Live data from Hacker News

Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs

blog.google

571–580 of 988 posts

Re: Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs

#571
post #412

Earlier quoted context omitted.

Their entire business is predicated on search dominance, which is now on much shakier footing. Agreed that those comments were ridiculous pre-ChatGPT, but now they have a genuine challenger.

This take only makes sense if you don’t know what people use Google search for. If you’re looking for your local Ford dealership, or to book a Carnival cruise, or need a disability lawyer, or to refinance your credit card debt, you’re searching for an ad. You’re not using Chat-GPT for that. You’re asking Chat-GPT about a research project. It’s stealing away all the hard to monetize searches and leaving the incredibly…

> You’re not using Chat-GPT for that

And that's where you're dead wrong. Absolutely wrong.

Re: Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs

#572

Earlier quoted context omitted.

People 100% are asking ChatGPT for all those things.

Someone deleted a reply to this comment “And where does ChatGPT get the data for those answers?” they wrote and I think the question is important: LLM chatbots can crawl the web just as well as Google.

The answer is Exa

Re: Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs

#573

I think there's one weirdly simple reason DeepMind isn't doing as well as OpenAI and Anthropic. I may be wrong on this. OpenAI and Anthropic went from tiny startups to huge companies. As a consequence the stock options/RSU's offered to the employees paid off a far higher percentage ROI than any stock options a DeepMind (and thus Google) employee would get (since Google is already huge). This disincentivizes people wh…

One theory I've been entertaining is that whenever GPT-3.5 came out a lot of people were talking about the "bitter lesson" and how scale was all we really needed to get to AGI. No need for any fancy tricks, just release a larger model trained on more data, by the time we released a hypothetical "GPT-5 sized" model we'd have AGI. Anyway, the actual theory is that Google and Meta have fallen behind because they've been…

Google has a ton of smart researchers trying all sorts of stuff. They didn’t really go all in on any one thing. They hedge their bets.

But they suck at harnesses, developer tooling etc. their internal environment is very complex and an intern on a MacBook with codex can probably move faster than a seasoned deepmind engineer

Re: Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs

#574

Earlier quoted context omitted.

Losing Ive was probably a good thing for Apple. I often feel that after a few years these big guys have done what they could do and somebody else can take it from there. So far even losing Jobs didn’t hurt Apple as it looks.

Jony Ive wanted to chop down all the trees at De Anza Community College for something like 11 million dollars for an Apple event. There's a lot of ways he was demanding in the wrong ways for Apple

Man, I'd never heard that story, so I verified it. Turns out it was about two dozen trees and 25 million dollars, just to put up a big tent for an Apple launch event. (Yes, the trees were removed.) What a tool, he just lost my respect.

Re: Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs

#575

Earlier quoted context omitted.

Someone deleted a reply to this comment “And where does ChatGPT get the data for those answers?” they wrote and I think the question is important: LLM chatbots can crawl the web just as well as Google.

The answer is Exa

Nah, just another startup that’ll get acquired or a similar fate. Building the capability is a superior option imho, if you’re at Anthropic or OpenAI scale. Cut out the middleman.

Re: Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs

#576

Earlier quoted context omitted.

Looked up LessWrong. Eh, philosophy people, also known for that "Roko's basilisk" meme. I'm gonna pass.

It's a rationalist community, IMHO their methodology of thought leads to much more well-reasoned takes on AI than on HN, where the discussion here is often very emotionally charged or led by wishful thinking. When it comes to AI, LessWrong has been discussing topics for years, that HN has just started to consider, so they are much further along in the philosophical "chain of thought" so to say. LW fully understood an…

Well at least HN hasn't (yet?) spawned a murderous death cult.

https://en.wikipedia.org/wiki/Zizians

Re: Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs

#577

Earlier quoted context omitted.

The answer is Exa

Nah, just another startup that’ll get acquired or a similar fate. Building the capability is a superior option imho, if you’re at Anthropic or OpenAI scale. Cut out the middleman.

They can't do it, which is why they acquire.

Re: Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs

#578

I think there's one weirdly simple reason DeepMind isn't doing as well as OpenAI and Anthropic. I may be wrong on this. OpenAI and Anthropic went from tiny startups to huge companies. As a consequence the stock options/RSU's offered to the employees paid off a far higher percentage ROI than any stock options a DeepMind (and thus Google) employee would get (since Google is already huge). This disincentivizes people wh…

It's not a ML talent problem. You don't need to be a genius deep learning researcher to think "Hey, maybe if we massively throttle and degrade the quality of our model while still charging the same price, that might drive people away" (as happened with Gemini 2.5 Pro, the one model where Google really was SOTA). Google's likely been providing insufficient training compute to DeepMind the same way they've been nickel-…

Is that what's happening? Search has sucked for a while, and I'd just assumed it was partly because they'd been pushing compute to AI-related activities.

Re: Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs

#579

Earlier quoted context omitted.

> Hassabis seems to have been pushed aside. Yeah he seemed too reasonable to me, relative to the fervor. Whoever ends up in charge needs to do a lot of frothing to catch up to the ferver that would justify their valuations and investments

Google's capex is to capture the compute purchases from the other labs. The amount they are spending on Gemini is probably shockingly low...and hence researchers leaving for better watered pastures.

this is roughly how i've been reading their approach, but i wonder when/if this changes. what actually makes them start caring about having the most capable model? maybe nothing. i suppose they could be happy in a world where they (a) are still the agents' preferred search index and (b) own or produce a good percentage of the hardware that anthropic and openai run on.
Post reply on HN