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

> What beat them down were the gauntlet of reviews, the frequent re-orgs, the institutional scar tissue from past failures, and the complexity of doing even simple things on the world stage. Startups can afford to ignore many concerns, Googlers rarely can.

I wonder if this helps explain why Google is getting smoked in the LLM space right now.



Eh, as outsiders we're all quick to judge.

OpenAI and friends are able to move quickly, but (so far) they're not able to translate their LLM innovations into high-margin revenue with any significant moat.

Give it a couple years to see where all the cards settle and who's actually making money "with" LLMs.


Google hasn’t innovated on anything in over a decade. Just continuing to ride that search monopoly. Entire company of rest and vesters.


Except, as pointed out, a lot of the tech was literally developed at Google. It’s like Xerox Labs all over again.


They literally invented transformers, one of the key innovations that enabled this LLM boom.


Google didn't invent shit, the following people did. None are still with Google.

* Ashish Vaswani - Founder, Stealth Startup

* Noam Shazeer - Founder, Character.AI

* Niki Parmar - Founder, Stealth Startup

* Jakob Uszkoreit - Founder, Inceptive

* Llion Jones - Founder, Sakana AI

* Aidan Gomez - Founder, cohere

* Lukasz Kaiser - OpenAI

* Illia Polosukhin - Founder, NEAR


Then by your logic no company ever invents anything.


You’re inferring things I didn’t say. Google’s inaction on the invention and letting that entire team leave the company is proof they’re inept.


if the same invention happened at a university, no one would say MIT/Stanford invented transformer. We credit the people involved. Somehow, if it happens at a company, company gets most of the credit. Even when the papers and authors are publicly available. this is different from say iPhone which probably would not be developed at a university.


"researched" it, yes. But they are completely unable to operationalize it.


I used Google Bard today.

Is that somehow not operationalized?


Ahh so you were our user today. How was the experience?


Bard is terrible at coding. It makes a lot of mistakes but is too insistent it got it right. My feeling is Bard confabulates more often than the recent version of ChatGPT.

The good thing is one can verify the output of Bard easily with the Google button and when asking for links it will give you reference (real or not). Other plus is the ability to access internet, so you can give Bard directly your own document for processing.

ChatGPT also seems to understand the intention of the user better while the filtering (clam down) of bard is very strict.




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

Search: