Hacker News Reader

Comment by taurath | Hacker News Reader

> 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

View on HN

Probably not going to happen, but I'd love to see Isomorphic Labs separate from Alphabet with Hassabis still as CEO. He's too pure minded to be at company like Google.

He won a nobel prize for his work on protein folding while at Google...it seems like they give him a lot of room to explore problems that aren't directly related to serving ads

But it was Google fault that they really cannot productize all that innovation that happened DeepMind. Google probably now fumbled with world models and this exodus will create next giant in that space.

> Google probably now fumbled with world models [...].

Can you explain this more precisely?

I have a feeling this might negatively impact Isomorphic as well since they also fall under Alphabet

Hopefully not - Isomorphic also have external investors (maybe 10-20% ownership?), who may be willing to buy it outright from Google if they lose interest, and Hassabis is now said to be "leaning into" the role of Isomorphic CEO, which is at least pursuing his goal for scientific advance via AI, if only in one specific (important) direction.

https://endpoints.news/demis-hassabis-leaning-into-isomorphi...

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.

> The amount they are spending on Gemini is probably shockingly low

IDK, haven't Google been putting Gemini front and centre in pretty much all of their products?

I'm seeing Gemini on my slide decks, Gemini on my e-mails, Gemini on my searches, Gemini on my videoconferences, Gemini on my database query console. My impression was they were doing a Google Plus style attempt to marshal all the company's efforts behind one product.

Their capex is primarily in building DCs (and what goes into them), which they fully intend to monetize in order to capture that $500B backlog. In that context, the amount of new capacity that will be given to Deepmind could be surprisingly small, since the ROI of selling pickaxes rather than mining gold is so much better right now.

I’ve heard the pay at deep mind is low and people are complaining about it

How low are we talking?

Nowhere near what other ai labs are paying. Most deep mind employees make what typical Google employees make

Not true, they get paid at least a whole level (or two) above what we make. When I was discussing offers a few years ago, the GDM candidates regularly had six figure differentials even though they were the same level.

Maybe this isn't the same as the eight figure comp they'd get at Meta when they did their hiring spree, but no one thinks that's sustainable.

its still not enough, and the up leveling is a problem because the people getting hired arent qualified to be upleveled

So, how low?

they are paying them like rank and file google employees, aside from a few star researchers

Of course it does matter what other companies are paying, but I wonder how many DeepMind employees actually deserve to be paid more than any other "rank and file" Google employee? What unique skill do they have (esp. relative to what Google are now doing), and what value are they creating?

I don't get the impression that the difference between one company succeeding to build SOTA LLMs, and another struggling, comes down to individual employees - it seems to be more about the organization itself and their ability to manage teams and projects of this type. No doubt there are a few rockstars generating huge value, such as Noam Shazeer had been, but they are the exceptions.

When DeepMind was first created, before Google acquired it, they were famous for the high salaries, especially for the UK, but this was an assemblage of the brightest and best PhDs, expected to be solving challenging research problems along the unknown path to AGI. Many of these original employees may still be there, but it seems their job and value proposition has changed - are they any more capable, or key to, helping Gemini catch up with the competition than some "rank and file" employee familiar with LLMs? And if so, why haven't they done it?

the fate of the company partially rests on whether deep mind does well, you sound like an MBA

>the fate of the company partially rests on whether deep mind does well

I'm honestly not sure about that. They need a decent LLM to fight off the threat of LLMs replacing search, but I'd say that Gemini 3.6 Flash is more than good enough for that, with a smaller/cheaper model being preferred to a larger one. If you are "searching" for a proof to the Jacobian conjecture, then try Fable, and I doubt Google will miss the advertising revenue if Anthropic manage to sell Terrance Tao a pair of socks.

More to the point, DeepMind seems to have become a product division charged with building LLMs, not a blue sky research institute chasing AGI. To the extent that continued LLM improvement is important to Google, the relevant question is how much do you need to pay for a competent ML/LLM developer?

>you sound like an MBA

Well, no - techie here.

I wasn't sure if you were suggesting that DeepMind should pay more just because people developing LLMs at other companies are paid more, or because these are elite DeepMind researchers and are objectively worth more than other Google developers. My point being that it seems they are no longer being used as elite researchers - they are LLM developers. Does Meta need to pay FAANG salaries to employees that have been repurposed as data labellers?

what are you even trying to argue? that the most critical part of the biggest companies on the planet dont need to pay above market?

i know people at deep mind, its impacting their ability to deliver good products

Right, but deepmind won't do well if they can't get TPUs, if the TPU network doesn't work, if their borg jobs aren't coming up fast enough, etc.

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.

> 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.

If the goal is to maximize share price (which it is) this is probably the safest approach. Why do they have to keep chasing developing the best model when they can

a) charge everyone for the cloud infra

b) have a "good enough" experience for normal consumers

Their current setup will be worth trillions. Already is. Let anthropic get paid for the most expensive queries while they get a bill from Google for their cloud/TPU use. GCP grew 82% YOY with 24B revenue and improving margins. So GCP became a 100B business. Anyone doubts it's gonna double in less than 5 years? Gemini just needs to be good enough for normal folks who want personal agents for everyday use. I don't think Google has to compete with Anthropic on making the best agentic programming model.

Same thing happened to Yan LeCun.

Real scientists are skeptical. Wall St and the people who serve it don’t like that.

> Whoever ends up in charge needs to do a lot of frothing

I don't get the impression from interviews that Kavukcuoglu is that guy - he seems like a safe pair of hands, but not someone that is on a mission.

OTOH I don't even think this is the right race to be in.

Reply on news.ycombinator.com