Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Catching up right now is not a matter of tech innovation, but raw energy and compute.

Of course, the next -revolution- in AI could very well come from Ilya. But why would he bestow that honor to anyone? He can self fund it if he wants. It's an R&D project, not a scaling problem.



> Catching up right now is not a matter of tech innovation, but raw energy and compute.

I don't really understand that. ChatGPT is not all that impressive as an actual tool for many things. It doesn't really seem to matter to me what energy and compute is going into it. Will it make it actually work?


If you consider the best frontier LLM in the world "not that impressive," I'm curious to know what you think is


I think the point is not "chatGPT isn't good", it's that no one is beating physics, you can rough out the coefficients on a napkin, and they're playing chicken with who wants to set more piles of money on fire.

It _is_ the best frontier LLM in the world, and virtually the entire global population of people who care about that are in this thread


This will change when the rumors turn out to be true and Apples Siri is powered by OpenAI and a billion people have a working conversational AI in their pocket.


I agree that this will happen (I said in another post), but I can't for the life of me attach it to one of the points I made


If that is the case, why hasn't Google caught up yet?


Is Google significantly behind? I would say from a technological perspective, they're very closely trailing or even surpassing OpenAI in many ways; they've built a formidable ChatGPT competitor in Gemini—not to mention they have a huge home court advantage with billions of people on Search, Chrome, Android, and G Suite. If you zoom out a bit, it's not really a fair fight between the two. More likely, Apple and Microsoft win against Google because they use/buy OpenAI.


Technical chops has never been Google's problem. Gemini for all the hype Google has thrown at it, has continued the recurring trend of G not knowing how to win with their product launches.


When you push either model to its limits, yes, Google is pretty far behind GPT-4 with Gemini.


Because Google is huge. Presumably the same reason why Microsoft finds it easier to shovel billions into OpenAI instead of doing it in-house.


Because their weakness is in focus and execution, not resources.

If they focused all of their energy on, simply, a frontier AI model, and not trying to shoehorn a half-complete model into all of their products, there is no doubt they would be ahead.

But this is the innovator's dilemma, and why it is that startups are the disruptors.

Big companies move slow and lack focus. Small companies move fast and can only focus on one thing.


and politics, although they seem to have woken up (no pun intended) and started de-politicizing.


Wrong leadership at every level: Sundar as CEO, Prabhakar as Search SVP, Sissie as Assistant/Gemini VP. Maybe they should hire Ilya instead?


probably because they have identified (correctly) that slowly integrating AI into their existing products that make them hundreds of billions is smarter than just burning that money to get upvoted on HN. Tall trees catch too much wind, ancient Chinese proverb. If you can afford to being second is usually less painful than being first.


There are clear advantages and disadvantages to that approach, but assuming this is the correct choice ( seems plausible, but I am not automatically convinced ), is that the actual reason for being as behind as they are?

I would argue that is not the case. I won't re-list some of the reasons other posters mentioned, which, based on past year, appear more likely ( decisions hamstrung by corporate committees, data governance bureaucracy, and last, but not least, ideology focus ). Leadership that is actually focused on 'delivering value to the shareholder' or not being worried about first mover advantage seems only a part of it.

edit: added first mover wording


my guess is that a significantly underrated reason is that Google holds itself to stricter data compliance standards than OAI


That is openly laughable. Incognito mode claims, just one of a long list of examples of this politicized monopoly engaging in criminal behavior.


I don’t think that’s a fair take nor does it really disprove mine.

The compliance regime at big corps is definitely more sophisticated than at a player like OAI


Cynicism is just giving yourself an excuse to be wrong because it sounds cooler.


Google is ideologically compromised which ties them into knots and creates endless internal struggles.

this led to the release of Gemini which was absurdly biased when viewed by the typical American.


I feel like the lowest hanging fruit right now lies with the UI and already established techniques for reducing latency and making the experience smoother.

Just executing the client right would give someone a competitive advantage right now...


I'd argue it's mainly a matter of people. Otherwise Meta, Amazon, Google, etc would've released their versions like yesterday.


I mean, that's probably 90% or 95% true, but the remaining 5-10% is almost certainly worth a $100M offer to Sutskever from Google or Meta (or possibly Amazon or Apple).


it’s a matter of data




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: