@@CODM100x i mean, it's kinda the point. it would be really weird for google to NOT dominate generative video. I still can't get their 2.0 flash to respond with more than 100 lines of code so anthropic is still beating everyone there.
@@homewardboundphotos Except Chat GPT o1 and o1-mini, which are arguably (significantly? sufficiently?) better at coding, and have *crazy* long max. token output length.
@@CraicDealerEire I guess you could say Open AI is the Apple of AI. Doesn't do anything particularly better but did set the whole thing in motion and will continue to put out decent products.
That's really true, google are specialist at selling stuff, and ....... and ...... and...... and..... and..... (still waiting) and...... and........ 😴😴😴
@@selpharessecret3899 Sora is 10 times worse than whatever edited and fake stuff they showcased months ago. Veo is literally generation better than the rest right now. And it generates in 4k and 4 videos at a time.
The data is public. Otherwise they would have built a better llm model from the get go. Whoever has the best ai researchers will win, and that's not always google. In fact Elon tends to have top notch engineers in his companies and that makes him a dark horse
The Ultimate AI Showdown is (not) Over. One thing about the industry we seem to forget is that it's always a temporary victory, and each company will play leapfrog for first place. Google is the winner of the moment. We'll see what happens on the next cycle.
I’m sure it’s in their terms for TH-camrs that they can use TH-cam videos for training etc. they just don’t want others to train off their data. So I would actually expect them to train off TH-cam videos
They were only lagging in AI consumer products. Which literally did not even exist as a category before Chad GPT launched. They had every other AI advantage and were always ahead of everybody else.
@@PseudoProphet Right.... you know I was trying to think back to when I could use google with natural language query ...... I keep think it was around 2009 that i could pretty much just it a question as I would ask you, and to see the LLM's of today makes me question as to how far back that the government agencies like DARPA... Its all pretty bizarre when you think about it. The cross roads that we are standing at now is almost unfathomable.
Google will win this easily . Google has data , funds, then they’re leading in quantum research too. They can easily offer Veo2 for free to users for years and destroy any competition
I wish man. But I have nearly 30 friends working there and noone has confidence in leadership. They have the talent that can make that happen but the nonsense politics make Meta or openAI far more effective despite them having 1/10th of the talent. If they completely get rid of their leadership, culture change will have google win the AI race in under 6 months
Gemini is a game-changer for voice chat! It's so natural, it's like talking to a real person. Unlike other AI, Gemini only asks questions when it needs to, and they're always spot-on. I was hooked after just one try! GPT and Copilot constantly interrupt with unnecessary questions, ruining my flow. With Gemini, I don't have that distraction. It's the closest I've come to having a genuine conversation with an AI.
Was messing around with gemini flash's pc viewing last night. Really strange to actively have a conversation with it about the things happening on the screen.
I don't know about the pro users, but Chat GPT has yielded me far better creative writing results than Gemini 2.0 Flash or Claude. Its memory function helps a lot developing a concept or a project. Though, Gemini could be better at prioritizing some things, like when I asked it a question that hinted at a specific audience, it offered a response taking that into consideration, offering me insights into the nature of the audience. Chat GPT failed to do so.
“Not everything in time most everyone “ “Why fake a faker, only to be fooled into believing we are not faked into believing we are not already.” Best practice is to be truthful or silent. “Listen long enough and surrounded by truthfulness… one can hear a pin drop that has a lie attached to it round the world “ Take care Jeremy
Everything is advancing so quickly in the AI space, with everyone striving to outdo one another. It’ll be interesting to see how things shape up once the holidays pass and the dust settles.
Two things - first, Sora hasn't been released yet, only Sora Turbo, and secondly, have you or anyone you know actually used Veo 2 ? Ahh, that'll be a no....
I don't know why everyone likes Claude; for me, Claude's response is so robotic. It gives information about what I ask, but the way it responds, I don't like. I like ChatGPT and Gemini responses better. Gemini sounds human to me.
I got so annoyed with Gemini refusing to answer anything even remotely political. I even asked who the first president of the United States was and it wouldn't answer. Does it still do this?
@@CrankyLawyer I normally don't ask the question related to politics. Maybe it is restricted on politics; I don't know why they have to do that. But for ChatGPT, it can answer.
@@BorSam , yup ChatGPT does politics but always has silly caveats etc. But Gemini seriously will not tell you who the first (or current) President of the United States is
@@BorSam , I just checked again and it will say who the current president is but still won't tell me the first president, or second. Lol Edit: but it It won't tell me who won the 2020 election even though it has told me Biden is the president. 😂😂
A person will need an AI just to figure out which versions of which Google AI services are where and when/if they are available in their locality. It's a giant stew of half-finished/half-released splatter. Also, it's easy to claim Google's AI is number one when you don't include OpenAI's top model (o1/o1 Pro) in the benchmarks.
The thing that actually matters is which model can users actually access? I cancelled my Gemini account shortly before the big release because I can't actually use the stuff they apparently have.
“We have no moat, neither does OpenAI” - Google. But I wouldn’t be so quick to write off OpenAI though. They’re sitting on a huge release. If you’ve been paying attention to their livestreams, there are lots of hints they’ve been throwing.
Sora is coming home for Christmas is not a rumor. Those humans are torn between affirmation and desire. This conflict occurs when affirmations are used superficially without addressing underlying emotions or when desires aren't fully acknowledged.
Google has a long history of showing vaporwear that never materializes. Until this is released for everyone to try themselves, assume this is all smoke and mirrors.
Ai can create realistic and such good videos but they havent managed to make realistic and proper images yet. Aleast most mainstream companies who have their products released.
Google has extensive resources and unparalleled expertise, making it unrealistic to assume they aren't actively working on something innovative and impactful.
The problem is that if Google gets the top spot for too long, it will just stagnate and try to sell more ads. This new type of AI-driven search is so much more expensive than the plain Google and it will deteriorate quickly to some sponsored ad-serving tool.
Googles access to videodata means it had training data that was next level, I don't think this translates into them being ahead in the game as the architecture underpinning these things is what actually determines who is at the bleeding tip of the spear.
Y'all keep forgetting Google invented the GPT architecture. Remember 2 years ago when Microsoft bought openAI Satya Nadela, Microsoft CEO didn't even believe openAI had what it takes to dethrone google, his exact words were, "we are going to make Google dance and when they dance, remember we made them dance" I guess Google finally danced. Lol
There is no king - and we want it to stay this way. The market is incredibly dynamic. It's more like a horse race. The competitors are jockeying for position. Competitive markets are great for us consumers.
I’m not gonna watch this video, but titles like this really bug me. Remember when TH-camrs were so sure Meta was gonna go bankrupt after all their failed Metaverse attempts? Or when Gemini couldn’t get it right and everyone was quick to say it was the end for Google? As long as these companies are still operating, it’s NEVER gonna be over.
Really??!! Over?!! O1 is by far the most intelligent model so far. And Sora is out, while Veo is…. A DEMO. Like all the previous demos that resulted to be fake in the past. I really don’t get it. You guys may not like Sam Altman, you can pick any side you want on the game of thrones AI drama… but that Open AI is in the lead is JUST A FACT. This may change quite soon, who knows? But I really don’t get the need of people Le of being dishonest.
In terms of capabilities at this point in time sure, Google seems good, however, these models are too censured and being big tech they are not trusted. Being censured and not trusted with private information means that anyone planning to design the next electric car, won't be using these models except maybe in school. Meta is still the model which I think will power innovation.
the reason i shifted to gemini advanced is that they have so much to offer. like 2tb drive space. phone integration with my apps and with the ok google voice prompt. and i saw it was faster to getting results than open ai.
GEMINI - ಗೇमिनि (two go, grow and move, down) Gē (ಗೇ):- 1) [verb] to do, perform (an action etc.) to carry out; to fulfil. 2) [verb] to cultivate a) to prepare and use (soil or land) for growing crops; to till; b) to break up the surface soil around (plants) in order to destroy weeds, prevent crusting, and preserve moisture; c) to grow (plants or crops) from seeds, bulbs, shoots, etc. Mi - Mi (मि). mīmati) 1) To go, move. 2) To sound. Derivable forms: mīm (मीम्). Ni(नि) ni, meaning down, lower. As above so below (Ge Mi Ni) Fulfill till nurture growth to go, move to sound all derivable forms meaning on the down low. - sanskrit
GEMINI - twins Gem in eye - precious to see g em in aye - gather them in, yes Jim aNd I - Jim, a masculine name that means "supplanter." - supplant - seize, or replace, Phonics J em N eye - hookem faithfulness see ge many - Gee many, gay many (many, gay / happy) ge mini - carry small Twins precious to see, gather them in yes. My replacement and myself who can faithfully see many happy small carriers. -babble
These Google language models are so assumptive and don't see nuance in your prompts. Grok 2 is really the only model that can kind of see nuance. And OpenAI is just behind them. ...these blind tests are meaningless as far as I'm concerned.
I have gotten precisely NOTHING useful out of Sora. It's SO MUCH worse than I was expecting that it's kind of breathtaking. If I have to regenerate a video 50 times to get something that merely doesn't suck, there's no value in it. Same goes for every other generative video platform I've tried, including Runway. The time and expense of generating hundreds of videos to get something only kinda usable would be better spent just shooting something for real or using more traditional CGI. Look no further than that Coke commercial - it represents the very best a team of professionals with lots of money can do and it can't stand up to the slightest scrutiny. All that said, I'm very curious to see what Veo 2 can do once I'm able to get my hands on it. Hope springs eternal...
Curious what the budget was for that Coke ad, because the result looked incredibly bad. I've definitely seen shorts where generative video should've been used as part of a supplemental workflow rather than the whole thing, and it would've been so much better. Looking forward to trying Veo 2 when it's more accessible.
I'm really impressed with the physic depiction Veo2 displayed it's really good. I don't think they train the same way as SORA did, bcoz training with existing videos won't really taught the model about real world physic interaction.
I like the OpenAI API-I have no issues using it. However, Gemini is definitely a very interesting alternative. I need to test it further in production.
Neither Google AI models nor those of Open AI can answer correctly the following questions designed for CS: PLANNING A CRUISE SCHEDULE A cruise line is scheduling 7 week-long voyages for the ship Freedom. Each voyage will occur in exactly one of the first 7 weeks 1 through 7. Each voyage will be to exactly one of the 4 destination Guadeloupe, Jamaica, Martinique, or Trinidad. Each of the destination will be scheduled for at least one of the weeks. The following conditions apply to Freedom's schedule Jamaica will not be its destination in week 4 Trinidad will be its destination week 7 Freedom will make exactly 2 voyages to Martinique, and at least 1 voyage to Guadeloupe will occur in some week between those 2 voyages Guadeloupe will be its destination in the week proceeding any voyage it makes to Jamaica No destination will be scheduled for consecutive weeks 1. Which of the following is an acceptable schedule of destinations for Freedom in order from week 1 through week 7? A. Guadeloupe, Jamaica, Martinique, Trinidad, Guadeloupe, Martinique, Trinidad B. Guadeloupe, Martinique, Trinidad, Martinique, Guadeloupe, Jamaica, Trinidad C. Jamaica, Martinique, Guadeloupe, Martinique, Guadeloupe, Jamaica, Trinidad D. Martinique, Trinidad, Guadeloupe, Jamaica, Martinique, Guadeloupe, Trinidad"
E. Martinique, Trinidad, Guadeloupe, Trinidad, Guadeloupe, Jamaica, Martinique" -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Here is Google AI model's answer: To determine which schedule is acceptable, let's apply the given conditions: Jamaica will not be its destination in week 4. Trinidad will be its destination in week 7. Freedom will make exactly 2 voyages to Martinique, and at least 1 voyage to Guadeloupe will occur in some week between those 2 voyages. Guadeloupe will be its destination in the week preceding any voyage it makes to Jamaica. No destination will be scheduled for consecutive weeks. Now, let's analyze each option based on these conditions: A. Guadeloupe, Jamaica, Martinique, Trinidad, Guadeloupe, Martinique, Trinidad Week 4 is Jamaica (valid). Week 7 is Trinidad (valid). 2 Martinique voyages (valid). Guadeloupe (week 1) precedes Jamaica (week 2) (valid). No consecutive weeks with the same destination (valid). Acceptable. B. Guadeloupe, Martinique, Trinidad, Martinique, Guadeloupe, Jamaica, Trinidad Week 4 is Jamaica (valid). Week 7 is Trinidad (valid). 2 Martinique voyages (valid). Guadeloupe (week 1) precedes Jamaica (week 6) (valid). No consecutive weeks with the same destination (valid). Acceptable. C. Jamaica, Martinique, Guadeloupe, Martinique, Guadeloupe, Jamaica, Trinidad Week 4 is Jamaica (invalid). Week 7 is Trinidad (valid). 2 Martinique voyages (valid). Guadeloupe precedes Jamaica (week 5) (valid). No consecutive weeks with the same destination (valid). Not Acceptable (due to Jamaica in week 4). D. Martinique, Trinidad, Guadeloupe, Jamaica, Martinique, Guadeloupe, Trinidad Week 4 is Jamaica (invalid). Week 7 is Trinidad (valid). 2 Martinique voyages (valid). Guadeloupe (week 3) precedes Jamaica (week 4) (invalid). No consecutive weeks with the same destination (valid). Not Acceptable (due to Jamaica in week 4 and Guadeloupe preceding Jamaica). E. Martinique, Trinidad, Guadeloupe, Trinidad, Guadeloupe, Jamaica, Martinique Week 4 is Jamaica (invalid). Week 7 is Trinidad (valid). 2 Martinique voyages (valid). Guadeloupe (week 3) precedes Jamaica (week 6) (valid). No consecutive weeks with the same destination (valid). Not Acceptable (due to Jamaica in week 4). Conclusion The acceptable schedules are: A. Guadeloupe, Jamaica, Martinique, Trinidad, Guadeloupe, Martinique, Trinidad B. Guadeloupe, Martinique, Trinidad, Martinique, Guadeloupe, Jamaica, Trinidad And here is Open AI model's answer: Let’s analyze the given options step by step, ensuring all the conditions are satisfied. Key Conditions: Jamaica is not the destination in Week 4. Trinidad is the destination in Week 7. Exactly 2 voyages to Martinique occur. At least 1 voyage to Guadeloupe occurs between the 2 voyages to Martinique. Guadeloupe is the destination in the week preceding any voyage to Jamaica. No destination is scheduled for consecutive weeks. Analysis of Each Option: Option A: Guadeloupe, Jamaica, Martinique, Trinidad, Guadeloupe, Martinique, Trinidad Week 4: Trinidad (Satisfies condition 1). Week 7: Trinidad (Satisfies condition 2). Martinique appears twice, and Guadeloupe is scheduled between them (Satisfies conditions 3 and 4). Guadeloupe precedes Jamaica (Satisfies condition 5). No consecutive weeks have the same destination (Satisfies condition 6). This schedule satisfies all conditions. Option B: Guadeloupe, Martinique, Trinidad, Martinique, Guadeloupe, Jamaica, Trinidad Week 4: Martinique (Satisfies condition 1). Week 7: Trinidad (Satisfies condition 2). Martinique appears twice, and Guadeloupe is scheduled between them (Satisfies conditions 3 and 4). Guadeloupe precedes Jamaica (Satisfies condition 5). No consecutive weeks have the same destination (Satisfies condition 6). This schedule satisfies all conditions. Option C: Jamaica, Martinique, Guadeloupe, Martinique, Guadeloupe, Jamaica, Trinidad Week 4: Martinique (Satisfies condition 1). Week 7: Trinidad (Satisfies condition 2). Martinique appears twice, and Guadeloupe is scheduled between them (Satisfies conditions 3 and 4). Guadeloupe precedes Jamaica (Satisfies condition 5). No consecutive weeks have the same destination (Satisfies condition 6). This schedule satisfies all conditions. Option D: Martinique, Trinidad, Guadeloupe, Jamaica, Martinique, Guadeloupe, Trinidad Week 4: Jamaica (Violates condition 1, as Jamaica cannot be the destination in Week 4). This schedule is invalid. Option E: Martinique, Trinidad, Guadeloupe, Trinidad, Guadeloupe, Jamaica, Martinique Week 4: Trinidad (Satisfies condition 1). Week 7: Martinique (Violates condition 2, as Trinidad must be the destination in Week 7). This schedule is invalid. Correct Answers: A, B, C ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Both of the answers of the 2 different AI models are not entirely correct because schedules A and B do not meet the requirement that at least 1 voyage to Guadeloupe will occur in some week between the 2 voyages to Martinique, although Open AI's model seems to have got a partially correct answer, it does not recognise that the English preposition "between" is correctly used to indicate one's position between 2 parties only when there is no other party in the ways to the 2 parties of reference. Ex: Party B is between party A and D: A-------------- B ------------------D but if Part C is either in the way from party B to party A, or in the way to party B to party D, then it will be no longer correct to use the preposition "between" to express the same position because it will become between party A and C A--------------B -------C or part C and D: C-------B-----------------D At the 1st glance at the 5 schedules A, B, C, D, and E, one can see straight away that schedule E is invalid with no voyage too Trinidad in week 7, which violates the 2 condition, and schedule D is also invalid with a voyage to Jamaica in week 4. At the 2nd look at the 3 left schedules, A, B, C, one can see that no voyage to Guadeloupe between 2 voyages to Martinique in schedule B, which violates the 3rd condition, and thus , it is not valid either. Looking at schedule A, one can see that, a voyage to Guadeloupe appears too be between 2 voyages to Martinique, but looking at it more closely will show that the voyage to Guadeloupe is not exactly between the 2 voyages to Martinique because a voyage to Trinidad is right after the 1st voyage to Martinique. Hence, schedule A does not satisfy the 3rd condition either, and thus, it is not valid either. Now, only schedule C is left, which meets all the conditions, with no voyage to Jamaica in week 4, a voyage to Trinidad in week 7, a voyage to Guadeloupe in week 3 b/t the 1st voyage to Martinique in week 2, and the 2nd voyage to Martinique in week 4, and with a voyage to Guadeloupe in week 5 which proceeds week 6 in which a voyage is made to Jamaica, and there is no destination which is scheduled for 2 consecutive weeks in this schedule either. Therefore, schedule C is the only correct answer to the question
I agree with this video. I find that the current gemini in Google search out performs chatgpt in software instructions such as, how to do things in fl studio.
Once they restructured the AI departments at Google and put Demis in charge it seemed like it'd be a matter of time. Google had ignored client facing LLMs and standalone AI in general. Deepmind haven't tackled a problem they haven't been able to come out of on top in so far.
Google is a sales company, the only reason this comparison works is because most people don't understand this is the SORA Turbo, not SORA proper. The day Google doesn't release their AI first, is the day they've beaten OpenAI - until then they need to have their best compared to the previous version of OpenAI.
This is just a pet peeve but you need to stop saying Gemini experience, especially when one of your graphics shows it to be called Gemini experimental at 5:14.
I tried Gemini 2 Flash just yesterday with a coding optimization, first response had hallucinations, the model just made something up with was obviously not true from the code that I provided then in the second response hallucinated again with something else. And this was often my experience when I tried some new model .. it just makes stuff up with confidence, on the other hand I use Claude 3.5 Sonnet daily for programming and it rarely hallucinates, it's extremely useful. I also tried the Imagen3 my first experience was a disappointment I wanted a similar image with the same style of the image I have included and the outputted image didn't have similar style at all, and I didn't even liked it. Obviously this is only a very small sample size, but I am not very compelled to keep trying when it always keep disappointing me, I don't get how the hell it scores on these benchmarks and leaderboards.
thanks for your videos, very helpful. a humble comment...I understand that exaggerated headlines may attrack viewers, but over time diminish credibility.
Not just google. Chinese deepseek r1 is as good as o1, and Chinese video generators like Hunyuan video are better than sora. But Google are even better for now.
Well to be fair, google DEFINITELY used every video on TH-cam to train this model. No other company has that level of readily available videos.
Not the point
@@CODM100x i mean, it's kinda the point. it would be really weird for google to NOT dominate generative video. I still can't get their 2.0 flash to respond with more than 100 lines of code so anthropic is still beating everyone there.
@@homewardboundphotos Except Chat GPT o1 and o1-mini, which are arguably (significantly? sufficiently?) better at coding, and have *crazy* long max. token output length.
Where do you think Open AI is winning/will win? If anthropic for code, Google for video
@@CraicDealerEire I guess you could say Open AI is the Apple of AI. Doesn't do anything particularly better but did set the whole thing in motion and will continue to put out decent products.
Until Veo2 is released to the public, it is just fiction
Sora really was a fiction and it's already been proved to everybody.
That's really true, google are specialist at selling stuff, and ....... and ...... and...... and..... and..... (still waiting) and...... and........ 😴😴😴
Veed to
@@PseudoProphet unless you are in the EU...
@@selpharessecret3899 Sora is 10 times worse than whatever edited and fake stuff they showcased months ago.
Veo is literally generation better than the rest right now.
And it generates in 4k and 4 videos at a time.
Google has ALL the data, CUSTOM AI CHIPS (TPUs), BILLIONS OF DOLLARS to burn, a whole lot of talent, and so on? It was inevitable.
Google maps alone is a trillion dollar behemoth imo
And logical...
The data is public. Otherwise they would have built a better llm model from the get go. Whoever has the best ai researchers will win, and that's not always google. In fact Elon tends to have top notch engineers in his companies and that makes him a dark horse
But still can't fucking figure it out captcha. (sarcasm)
The Ultimate AI Showdown is (not) Over. One thing about the industry we seem to forget is that it's always a temporary victory, and each company will play leapfrog for first place. Google is the winner of the moment. We'll see what happens on the next cycle.
Facts!
we have 3 days of openai left they still can pull a rabbit
Why did you put the word “not” in brackets? (Strange)
Google is way bigger company with more really good engineers, I doubt that OpenAi has a chance
Results of the brain drain as people flee the liar with the vocal fry. People are still necessary.
I expected google to take the lead, but not that fast.
Google is strong
They aren't ahead. They have a long way to go.
Google has 20 years of youtube videos and no one can tell me they didn't use that for training Veo.
I’m sure it’s in their terms for TH-camrs that they can use TH-cam videos for training etc. they just don’t want others to train off their data. So I would actually expect them to train off TH-cam videos
"Showdown is OVER" That's overexaggerated.
It always is
On the internet? Can't believe it 🤣
its just begun
It’s not game over until they actually release a product
lol true - and how about the pricing when it is out? Didn't bother to watch the video.
Google finally has a product I might pay real money for
Google has TH-cam. They should win this race easily. I was surprised last year when I saw that they were lagging.
They were only lagging in AI consumer products. Which literally did not even exist as a category before Chad GPT launched.
They had every other AI advantage and were always ahead of everybody else.
Right they have more pure training data then anyone in the world.
@@PseudoProphet Right.... you know I was trying to think back to when I could use google with natural language query ...... I keep think it was around 2009 that i could pretty much just it a question as I would ask you, and to see the LLM's of today makes me question as to how far back that the government agencies like DARPA... Its all pretty bizarre when you think about it. The cross roads that we are standing at now is almost unfathomable.
They have DeepMind, they definitely have enough firepower to lead the field
google has youtube and they design their own tensor processors. openai is dependent on nvidia graphics cards
Google will win this easily . Google has data , funds, then they’re leading in quantum research too. They can easily offer Veo2 for free to users for years and destroy any competition
I wish man. But I have nearly 30 friends working there and noone has confidence in leadership. They have the talent that can make that happen but the nonsense politics make Meta or openAI far more effective despite them having 1/10th of the talent.
If they completely get rid of their leadership, culture change will have google win the AI race in under 6 months
They also have their own chips. And those chips are version 4 now
Google is a monopoly if they accomplish this, they need to be broken up
The Elite Society's Money Manifestation ebook made me realize so much about attracting wealth, it’s insane
Veo2 is better than the CGI in most movies today. Can't tell it isn't real.
Gemini is a game-changer for voice chat! It's so natural, it's like talking to a real person. Unlike other AI, Gemini only asks questions when it needs to, and they're always spot-on. I was hooked after just one try!
GPT and Copilot constantly interrupt with unnecessary questions, ruining my flow. With Gemini, I don't have that distraction. It's the closest I've come to having a genuine conversation with an AI.
Perplexity voice is quite good.
Was messing around with gemini flash's pc viewing last night. Really strange to actively have a conversation with it about the things happening on the screen.
I dont know w Gemini are u using, but ChatGPT is by far better
To make absolute (definitely ??) statements in AI is quite bold. Just a way to get readers to see the channel
Google has TH-cam as its data source. Great advantage
I don't know about the pro users, but Chat GPT has yielded me far better creative writing results than Gemini 2.0 Flash or Claude. Its memory function helps a lot developing a concept or a project.
Though, Gemini could be better at prioritizing some things, like when I asked it a question that hinted at a specific audience, it offered a response taking that into consideration, offering me insights into the nature of the audience. Chat GPT failed to do so.
Did you adjust Gemini’s style how to answer? Makes a massive difference. 😊
Google has youtube.
I've had Gemini research 2.0 working on something for me for 10 days now, the longer it takes, the more excited I'm getting.
Seriously? That's cool! How did you get it?
10 days...? lol. on a single prompt?
na your laptop is frozen 😂
We can fake anything...
“Not everything in time most everyone “
“Why fake a faker, only to be fooled into believing we are not faked into believing we are not already.”
Best practice is to be truthful or silent.
“Listen long enough and surrounded by truthfulness… one can hear a pin drop that has a lie attached to it round the world “
Take care
Jeremy
@@Jeremy-Aicool story
There's no way in hell I'm coming back to Chrome for their AI addons. I hope they come to Firefox.
Haha, Firefox is pretty much a dead man walking. If the Justice Department has its way, Google will be forced to stop funding Firefox.
Good luck. Chrome is soooo yummy.
What’s firefox? Feels like an edgelord browser
Everything is advancing so quickly in the AI space, with everyone striving to outdo one another. It’ll be interesting to see how things shape up once the holidays pass and the dust settles.
What does the holidays have anything to do
Two things - first, Sora hasn't been released yet, only Sora Turbo, and secondly, have you or anyone you know actually used Veo 2 ? Ahh, that'll be a no....
I have access to Veo 2. Anything you want me to try?
yea there are PPL who got access to veo 2
Google is better. OpenAI is just a rush.
Ignorant comment
@@rigellroy Great. What's your feeling about it...
Creating videos is a tiny portion of what AI should do for mankind. Let’s see how this plays out.
I don't know why everyone likes Claude; for me, Claude's response is so robotic. It gives information about what I ask, but the way it responds, I don't like. I like ChatGPT and Gemini responses better. Gemini sounds human to me.
I got so annoyed with Gemini refusing to answer anything even remotely political. I even asked who the first president of the United States was and it wouldn't answer. Does it still do this?
@@CrankyLawyer I normally don't ask the question related to politics. Maybe it is restricted on politics; I don't know why they have to do that. But for ChatGPT, it can answer.
@@BorSam , yup ChatGPT does politics but always has silly caveats etc. But Gemini seriously will not tell you who the first (or current) President of the United States is
@@BorSam , I just checked again and it will say who the current president is but still won't tell me the first president, or second. Lol
Edit: but it It won't tell me who won the 2020 election even though it has told me Biden is the president. 😂😂
And Claude's usage limits are crazy
A person will need an AI just to figure out which versions of which Google AI services are where and when/if they are available in their locality. It's a giant stew of half-finished/half-released splatter. Also, it's easy to claim Google's AI is number one when you don't include OpenAI's top model (o1/o1 Pro) in the benchmarks.
The thing that actually matters is which model can users actually access? I cancelled my Gemini account shortly before the big release because I can't actually use the stuff they apparently have.
“We have no moat, neither does OpenAI” - Google.
But I wouldn’t be so quick to write off OpenAI though. They’re sitting on a huge release. If you’ve been paying attention to their livestreams, there are lots of hints they’ve been throwing.
Sora is coming home for Christmas is not a rumor. Those humans are torn between affirmation and desire.
This conflict occurs when affirmations are used superficially without addressing underlying emotions or when desires aren't fully acknowledged.
Google has a long history of showing vaporwear that never materializes. Until this is released for everyone to try themselves, assume this is all smoke and mirrors.
Ai can create realistic and such good videos but they havent managed to make realistic and proper images yet. Aleast most mainstream companies who have their products released.
Google has extensive resources and unparalleled expertise, making it unrealistic to assume they aren't actively working on something innovative and impactful.
The problem is that if Google gets the top spot for too long, it will just stagnate and try to sell more ads. This new type of AI-driven search is so much more expensive than the plain Google and it will deteriorate quickly to some sponsored ad-serving tool.
Google is killing it currently... but I still prefer Chatgpt over Gemini as an AI Chatbot 😊
ChatGPT is like ICQ back in the days. 😅
That car's dashboard instruments are not at all correct. It is downright bizarre.
Googles access to videodata means it had training data that was next level, I don't think this translates into them being ahead in the game as the architecture underpinning these things is what actually determines who is at the bleeding tip of the spear.
Anyone else notice the 13.4 percent decrease in long context?? Probably one of the most important factors in an LLM.
False, this probably requires too much compute to be released to customers.
Same thing with Sora, you are using the nerfed Sora turbo.
Y'all keep forgetting Google invented the GPT architecture. Remember 2 years ago when Microsoft bought openAI Satya Nadela, Microsoft CEO didn't even believe openAI had what it takes to dethrone google, his exact words were, "we are going to make Google dance and when they dance, remember we made them dance" I guess Google finally danced. Lol
Sora was showcased in February 2024. What tech they have cooking for Sora 2?
OpenAI is the Lycos search engine of AI. Google is Google.
There is no king - and we want it to stay this way. The market is incredibly dynamic. It's more like a horse race. The competitors are jockeying for position. Competitive markets are great for us consumers.
Except video generation is a totally niche use case that has near zero general business value.
ads?
I have always expected for Google to be better at some point. They have way more data for many years.
200 dollars for Sora is a fkn joke
Google's got more cash, more talent (especially since OpenAI lost some of its talents), and way more data from the intire internet.
I’m not gonna watch this video, but titles like this really bug me. Remember when TH-camrs were so sure Meta was gonna go bankrupt after all their failed Metaverse attempts? Or when Gemini couldn’t get it right and everyone was quick to say it was the end for Google? As long as these companies are still operating, it’s NEVER gonna be over.
Google is superior!
Really??!! Over?!! O1 is by far the most intelligent model so far. And Sora is out, while Veo is…. A DEMO. Like all the previous demos that resulted to be fake in the past.
I really don’t get it. You guys may not like Sam Altman, you can pick any side you want on the game of thrones AI drama… but that Open AI is in the lead is JUST A FACT.
This may change quite soon, who knows? But I really don’t get the need of people
Le of being dishonest.
No matter how good the product, I wont go out of my way to use a google product. Im already halfway out the door on TH-cam.
In terms of capabilities at this point in time sure, Google seems good, however, these models are too censured and being big tech they are not trusted. Being censured and not trusted with private information means that anyone planning to design the next electric car, won't be using these models except maybe in school. Meta is still the model which I think will power innovation.
🇧🇷🇧🇷🇧🇷🇧🇷👏🏻, Google is impressive!
the reason i shifted to gemini advanced is that they have so much to offer. like 2tb drive space. phone integration with my apps and with the ok google voice prompt. and i saw it was faster to getting results than open ai.
Again, we saw cherry-picked videos, it was the same when Sora was first announced
Wasn’t cherry picked at all trust me on this one
They posted the limitations on their own page which is a lot more honest than what other companies have done
Maybe OAI needs to announce more days of Christmas?
Everything is moving so fast with AI and everone trying to one up the other. It will be interesting after the holidays and the dust settles
The race is on. 🏁💰
Not gonna lie, open ai presenters on day 10 did look upset. 😂
Google is the googol of the universe. The name Gemini and being in Gemini as this happens is synchronicity, no coincidence.
What do you mean “being in Gemini as this happens”?
GEMINI - ಗೇमिनि (two go, grow and move, down)
Gē (ಗೇ):-
1) [verb] to do, perform (an action etc.) to carry out; to fulfil.
2) [verb] to cultivate a) to prepare and use (soil or land) for growing crops; to till; b) to break up the surface soil around (plants) in order to destroy weeds, prevent crusting, and preserve moisture; c) to grow (plants or crops) from seeds, bulbs, shoots, etc.
Mi - Mi (मि).
mīmati) 1) To go, move. 2) To sound. Derivable forms: mīm (मीम्).
Ni(नि) ni, meaning down, lower.
As above so below (Ge Mi Ni)
Fulfill till nurture growth to go, move to sound all derivable forms meaning on the down low.
- sanskrit
GEMINI - twins
Gem in eye - precious to see
g em in aye - gather them in, yes
Jim aNd I - Jim, a masculine name that means "supplanter." - supplant - seize, or replace,
Phonics
J em N eye - hookem faithfulness see
ge many - Gee many, gay many (many, gay / happy)
ge mini - carry small
Twins precious to see, gather them in yes.
My replacement and myself who can faithfully see
many happy small carriers.
-babble
@@joecaves6235 lol… ok….
@@dkae-designs He means the position of the cosmos. Like astrology / astronomy
These Google language models are so assumptive and don't see nuance in your prompts. Grok 2 is really the only model that can kind of see nuance. And OpenAI is just behind them. ...these blind tests are meaningless as far as I'm concerned.
Gemini (2.0 flash experimental) still fails STRAWBERRY (miserably) and cannot do my good ol' pianokeyboard.
I have gotten precisely NOTHING useful out of Sora. It's SO MUCH worse than I was expecting that it's kind of breathtaking. If I have to regenerate a video 50 times to get something that merely doesn't suck, there's no value in it. Same goes for every other generative video platform I've tried, including Runway. The time and expense of generating hundreds of videos to get something only kinda usable would be better spent just shooting something for real or using more traditional CGI. Look no further than that Coke commercial - it represents the very best a team of professionals with lots of money can do and it can't stand up to the slightest scrutiny.
All that said, I'm very curious to see what Veo 2 can do once I'm able to get my hands on it. Hope springs eternal...
Curious what the budget was for that Coke ad, because the result looked incredibly bad. I've definitely seen shorts where generative video should've been used as part of a supplemental workflow rather than the whole thing, and it would've been so much better. Looking forward to trying Veo 2 when it's more accessible.
I tried to get Google to write some very simple code for me, it failed! So I don't see how it snatches anything!
I'm really impressed with the physic depiction Veo2 displayed it's really good. I don't think they train the same way as SORA did, bcoz training with existing videos won't really taught the model about real world physic interaction.
Has the general public actually used Sora?
Finally non uncally valley ai generated video
They need to integrate all this into android OS insteadnof having each AI tool on a seperate island.
Google collecting data will be very dangerous.
I like the OpenAI API-I have no issues using it. However, Gemini is definitely a very interesting alternative. I need to test it further in production.
Talk in numbers man! Don’t give us hyped generalities!
What’s the smart glass shown in the video? Is it a product of Google empowered with Gemini 2.0?
Let’s see what happens since many Chinese AI ones defeated SORA
Every company that I know of cherry picks their examples...
I would assume it's better, but how much better really? If it even is..
Neither Google AI models nor those of Open AI can answer correctly the following questions
designed for CS:
PLANNING A CRUISE SCHEDULE
A cruise line is scheduling 7 week-long voyages for the ship Freedom. Each voyage will occur in exactly one of the first 7 weeks 1 through 7. Each voyage will be to exactly one of the 4 destination Guadeloupe, Jamaica, Martinique, or Trinidad.
Each of the destination will be scheduled for at least one of the weeks. The following conditions apply to Freedom's schedule
Jamaica will not be its destination in week 4
Trinidad will be its destination week 7
Freedom will make exactly 2 voyages to Martinique,
and at least 1 voyage to Guadeloupe will occur in some week between those 2 voyages
Guadeloupe will be its destination in the week proceeding any voyage it makes to Jamaica
No destination will be scheduled for consecutive weeks
1. Which of the following is an acceptable schedule of destinations for Freedom in order from week 1 through week 7?
A. Guadeloupe, Jamaica, Martinique, Trinidad, Guadeloupe, Martinique, Trinidad
B. Guadeloupe, Martinique, Trinidad, Martinique, Guadeloupe, Jamaica, Trinidad
C. Jamaica, Martinique, Guadeloupe, Martinique, Guadeloupe, Jamaica, Trinidad
D. Martinique, Trinidad, Guadeloupe, Jamaica, Martinique, Guadeloupe, Trinidad"
E. Martinique, Trinidad, Guadeloupe, Trinidad, Guadeloupe, Jamaica, Martinique"
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Here is Google AI model's answer:
To determine which schedule is acceptable, let's apply the given conditions:
Jamaica will not be its destination in week 4.
Trinidad will be its destination in week 7.
Freedom will make exactly 2 voyages to Martinique, and at least 1 voyage to Guadeloupe will occur in some week between those 2 voyages.
Guadeloupe will be its destination in the week preceding any voyage it makes to Jamaica.
No destination will be scheduled for consecutive weeks.
Now, let's analyze each option based on these conditions:
A. Guadeloupe, Jamaica, Martinique, Trinidad, Guadeloupe, Martinique, Trinidad
Week 4 is Jamaica (valid).
Week 7 is Trinidad (valid).
2 Martinique voyages (valid).
Guadeloupe (week 1) precedes Jamaica (week 2) (valid).
No consecutive weeks with the same destination (valid).
Acceptable.
B. Guadeloupe, Martinique, Trinidad, Martinique, Guadeloupe, Jamaica, Trinidad
Week 4 is Jamaica (valid).
Week 7 is Trinidad (valid).
2 Martinique voyages (valid).
Guadeloupe (week 1) precedes Jamaica (week 6) (valid).
No consecutive weeks with the same destination (valid).
Acceptable.
C. Jamaica, Martinique, Guadeloupe, Martinique, Guadeloupe, Jamaica, Trinidad
Week 4 is Jamaica (invalid).
Week 7 is Trinidad (valid).
2 Martinique voyages (valid).
Guadeloupe precedes Jamaica (week 5) (valid).
No consecutive weeks with the same destination (valid).
Not Acceptable (due to Jamaica in week 4).
D. Martinique, Trinidad, Guadeloupe, Jamaica, Martinique, Guadeloupe, Trinidad
Week 4 is Jamaica (invalid).
Week 7 is Trinidad (valid).
2 Martinique voyages (valid).
Guadeloupe (week 3) precedes Jamaica (week 4) (invalid).
No consecutive weeks with the same destination (valid).
Not Acceptable (due to Jamaica in week 4 and Guadeloupe preceding Jamaica).
E. Martinique, Trinidad, Guadeloupe, Trinidad, Guadeloupe, Jamaica, Martinique
Week 4 is Jamaica (invalid).
Week 7 is Trinidad (valid).
2 Martinique voyages (valid).
Guadeloupe (week 3) precedes Jamaica (week 6) (valid).
No consecutive weeks with the same destination (valid).
Not Acceptable (due to Jamaica in week 4).
Conclusion
The acceptable schedules are:
A. Guadeloupe, Jamaica, Martinique, Trinidad, Guadeloupe, Martinique, Trinidad
B. Guadeloupe, Martinique, Trinidad, Martinique, Guadeloupe, Jamaica, Trinidad
And here is Open AI model's answer:
Let’s analyze the given options step by step, ensuring all the conditions are satisfied.
Key Conditions:
Jamaica is not the destination in Week 4.
Trinidad is the destination in Week 7.
Exactly 2 voyages to Martinique occur.
At least 1 voyage to Guadeloupe occurs between the 2 voyages to Martinique.
Guadeloupe is the destination in the week preceding any voyage to Jamaica.
No destination is scheduled for consecutive weeks.
Analysis of Each Option:
Option A:
Guadeloupe, Jamaica, Martinique, Trinidad, Guadeloupe, Martinique, Trinidad
Week 4: Trinidad (Satisfies condition 1).
Week 7: Trinidad (Satisfies condition 2).
Martinique appears twice, and Guadeloupe is scheduled between them (Satisfies conditions 3 and 4).
Guadeloupe precedes Jamaica (Satisfies condition 5).
No consecutive weeks have the same destination (Satisfies condition 6).
This schedule satisfies all conditions.
Option B:
Guadeloupe, Martinique, Trinidad, Martinique, Guadeloupe, Jamaica, Trinidad
Week 4: Martinique (Satisfies condition 1).
Week 7: Trinidad (Satisfies condition 2).
Martinique appears twice, and Guadeloupe is scheduled between them (Satisfies conditions 3 and 4).
Guadeloupe precedes Jamaica (Satisfies condition 5).
No consecutive weeks have the same destination (Satisfies condition 6).
This schedule satisfies all conditions.
Option C:
Jamaica, Martinique, Guadeloupe, Martinique, Guadeloupe, Jamaica, Trinidad
Week 4: Martinique (Satisfies condition 1).
Week 7: Trinidad (Satisfies condition 2).
Martinique appears twice, and Guadeloupe is scheduled between them (Satisfies conditions 3 and 4).
Guadeloupe precedes Jamaica (Satisfies condition 5).
No consecutive weeks have the same destination (Satisfies condition 6).
This schedule satisfies all conditions.
Option D:
Martinique, Trinidad, Guadeloupe, Jamaica, Martinique, Guadeloupe, Trinidad
Week 4: Jamaica (Violates condition 1, as Jamaica cannot be the destination in Week 4).
This schedule is invalid.
Option E:
Martinique, Trinidad, Guadeloupe, Trinidad, Guadeloupe, Jamaica, Martinique
Week 4: Trinidad (Satisfies condition 1).
Week 7: Martinique (Violates condition 2, as Trinidad must be the destination in Week 7).
This schedule is invalid.
Correct Answers:
A, B, C
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Both of the answers of the 2 different AI models are not entirely correct because schedules A and B do not
meet the requirement that at least 1 voyage to Guadeloupe will occur in some week between the 2 voyages to
Martinique, although Open AI's model seems to have got a partially correct answer, it does not recognise that
the English preposition "between" is correctly used to indicate one's position between 2 parties only when
there is no other party in the ways to the 2 parties of reference.
Ex:
Party B is between party A and D: A-------------- B ------------------D
but if Part C is either in the way from party B to party A, or in the way to party B to party D, then it will be no longer
correct to use the preposition "between" to express the same position because it will become between party A and C
A--------------B -------C or part C and D: C-------B-----------------D
At the 1st glance at the 5 schedules A, B, C, D, and E, one can see straight away that schedule E is invalid with
no voyage too Trinidad in week 7, which violates the 2 condition, and schedule D is also invalid with a voyage
to Jamaica in week 4.
At the 2nd look at the 3 left schedules, A, B, C, one can see that no voyage to Guadeloupe between 2 voyages to
Martinique in schedule B, which violates the 3rd condition, and thus , it is not valid either.
Looking at schedule A, one can see that, a voyage to Guadeloupe appears too be between 2 voyages to
Martinique, but looking at it more closely will show that the voyage to Guadeloupe is not exactly between the 2
voyages to Martinique because a voyage to Trinidad is right after the 1st voyage to Martinique.
Hence, schedule A does not satisfy the 3rd condition either, and thus, it is not valid either.
Now, only schedule C is left, which meets all the conditions, with no voyage to Jamaica in week 4, a voyage to
Trinidad in week 7, a voyage to Guadeloupe in week 3 b/t the 1st voyage to Martinique in week 2, and the 2nd
voyage to Martinique in week 4, and with a voyage to Guadeloupe in week 5 which proceeds week 6
in which a voyage is made to Jamaica, and there is no destination which is scheduled for 2 consecutive weeks
in this schedule either. Therefore, schedule C is the only correct answer to the question
I agree with this video. I find that the current gemini in Google search out performs chatgpt in software instructions such as, how to do things in fl studio.
Once they restructured the AI departments at Google and put Demis in charge it seemed like it'd be a matter of time. Google had ignored client facing LLMs and standalone AI in general. Deepmind haven't tackled a problem they haven't been able to come out of on top in so far.
I’ve been thinking of the same thing.
Gemini isn't even close to GPT4o, so yeah. Wayyyy overstated dude
Gemini is Woke.
Copilot is Woke.
ChatGPT is Woke.
Google is a sales company, the only reason this comparison works is because most people don't understand this is the SORA Turbo, not SORA proper. The day Google doesn't release their AI first, is the day they've beaten OpenAI - until then they need to have their best compared to the previous version of OpenAI.
Waiting for what SSI is cooking.
I would be happier when open source imagine generators coming out google is way to biased.
It would make sense google owns TH-cam… 🤷🏾♂️
This is just a pet peeve but you need to stop saying Gemini experience, especially when one of your graphics shows it to be called Gemini experimental at 5:14.
if the public can have it for free then that is where my vote is
This battle is worse than Kendrick vs Drake
don't call me weird but all i'm going to use this for is making videos of power lines falling down
To be fair
ChatGPT is widely used than Gemini
Is it a sponsored video by Google ??😊😂
It's developing very fast, it's really amazing technology. 😅
Kling is even better than Sora
youre falling for the same lies as when sora was shown first lol. they, same as openai wont let you use undistilled version
I tried Gemini 2 Flash just yesterday with a coding optimization, first response had hallucinations, the model just made something up with was obviously not true from the code that I provided then in the second response hallucinated again with something else. And this was often my experience when I tried some new model .. it just makes stuff up with confidence, on the other hand I use Claude 3.5 Sonnet daily for programming and it rarely hallucinates, it's extremely useful. I also tried the Imagen3 my first experience was a disappointment I wanted a similar image with the same style of the image I have included and the outputted image didn't have similar style at all, and I didn't even liked it. Obviously this is only a very small sample size, but I am not very compelled to keep trying when it always keep disappointing me, I don't get how the hell it scores on these benchmarks and leaderboards.
Veo2 closed preview vs Sora Turbo (mini). Yeah I'll believe it when it's released
SORA hallucinates
Whatever the one is they put on the phone messages its horrible.
The GPS is good
Why does this guy look like he's always about to cry.
Competition is good; it's welcomed! ;)
Should I Switch my subscription to Gemini?
thanks for your videos, very helpful. a humble comment...I understand that exaggerated headlines may attrack viewers, but over time diminish credibility.
Not just google. Chinese deepseek r1 is as good as o1, and Chinese video generators like Hunyuan video are better than sora. But Google are even better for now.