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

321–330 of 989 posts

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

#321

Post by raincole »

firasd wrote: post #312

Who knew the real revenue unlock wouldn’t be based on how much paranoid red-teaming the model underwent to resist users jailbreaking its ‘alignment’, and instead more on whether the model is post-trained to use ‘sed’ and ‘git’? Poor Gemini Scientist-heavy orgs that want to solve everything in token space may overtook tool use; meanwhile Anthropic has been super focused on MCP, Claude Code etc for over a year

Out of the three main US AI companies' models, Gemini is obviously the less aligned (read: censored). So I really don't know what you're talking about.

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

#322

Post by asdfman123 »

compiler-guy wrote: post #278
asdfman123 wrote: post #198

Earlier quoted context omitted.

A "Maria's not an asset to the abbey" situation for you Sound of Music enjoyers.

Except no one was singing "How do you solve a problem like Jeff and Sanjay?" prior to this.

You don't know that for sure.

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

#323

Post by gbalduzzi »

simianwords wrote: post #295
bigyabai wrote: post #269

Earlier quoted context omitted.

> Why wouldn't you? It reduces your overall cost! Because real Fable usage starts at $20/month, and has oppressive usage limits even at that (ridiculous) monthly price. Compared to my $3/month GLM-5.2 subscription, I have never felt like I was leaving capabilities on the table by refusing to cough up $20 for 15 minutes of Fable use per day.

This is the wrong way to look at it. If you have a complicated task , you can solve it for cheaper if you used Fable. It will use fewer turns to achieve the same result. You can solve it for cheaper if you use GLM but if you are involved in it more, but that defeats the purpose.

The point is that there aren't many complex tasks were fable delivers a significant value increase over cheaper models.

Single prompting a very complex tasks is rare even on frontier models, because it can be done successfully only for specific situations (e.g. you have a very strong verification step the model can iterate on).

Most of my everyday usage is for smaller takes, were you don't really get the benefit of the most expensive models, and my guess is that is the case for the most users

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

#324

Post by simianwords »

bigyabai wrote: post #306
simianwords wrote: post #295

Earlier quoted context omitted.

This is the wrong way to look at it. If you have a complicated task , you can solve it for cheaper if you used Fable. It will use fewer turns to achieve the same result. You can solve it for cheaper if you use GLM but if you are involved in it more, but that defeats the purpose.

It's not cheaper if the price of admission is $20 for the first taste. And it's definitely not cheaper to pay per-token versus using my GLM-5.2 quota.

Again this is a resolution problem. Your tasks are small enough that fit into a nice $3 quota. If you are an enterprise or a power user, the right-sizing argument doesn't work.

I'm talking about API prices - subscription is a different game.

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

#325

Post by xnx »

user43928 wrote: post #282
xnx wrote: post #95

Earlier quoted context omitted.

OpenAI and Anthropic have the freedom to do absolutely insane things like negligently hack other companies. It would be stock price suicide if anything even remotely happened with Google.

If Google put out a blog post about Gemini 4 having escaped its sandbox via 0-day exploit to then hack other companies, I unironically believe this would result in a boost to their stock price. They could really use some encouraging news about the competitiveness of their AI lab.

I wonder if we'll ever know the story of Gemini 3.5 Pro. Is it possible Google saw its potential for hacking, tried to nerf it, and ended up ruining the training run?

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

#326

Post by tuesdaynight »

Analemma_ wrote: post #225
xnx wrote: post #9

> Lastly, after an incredible 27-year run, Jeff Dean is at a moment where he wants to try something new, and we’re excited to support him in that. Jeff and Google Senior Fellow Sanjay Ghemawat are launching an independent public benefit corporation to accelerate discoveries in ML, science, and engineering. Oof. Good for Jeff and Sanjay (who just joined Twitter), bu that is a big loss for Google. Google stock is down…

For at least a year now the standard reflexive reply to “Google seems way behind OAI and Anthropic” has been “it’ll be ok, they’ve got Dean and Hassabis.” And now they don’t. What reason is there to be bullish about Google now?

It just depends on what are the most upvoted comments in HN. If they are bearish, you can be sure that you should be bullish in your investments. Meta was supposed to be broken as a business already, OpenAI and Anthropic would be failures as well.

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

#328

Post by simianwords »

gbalduzzi wrote: post #323
simianwords wrote: post #295

Earlier quoted context omitted.

This is the wrong way to look at it. If you have a complicated task , you can solve it for cheaper if you used Fable. It will use fewer turns to achieve the same result. You can solve it for cheaper if you use GLM but if you are involved in it more, but that defeats the purpose.

The point is that there aren't many complex tasks were fable delivers a significant value increase over cheaper models. Single prompting a very complex tasks is rare even on frontier models, because it can be done successfully only for specific situations (e.g. you have a very strong verification step the model can iterate on). Most of my everyday usage is for smaller takes, were you don't really get the benefit of t…

> The point is that there aren't many complex tasks were fable delivers a significant value increase over cheaper models.

Strong disagree on this. Any decently complicated task like a refactor is going to be more likely to be solved by Fable than by Gemma 3B or whatever.

I have personally tried to use Sonnet over Opus for tasks and Sonnet gets things right sometimes and at other times I wish I had just paid higher.

This is the standard pattern I keep seeing and I can have a bet with you that it would stay like this.

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

#329

Post by epiccoleman »

bigyabai wrote: post #269
simianwords wrote: post #266

Earlier quoted context omitted.

> I said for a few years to many a downvote on HN, everyone wants AI, nobody wants to pay the true costs, the AI race will turn into a "race to the bottom" that is, who can give you the most compute for the lowest cost, and still remain profitable? I keep seeing this but this line of thinking doesn't make any sense. What does it really mean? There are expensive models that increase the probability of you doing your t…

> Why wouldn't you? It reduces your overall cost! Because real Fable usage starts at $20/month, and has oppressive usage limits even at that (ridiculous) monthly price. Compared to my $3/month GLM-5.2 subscription, I have never felt like I was leaving capabilities on the table by refusing to cough up $20 for 15 minutes of Fable use per day.

where are you subbing to GLM-5.2? i've been meaning to try it out and for $3 it's a no-brainer to just load it up and give it a shot.

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

#330

Post by CobrastanJorji »

Mistletoe wrote: post #192

Lots of talk of us being at the cusp of AGI, but how will we even know when we get there? When AGI develops a religion for itself?

We'll know that we've reached AGI when a key conservative political belief in the US is that AIs are not people and do not deserve rights.

We'll know that we've reached the singularity when there's a mysterious and superintelligent AI entity with technology and motivations that we as humanity do not understand and have no control over. Not sure why we want that, exactly, but that'll be the big sign.

Post reply on HN