# Googles NEW AI Gemini 2.0 Details Were Just LEAKED! ## Метаданные - **Канал:** TheAIGRID - **YouTube:** https://www.youtube.com/watch?v=cCJFq3Ul0yI - **Дата:** 26.10.2024 - **Длительность:** 11:48 - **Просмотры:** 18,155 ## Описание Prepare for AGI with me - https://www.skool.com/postagiprepardness 🐤 Follow Me on Twitter https://twitter.com/TheAiGrid 🌐 Checkout My website - https://theaigrid.com/ Links From Todays Video: https://www.theverge.com/2024/10/25/24279600/google-next-gemini-ai-model-openai-december Welcome to my channel where i bring you the latest breakthroughs in AI. From deep learning to robotics, i cover it all. My videos offer valuable insights and perspectives that will expand your knowledge and understanding of this rapidly evolving field. Be sure to subscribe and stay updated on my latest videos. Was there anything i missed? (For Business Enquiries) contact@theaigrid.com #LLM #Largelanguagemodel #chatgpt #AI #ArtificialIntelligence #MachineLearning #DeepLearning #NeuralNetworks #Robotics #DataScience ## Содержание ### [0:00](https://www.youtube.com/watch?v=cCJFq3Ul0yI) Segment 1 (00:00 - 05:00) so The Verge has actually come out with another article in which they discuss leaks pertaining to Google's next Frontier Model now this video I think is rather fascinating because when we look at this article it actually talks about some things that we kind of didn't expect from the next iteration of model cycles and I think things are about to change but not in a way that most people do think so you can see here that it says Google plans to announce its next Jam model soon this is of course something thing that we knew was on the horizon given the amount of time that it has been since the last model release and it says that December is shaping up to be a month of jeweling announcements from open Ai and Google and of course it talks about how anthropic wants to take over your computer and satcha nadella's good year but the main focus of this video is going to be Google because of course as you know this time last year they released Gemini 1. Pro which was outstanding it man to surpass GPT 40 there were just a bunch of things that were just absolutely incredible but now all eyes are on Google again to see what do they have for their next Frontier Model now there's a lot of speculation going on but I've been doing some research I'm digging I found a few things and I want to share them with you today so let's dive into exactly what could potentially be coming in December for Google and of course the next iteration of Frontier Model it says here that as my colleagues Kylie Robinson and Tom Warren reported open AI is eyeing a December debut for its next Flagship model and fittingly Google is aiming to release its next major Gemini 2. 0 model in the same month the source is familiar with the plan tell me of course this is going to be the contined trend as Google and openi continue to try and one up each other if you aren't familiar you have to remember that this time last year it was Google that managed to release Google Gemini Pro which actually surprisingly demolished opening eyes GPT 4 at the time across several different benchmarks this was as to many people's surprise as it demonstrated Advanced reasoning capabilities across coding across different contexts and across long context Solutions now with that being said it seems like not only is open aai and Google doing this but anthropic and x. and of course meta erasing to debut their next Frontier models now this is pretty interesting considering the fact that we've actually still been getting a frequent amount of updates with open aai and with Google with these Frontier companies we've recently had metas Lama 3. 2 open ai1 and even some secret Google models that seemed to slip under the Thea for most people so we'll be talking about that later on in the video but I want to show you guys a part of this article that might surprise you now do remember though that this is pure speculation meaning that these articles could have the information completely wrong that sometimes does happen and it's the only article that is reporting on this but they are saying unfortunately that the next iteration of models aren't showing the performance gains that these teams are hoping for you can see it says that apparently open AI CEO Sam mman is doing a phase roll out of these successes to GPT 4 starting first with their business partners but the sources apparently say that Google is planning to just simply widely release the next version of Gemini at the outset basically meaning that there's not going to be like any pre access any special accounts it's just going to be a blanket release meaning that everyone gets access to the new model at the same time now if these models aren't going to provide the reasoning capabilities that many have suspected it would Mark a very interesting turning point for the AI industry and I think this is going to be one of those ones that we're going to have to wait and see the reason I say this is because there's often so much speculation about models before they come through and often times sometimes they are accurate and sometimes they are completely off now I do think that if the model isn't as good as we think I think that of course once again they may iterate to the next Paradigm which is of course test time compute or inference time compute where they actually use the compute at inference time to get better reasoning abilities after all Google did publish a paper basically stating that this is the direction that they might be heading in now you can see that this article also States the fact that this chat p is the same that is happening across companies developing leading large models which means that once again it seems that if companies are starting to Pivot away from traditional llm models and move towards inference or test time compute then it does mean that open AI would be leading the charge once again with as to where the next iteration of models are going to go now there was a screenshot from a small event where I did see Logan killpatrick the league product for Google AI Studio who's working on Google G API and AGI ### [5:00](https://www.youtube.com/watch?v=cCJFq3Ul0yI&t=300s) Segment 2 (05:00 - 10:00) showing us exactly where we are headed you can see that he has on the presentation Google's Gemini 2. 0 saying that these next Generations of Frontier models are to be decided they've also got the ability to tune everything and of course Gemini 2. 0 vo image and 3 and various things like notebook LM so it was clear that of course the next iteration of models is going to be Gemini 2. 0 so at least we know the name of the model and what we can kind of expect now the thing is though even if these models aren't as crazy in terms of their reasoning ability I think Google is remarkably underrated in their ability to provide you with a variety of different things for example just because they might not be as useful I just think Google has a UI problem because if you actually use the Google AI Studio you can input a variety of different things audio video image and you can put things like videos that are an hour long audios that are hours long this is something that we simply just don't have from any major other platform so I completely wouldn't bet against Google in that regard as they still do have some serious Innovations under their belt and there was also a very cryptic tweet from Jimmy apples which was rather fascinating that actually does talk about Google's model and if you aren't familiar with why I'm referencing Jimmy apples it's because this is someone who last week knew that anthropic were're going to release something API related and agentic he says that someone may have got too drunk and said Gemini 2. 0 has already been deployed to select B2B customers but this might have been the ramblings of a drunk honestly I find it sometimes hard to decrypt what he's saying here but maybe Gemini 2. 0 has already been rolled out to select B2B customers for initial testing I mean this wouldn't be the first time a major company has done such a thing and of course Google might be trying to avoid any kind of backlash they do get as Google has faced numerous levels of scrunity for having models that have been deemed a little bit too woke in certain scenarios now that's not my biggest concern is whether or not these models are going to live up to the height and actually become a part of people's everyday use in the AI ecosystem now like I said before even if these models do not live up to the quote unquote hype I still think that Google are going to win this AI race in many different areas for example I'm also hearing that Noom the famed AI researcher that Google paid an ungodly amount of money to abandon character. a is working on a separate reasoning model to advance open ai1 however Google declined to comment on this story now I'm not sure if Noom is going to be working on this or not but I would say that it is most likely that Google is going to be working on an advanced AI model because they aren't just going to be doing one aspect of AI at all Google have brought us the innovations that have been long context and they are continuing to bring us a number of other breakthroughs that are going to make these AI systems more useful the only thing is that when we're going to get these in products I don't know there are a few things that Google announced at recent events that some people may have forgotten and I'm going to use this time to jog your memory on a few things that Google have in the pipeline because everyone seems to forget Google are a terrific company that seriously has a lot that we really shouldn't bet against okay let's do some tests tell me when you see something that makes sound I see a speaker which makes sound what is that part of the speaker called that is the Tweeter it produces high frequency sounds give me a creative alliteration about these creative crayons color cheerfully they certainly craft colorful creations what does that part of the code do this code defines encryption and decryption functions it seems to use aebc encryption to encode and decode data based on a key and an initialization Vector IV that's right what neighborhood do you think I'm in this appears to be the king's cross area of London it is known for its railway station and transportation connections do you remember where you saw my glasses yes I do your glasses were on the desk near a red apple what can I add here to make this system faster ### [10:00](https://www.youtube.com/watch?v=cCJFq3Ul0yI&t=600s) Segment 3 (10:00 - 11:00) adding a cache between the server and database could improve speed what does this remind you of shringer cat all right uh give me a band name for this Duo golden Stripes nice thanks jini and one of the things I do want to leave you guys with is the fact that if you haven't been using Google Gemini maybe you might want to reconsider that one thing that did slip under the radar and at the time I didn't even make a video on this because I didn't even realize it happened until a little bit later and it was something that just missed the mark for me in terms of news that I was aware of so Google did release an updated version of Google Gemini Pro that did actually perform significantly better on a range of different benchmarks including math including natural code mode you can see that the mmu the math Vista even on benchmarks that other models can't even compete in the models do perform exceptionally well just to be clear this is Google's Gemini 1. 5 pro2 so if you go into the Google AI studio and you want to take a look at Google's more advanced model the one that is actually quite useful I'd say you go ahead and go to the AI Studio because it's a place where you definitely want to be to get the most out of Google's products and services so let me know what you think about this stuff do you think that Google still has in the bag I still am not writing them off at all after all they've got many different products in the pipeline and the recent debut of notebook LM has left many people surprised and of course satisfied so that being said let me know your thoughts comment down below I'll see you in the next one --- *Источник: https://ekstraktznaniy.ru/video/13912*