So <a href='https://ainewsera.com/google-offers-free-ai-cyber-tools/make-money-ai/' title='Google offers free AI cyber tools to enhance online security' >Google</a> Deep Mind drops Gemini and the world goes wild

So Google Deep Mind drops Gemini and the world goes wild

I can honestly say I don’t think anyone saw this coming when Sam Alman got fired from openi for 3 days. I think we all felt that we had a front row seat to all those events because of the massive amounts of leaks that were happening. I feel like I was there while it was happening, we knew everything about everybody, but this massive news out of Google, not a peep!

The Gemini Era

The headline is “Welcome to the Gemini era and we beat GPT 4”. Here’s GPT 4, here’s Gemini, this line is the humans, and this is where the humans end up. Gemini is shown to be able to solve various visual puzzles and even guess where the ball was in a cup. Everybody’s amazed!

Critiques Against Gemini

After the initial hype, critiques started pouring in about how the numbers were attained, how the cup shuffling game was played, and even questions about Alpha code 2. Are the claims about Gemini as impressive as Google made it seem to be?

Google’s Advantages

Google may have an unsalable advantage with their state-of-the-art AI model, microchips, penetration in the phone market, massive databases, and the vast amount of data they’ve gathered from the internet. It seems like Google has caught up and surpassed others in the field of AI.

Gemini vs. GPT 4

After analyzing the data, it’s clear that Gemini and GPT 4 are neck-to-neck. While there may have been some polishing of results, the gap between the two models has been significantly narrowed.

Future Developments

There are rumors of open AI coming up with new models to compete with Gemini. The competition in the AI field is heating up with more players entering the game.

Final Thoughts

So, what do you think about Gemini? Is it as good as people think, or is it just a bunch of smoke and mirrors? Let me know your thoughts in the comments!

If you made it this far, thank you for reading my article. Stay tuned for more updates in the world of AI!

47 COMMENTS

  2. I think Google is tricking you all. It's a massive coup in their leading you all to believe that Gemini isn't capable and they're going to lead you into a trap especially open ai and these other companies that are going to get stung and shut down as soon as they bust out the real Gemini. I know all of this for a fact I better put out my expose tonight while the getting is good huh..

  4. I would be a little nervous if I was nvidia cause their crazy valuation is based on them supplying the shovels, if there is a lot of other shovel makers who are dramatically cheaper than nvidia will have to cut prices to compete and that means less profit and less sales.

  6. I'd be interested in a view on Opensource development given how much techical improvements are going to reduce the cost of training a lot in the next year or so. Will a trainable LLM come with a computer in the not too distant future?

  7. Defending false marketing, it's simple, if it's not to the customers benefit and is intentionally missleading, we need to criticize it. Don't defend the mega corporations norms that are only to their own benefits.

  12. Completely disagree it's closer to GPT4. I was fighting Bard on its stupidity just like I had to with GPT3.5. It has a smaller context window that causes a lot of problems for my conversations when working on code (It would say, provide the code, and I'd have to remind it that it just wrote the code I'm asking it to revise), and Bard failed badly to answer a mathematical word problem that GPT4 handles perfectly. Bard is good but distinctly feels on the level of GPT3.5 to me.

  14. Great analysis of where we are today. The multi-modal approach that Google took is interesting and I am glad Wes covered it and the TPUs. There is nothing like competition between these big companies to lower our costs for doing AI.

  16. What I want to happen is for one of these companies to give free limited access to the API of their models so that I can use it in a game without worrying about having to charge customers subscriptions because that business model isn't going to work for gamers.

  20. Gemini Pro is a little less better than GPT-3.5 from what I've seen and Gemini Ultra appears to be about level to GPT-4 in text. What Gemini Ultra should better GPT-4 at is in video and possibly audio and image.

    The issue is Gemini Pro isn't out in a lot of places, the image features are not available where Pro is out, Ultra has no release date and there is talk it will never be released to the public.

    Overall, along with the terrible marketing, I'm very disappointed…again. I really want Google to do well here.

    It's looking more and more likely that the Transformer has peaked with GPT-4 and the AI labs now need a whole new AI model to get to AGI.

  23. Tbh i think at this point is who going to do the photonic chip first, i was seeing about photonic graphic card sometime ago, but this will be what rlly decide who going to win the "AGI" run.

  24. I find it interesting that GPT-4 and Gemini, are so neck to neck comparatively, this is exactly what I would expect as the limitations of these models are reached. The growth curve for AI is a logistic curve, not exponential. I'd expect this trend to follow even when the models can self-improve, but I think thats where most people would disagree with me.

  27. 3:35 After it updated, Bard Gemeni pro couldn't maintain a single programming language as the subject of the discussion. And not over multiple messages, just one of its responses.

  29. So you took Google at face value on your last video and now… you are still doing it. Even when they are shown to have been focused on Marketing as much as science. When has tech marketing ever held up to hands-on? Never. That is why they made this announcement BEFORE people could use it.

  31. Btw Bard does not have vision yet, it just uses Google Lens to describe the images to Gemini Pro, that's how it's been the whole time. Google just made it really confusing. That's why Bard's vision has been bad the whole time it's been available.

  32. Google was a cool startup promoting "don't be evil", then became evil, and finally ended up a slothy, obese corporate dinosaur who cant get out of its own way.

  33. Gemini Pro couldn’t do the correct formula calculations on US federal taxes based on a yearly salary I gave it. Very disappointing. Claude did it perfectly. I don’t care how book smart these AI models are. They need to be practical for daily use to uplift humanity.

  35. We need better developer access to Gemini/Bard using APIs – it is currently easier to work with OpenAI as an application developer. Interacting with Google Gemini/Bard is through their apps directly. If I missed something please point me/us to the API layer access documentation. Google seems to want to control directly how this tech is being used and not focused on enabling other developers without a direct partnership (for big players) with Google.

  37. I didn’t notice much difference between previous bard and current bard. Gemini pro is still trash on some instances it even felt like bard has become dumb. Waiting for Google to optimize the model.

  40. I'd say Anthropic is also up there. Claude had no multi-modality, but feels more stable, logical and… sane. If they are working on something, and releasing it soon, it might be better than the rest.

  41. Yeah this is interesting, I wonder if they will keep going all into TPUs?

    It seems like those hardware innovations will provide the leverages required for being the ‘best’ in the short term. Then software (algorithmic) in the long term.

  44. I think OpenAI are going to be in trouble. Google has too many advantages over everyone else. The sheer amount data they have access to. Youtbe alone would be put them way infront everyone else on Vidoe modality. They have their own hardware, everyone else is paying 1000% markup on NVIDIA hardware. They can design their next gen processors to perfectly be in tune with their models, making both training and inference cost cheaper to run. And let's not forget about the 120B cash they have on hand. Sorry OpenAI, imo in 5 years time you'll be the next Yahoo.

  45. Until the whole world or at least most of the world get access to Bard Gemini then it doesn’t exist as far as I’m concerned. And all there claims are BS until I can see for myself. The world isn’t just the USA. But I suppose it probably costs a lot to run those AIs so maybe they can’t afford to allow other countries to use it yet. But from the perspective of someone living outside the US I think it’s a total joke.

  47. I don’t think Google ever intended to have the public have access to advanced AI and eventually AGI. I think Google wanted to develop AGI and keep it internally in order to help them dominate the world. OpenAI kind of forced their hand by letting everyday people use it.

LEAVE A REPLY

Please enter your comment!
Please enter your name here