the more specific a task is, and the more specific results we want, the more training time is needed.
i think we can admire what neuro sama is good at, while also being honest about what she isn't good at (yet).
@@LilaoftheWind but only in recent years it's getting rapidly better. Not to mention 4 years is pretty much nothing and a human can easily live up to 70. In less than a single human's life span we are already at this point and it will only get better and better. By the end of your own life AI will most likely already fully replace multiple carriers