I've shared the video thinking that this "real-time input" would be a real feature, but it seems that that's not the case. I never expected that google would clickbait us. Thank you for the video, Luke! Greetings from Brazil!
The latest Gemini demo showed 2 cartoon pictures of a cat, one dead and one alive. Then they put a box next to the pictures and asked Gemini what they reminded it of (or something along those lines) and it said Schrödinger’s cat. I was amazed by that level of abstract thinking, something I haven’t even seen with ChatGPT but now I wonder if they didn’t fake it.
The demo is a bit misleading, but it’s merely a marketing technique to get more than just engineers to see the potential here. It’s really only a couple steps away from something like the final video. The PROBLEM is they didn’t consider their audience carefully. Marketing was targeted a daily folk, but they showed it to engineers. 👏
Completely agree about the marketing technique here and that this is a great demo of where AI can be. After watching this video the first time (of all teh videos) this one motivated me the most; like holy smokes imagine the potential of this thing working in real time with us as we complete tasks... And unfortunately i'm one of those engineers they showed it too 🤣
@@LukeBarousse Gemini is a pile of garbage. We often use Gpt4 in our software company and we tried Bard's Gemini yesterday. It failed miserably on TSQL and C# coding tasks as compared to GPT4. Google hype that people are buying without any critical thinking.
It used to be better, but they nerfed GPT-4 a lot to the point its useless to many of us power users now. Even OpenAI acknowledged this yesterday in a post that they are looking into this issue.
That's by design. They want to claim victory on an unavailable product. If Ultra was for us to test right now I think we would see how they've cherrypicked result and how GPT4 is probably better than Ultra in many scenarios not presented. Remember when they announced palm 2. Just looking at the fan fare you wouldn't guess that the palm 2 powered bard was even still worse than GPT3.5 in many cases. They were saying how it's very good at coding in many languages. But because palm2 was in bard at the time of the demo, within hours this was disproved.
@@elawchess fair point, again we are assuming that google engineer are not as capable as open ai engineers (which is unlikely!). We still have to remember that google is the one that introduced the original transformers paper..... O.o
@@moslehmahamud It's not really about google engineer not being as capable. Open AI have a head start in terms of time. GPT-4 is already running. They are collecting usage details and using that to improve the model. That may be what's making all the difference and google is playing catchup. They have to release a product that people can use and compare and see it matches and beats GPT-4 and simply pointing to the size, history or google and it's engineers isn't going to do the job. Also OpenAI can have a secret way they found that is making the model good, and just because google has good engineers doesn't mean that they would definitely re-invent that method within a year
Great work Luke 👏🏼✨❤️ I have a suggestions if you can tell us about the organizations or any companies that would like to hire junior data analyst... I have completed my Google Data Analytics certificate... But i don't know if there specific companies can hire me... If you can tell us about this i will be grateful...
I respect your perspective on this but I think this is more the case of people not doing a good enough job at following up with the material that orgs put out. The initial disclaimer is vague, I agree, but at 0:22 in their hands-on demo, they include another disclaimer about shortened prompts. I also do not think that reading the description of a video, and the very relevant blog-post attached is too high of a bar for people to cross. Is the video showing an inspiring use-case that isn't actually being used? Yeah, of course. But, if you have critique about this more creative presentation, and consider the paper that partners with this video to be an obfuscation of their methods, I think we'd need to agree to disagree. Just a little disappointed with the lack of diligence on reading what is essentially the Part 2 to the video they put out (the blog post). I don't think Google's scamming us, I think people just aren't like reading what they are putting out there, and that's on the end-viewer imo.
I appreciate you sharing this perspective, as I do think as consumers we do need to dig into the relevant documents. But TBH... the disclaimers could have been WAY more clearer for this video; they could have said "This video recreates some or our favorite interactions with Gemini".... instead they used the word "highlights" vs "recreates"
@@LukeBarousse great point! I probably had a more defensive feeling to the recent feedback of their demo. Not because I love Google or anything. I 100% agree with what you just wrote. And also thanks for making vids for all of us to watch!
Errm google and other companies know about human behavior and that very likely people would jump to conclusions after watching only the video, and they exploited that. It's quite deliberate. You are given google too much benefit of the doubt here @DaveTheChronicler
@@LukeBarousse They could have had someone narrating prior to the demo that in fact still images and text prompts were used but this is just a recreation or simulation using video, and then switch to the demo. Less people would have been misled. It's quite deliberate what they did and i'm happy it's blowing up in their face. They have a really bad habit of misrepresenting their AI results against competition and I've been annoyed since the chess AlphaZero vs Stockfish comparison where they ran stockfish on a machine hundreds of magnitudes smaller than theirs and didn't adjust for that in their results, but were instead running around claiming victory much like they are doing now to OpenAI
I've shared the video thinking that this "real-time input" would be a real feature, but it seems that that's not the case. I never expected that google would clickbait us. Thank you for the video, Luke! Greetings from Brazil!
check out chat-GPT 4o, it's a new mind blowing version of chat-GPT
I saw a couple articles about this and I'm so glad you and Fireship covered this!
Are we really surprised they wouldn't wanna be up front after the Bard fiasco from their last live demo?
I was really pulling for em 😔
@@LukeBarousse same, they're really screwing themselves over in the public perception by doing things like this!
Bummer, thought this was a true breakthrough for Google. Seems like they're showing that Microsoft and OpenAI still have the edge.
I see what you did there! XD
You snuck that one in sweetly.
Ahhh I just saw this Gemini demo today and thought it was absolutely insane! Now you're saying it's probably fake?!!! Ughh🙈🙈
yeah i'm a little disappointed as well 😔
The latest Gemini demo showed 2 cartoon pictures of a cat, one dead and one alive. Then they put a box next to the pictures and asked Gemini what they reminded it of (or something along those lines) and it said Schrödinger’s cat. I was amazed by that level of abstract thinking, something I haven’t even seen with ChatGPT but now I wonder if they didn’t fake it.
The demo is a bit misleading, but it’s merely a marketing technique to get more than just engineers to see the potential here. It’s really only a couple steps away from something like the final video. The PROBLEM is they didn’t consider their audience carefully. Marketing was targeted a daily folk, but they showed it to engineers. 👏
Completely agree about the marketing technique here and that this is a great demo of where AI can be. After watching this video the first time (of all teh videos) this one motivated me the most; like holy smokes imagine the potential of this thing working in real time with us as we complete tasks...
And unfortunately i'm one of those engineers they showed it too 🤣
@@LukeBarousse Gemini is a pile of garbage. We often use Gpt4 in our software company and we tried Bard's Gemini yesterday. It failed miserably on TSQL and C# coding tasks as compared to GPT4. Google hype that people are buying without any critical thinking.
If Google feels it needs to out out a fake video i feek its time ti dump my stock and buy their competitors
Not telling the whole truth is not telling the truth.
That’s a pretty hefty accusation! Clippy could have done all of that!
Bring back clippy!!!!!
Also what about the name Gemini, doesn't seem to fit the concept, except starting with G.
The Google Translate is not working properly at the moment; first, focus on fixing that, and then other projects can follow.
Just marketing. Even chat gpt is a hit or miss in the real life tasks. Google is very behind the competition and they know it.
Great Work Luke!
Great video, honest review of Gemini AI.
So that’s whats going on with the google stocks right now 😂
oh google 🤦🏼♂️
The stock drop down 😁
Google lied a lottttt of times
Fake demo, OpenAI is still better
It used to be better, but they nerfed GPT-4 a lot to the point its useless to many of us power users now. Even OpenAI acknowledged this yesterday in a post that they are looking into this issue.
@@Ghostrider-ul7xn v4 may be worse than it was, but it is still far better than Bard based on my experimentation
@@Ghostrider-ul7xn what are the examples of the use cases where it's now significantly worse.
Ok I think we would have accepted the link with the actual prompts. But still shows it’s still in early faces.
Google losing AI race 😢
Am I the only one who sees a pink panty on his neck instead of that pink shirt?!!
Scandalous data nerd!
i am no google fanboi, but we haven't gotten access to gemini yet. Hard to say if it is bad at this point...
That's by design. They want to claim victory on an unavailable product. If Ultra was for us to test right now I think we would see how they've cherrypicked result and how GPT4 is probably better than Ultra in many scenarios not presented.
Remember when they announced palm 2. Just looking at the fan fare you wouldn't guess that the palm 2 powered bard was even still worse than GPT3.5 in many cases. They were saying how it's very good at coding in many languages. But because palm2 was in bard at the time of the demo, within hours this was disproved.
@@elawchess fair point, again we are assuming that google engineer are not as capable as open ai engineers (which is unlikely!). We still have to remember that google is the one that introduced the original transformers paper..... O.o
@@moslehmahamud It's not really about google engineer not being as capable. Open AI have a head start in terms of time. GPT-4 is already running. They are collecting usage details and using that to improve the model. That may be what's making all the difference and google is playing catchup.
They have to release a product that people can use and compare and see it matches and beats GPT-4 and simply pointing to the size, history or google and it's engineers isn't going to do the job.
Also OpenAI can have a secret way they found that is making the model good, and just because google has good engineers doesn't mean that they would definitely re-invent that method within a year
If it chould do it from a video even if it took along time to process that would have been a great improvement
First time I watch this demo, I already have a doubt about it because I predict this kind of AI will not available so early.
GPT trumps Gemini by far. I hope Ultra is less disappointing
classic marketing goofiness
So, not the father of Skynet? 😢
🤣
stock market manipulation tactics. Shame on them.
Great work Luke 👏🏼✨❤️
I have a suggestions if you can tell us about the organizations or any companies that would like to hire junior data analyst...
I have completed my Google Data Analytics certificate...
But i don't know if there specific companies can hire me...
If you can tell us about this i will be grateful...
I respect your perspective on this but I think this is more the case of people not doing a good enough job at following up with the material that orgs put out.
The initial disclaimer is vague, I agree, but at 0:22 in their hands-on demo, they include another disclaimer about shortened prompts. I also do not think that reading the description of a video, and the very relevant blog-post attached is too high of a bar for people to cross.
Is the video showing an inspiring use-case that isn't actually being used? Yeah, of course. But, if you have critique about this more creative presentation, and consider the paper that partners with this video to be an obfuscation of their methods, I think we'd need to agree to disagree. Just a little disappointed with the lack of diligence on reading what is essentially the Part 2 to the video they put out (the blog post). I don't think Google's scamming us, I think people just aren't like reading what they are putting out there, and that's on the end-viewer imo.
you have too much faith in general viewers lmao
I appreciate you sharing this perspective, as I do think as consumers we do need to dig into the relevant documents.
But TBH... the disclaimers could have been WAY more clearer for this video; they could have said "This video recreates some or our favorite interactions with Gemini".... instead they used the word "highlights" vs "recreates"
@@LukeBarousse great point!
I probably had a more defensive feeling to the recent feedback of their demo. Not because I love Google or anything. I 100% agree with what you just wrote.
And also thanks for making vids for all of us to watch!
Errm google and other companies know about human behavior and that very likely people would jump to conclusions after watching only the video, and they exploited that. It's quite deliberate. You are given google too much benefit of the doubt here @DaveTheChronicler
@@LukeBarousse They could have had someone narrating prior to the demo that in fact still images and text prompts were used but this is just a recreation or simulation using video, and then switch to the demo. Less people would have been misled. It's quite deliberate what they did and i'm happy it's blowing up in their face.
They have a really bad habit of misrepresenting their AI results against competition and I've been annoyed since the chess AlphaZero vs Stockfish comparison where they ran stockfish on a machine hundreds of magnitudes smaller than theirs and didn't adjust for that in their results, but were instead running around claiming victory much like they are doing now to OpenAI
Use water in your hole bro;))
Of course it was fake
Tencent AI is better!
I love your videos but the "data nerds" greeting annoys me every time
Why does it annoy you, fellow data nerd?
@@danre64THAT data nerd fell right into that one!
Sounds like a self hating data nerd