Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

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

blog.google

Post reply on HN
Print view View original

341–350 of 988 posts

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

#341

Post by marking-time »

jgalt212 wrote: post #224
MikhailTal wrote: post #212

> Google DeepMind: We are building strong momentum: Flash is in high demand, our Cyber model is live, and Gemma models have surpassed 900M+ downloads Considering these are the best stats they could find, gemini usage+general situation must be really, really bleak. High demand means nothing. A model being live is nothing to brag about. And gemma downloads also can be from auto CI pipelines etc. Nothing concrete

> And gemma downloads also can be from auto CI pipelines etc. Nothing concrete I have always found NPM download numbers truly suspect. Is no one caching? Are they estimating true number of downloads base on some estimate of cache hits?

> And gemma downloads also can be from auto CI pipelines etc. Nothing concrete

Thank you for adding some clarity to this. When I calculated 900 million downloads divided by 8.3 billion people in the world, I came with a number that made it look like about one person in 10 were downloading this model.

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

#342

Post by yumraj »

Oras wrote: post #285
jcfrei wrote: post #154

Earlier quoted context omitted.

I think their move makes a lot of sense. Frontier models can probably not advance much further with the datasets we have currently available. Most of the text that goes in is online chatter, images and some scientific texts. That's great for chatbots, knowledge retrieval and programming. But with that database genuine discovery is hard to do. I think for the next step in intelligence the models need to have access to…

> Frontier models can probably not advance much further with the datasets we have currently available I’ve been reading this sentiment on HN since GPT4o, yet models got better and better

There’s a difference between things getting incrementally better and a step function.

LLMs will obviously keep getting incrementally better, but in order to get the kind on jump that LLMs themselves were, the sentiment is that we need something more.

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

#343

Post by GodelNumbering »

squidbeak wrote: post #284
GodelNumbering wrote: post #199

So, in last several months, all the prominent names Google lost: Demis Hassabis (technically still with google but these things are usually presented with a spin), Jeff Dean, Sanjay Ghemawat, Oriol Vinyals, Quoc Le, Noam Shazeer, John Jumper, Jonas Adler, Alexander Pritzel, David Silver, Denny Zhou, Fernando Pereira, Alex Turner And all the prominent names Google gained: NULL Combined with no gemini frontier GA relea…

Gemini 3 Pro was legitimately frontier and released 8 1/2 months ago, not 14.

Correct, but it was preview release. I was referring to GA in my comment.

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

#344

Post by ilaksh »

vineyardmike wrote: post #122
atleastoptimal wrote: post #30

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…

The answer is far more benign. They getting a higher ROI renting their TPUs to Anthropic et al instead of performing training and serving their own models. Google cloud has insane backlog, and has rapidly expanded to satisfy it. While those DCs get built, they’re cannibalizing their own products for it. This makes sense because (1) they are investors in Anthropic, so they still win and (2) they can always catch up on…

I think the reason Google hasn't prioritized larger models that are more intelligent than everyone else's, even though they probably could, is because they have 4 billion active users already. For example, they send AI Overviews for a large portion of Google searches now.

So I think they have to prioritize scaling for their models to a higher degree than other groups. Being within say 5% or so in most cases is probably adequate and matters more overall for their user base than being the absolute best coder. So they may be setting compute constraints for training or inference that are firmer than other teams.

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

#345

Post by tomnicholas1 »

jcfrei wrote: post #154
xtreak29 wrote: post #5

https://xcancel.com/JeffDean/status/2085034604172603724 > Announcing Discovery Loop! > I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat, @OriolVinyalsML and @quocleix, we are founding Discovery Loop (@DiscoLoopAI), a Public Benefit Corporation whose mission is to automate machine learning, science, and engineering to accelerate discoveries and progress. The four of…

I think their move makes a lot of sense. Frontier models can probably not advance much further with the datasets we have currently available. Most of the text that goes in is online chatter, images and some scientific texts. That's great for chatbots, knowledge retrieval and programming. But with that database genuine discovery is hard to do. I think for the next step in intelligence the models need to have access to…

> Data from physics, chemistry, biology experiments

There are PetaBytes of important scientific data locked in archival file formats. The first step is to make this efficiently readable.

https://www.earthmover.io/blog/virtual-zarr

https://news.ycombinator.com/item?id=46659254

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

#346

Post by ben_w »

closetheloopdev wrote: post #318
possibilistic wrote: post #293

Earlier quoted context omitted.

2023: Google is doomed. They're not building AI. No wonder the "Attention" authors left. Google just sat on this. Their PMs are steering them into oblivion. OpenAI is winning while Google takes no risks. Innovator's dilemma. Google Search is doomed. 2024: Google is on fire. Sergey coming back helped. Gemini, Veo. They've caught up. OpenAI is doomed. 2025: Google is seriously winning now. Nano Banana!! Google was dest…

It seems Google has to play both offense and defense: competing against frontier labs' models while protecting search and ads. They also have to be mindful of not doing anything that could hurt their own search or ads. That seems harder than a frontier lab just doing offense on both models and search/ads. Having said that, I think Google's moat is still strong with Cloud, Gmail, YouTube, Android, Chrome, etc.

> Cloud, Gmail, YouTube, Android, Chrome

Cloud and gmail, I agree.

I don't think YouTube can survive if GenAI keeps going like this. Android, same, but on a different timescale and for different reasons. The Play Store (and all other app stores including Apple's) will also face problems from GenAI making apps (it already replaces my need to buy, but I'm weird and a software dev ("but I repeat myself")).

Not sure how big a moat Chrome really is? It's more like a sales funnel than a product itself, I think?

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

#348

Post by rowyourboat »

krapht wrote: post #76
jerlam wrote: post #42

Earlier quoted context omitted.

Google specifically has a track record. You could have made the same argument about social media (Google+) a decade ago.

social media is a completely different market though - since there are massive returns to scale, it's incredibly hard for a new entrant to break in. model training and inference is the opposite. people switch LLMs like people change clothes in the morning. there are popular services (openrouter) that make moving as easy as changing a model string.

Which poses the question: Why does Google even need to catch up? At least currently, the name of the game is integration. The actual model is a commodity.

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

#349

Post by tonfa »

VirusNewbie wrote: post #179
tonfa wrote: post #162

Earlier quoted context omitted.

It usually isn't, Chairman doesn't involve day to day supervision and decision making. When Eric Schmidt became chairman or Ruth Porat became president, this was more a transition towards less involvement than an increase of impact.

He means the chief scientist part. Before Demis had no influence on cloud, tpu development, day to day dev tools, etc.

Demis took Jeff former title. (Jeff became Chief Scientist when Brain was dissolved into deepmind)

I suspect the SVP/CEO leading Deepmind has a lot more sway over the rest of the company than someone with a title and no reports. :)

(Also the 20y+ of contributions and having built much of the base Google infra probably gave more influence to Jeff than the remaining title)

Post reply on HN