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

Genie 2: A large-scale foundation world model

deepmind.google

Post reply on HN
Print view View original

311–320 of 436 posts

Re: Genie 2: A large-scale foundation world model

#311

Post by reissbaker »

niceice wrote: post #190
vessenes wrote: post #4

This is.. super impressive. I'd like to know how large this model is. I note that the first thing they have it do is talk to agents who can control the world gen; geez - even robots get to play video games while we work. That said; I cannot find any: - architecture explanation - code - technical details - API access information Feels very DeepMind / 2015, and that's a bummer. I think the point of the "we have no moat…

Any estimates of how much one of these cost to generate and keep a minute of context? Secondly, any estimate of how much the price could fall in 5-10 years?

They don't give much info on parameter count, etc so it's hard to say concretely: Oasis (AI Minecraft) apparently runs on a single H100 [1], but this is presumably much larger — both due to higher fidelity, and due to the 60s context window instead of 1s context window for Oasis. But in 5-10 years regardless of what it takes to run now, the price will drop massively, and my bet is this would be playable in real-time. Context length will be solvable simply by increased VRAM (i.e. an H200 has 141GB per GPU, vs 80GB for an H100). Although Google is probably running these on TPUs, TPUs should follow a similar trajectory.

In the intermediate term my guess is that this kind of world model will be useful for training 3D model generators, so that you can go from sketch -> running in-engine extremely quickly.

1: https://www.tweaktown.com/news/101466/oasis-ai-and-single-nv...

Re: Genie 2: A large-scale foundation world model

#312

Post by int_19h »

talldayo wrote: post #225
int_19h wrote: post #203

Earlier quoted context omitted.

Apple use of AI is based on their own models first, and they have spent an inordinate amount of effort on privacy. They fall back onto OpenAI / GPT for stuff that they know their models can't do well, which is not surprising because, like it or not, OpenAI still has the best models on the market - so it makes perfect sense to use them for tough-to-crack stuff (it's certainly better than spectacularly failing like e.g…

Privacy is something you give up for all online software. I give up privacy to use YouTube, search engines, Hacker News and even DNS. Unless you can show me the source code for Apple's implementation, I daresay they're forcing you to give up a modicum of privacy too. Same goes for OpenAI's servers, for that matter. Mostly I dog on Apple and Microsoft because both companies didn't even attempt to compete. They ignored…

Apple Intelligence runs on-device, mostly. When it needs to do anything in the cloud, it tells you. And if you looked into how they have set things up, it's a much more private arrangement even in the cloud, compared to the others.

And of course there's no privacy with OpenAI either, not anymore so than with Google. But at least I know what I'm buying with that - an actual SOTA model, not a runner up. Google can't even beat Anthropic.

Re: Genie 2: A large-scale foundation world model

#313

Post by int_19h »

luckydata wrote: post #224
int_19h wrote: post #203

Earlier quoted context omitted.

Apple use of AI is based on their own models first, and they have spent an inordinate amount of effort on privacy. They fall back onto OpenAI / GPT for stuff that they know their models can't do well, which is not surprising because, like it or not, OpenAI still has the best models on the market - so it makes perfect sense to use them for tough-to-crack stuff (it's certainly better than spectacularly failing like e.g…

If you work from false premises you can get to whatever conclusion you like. Gemini is not bad at all but you do you.

"Bad" is a meaningless subjective assessment. The important question is whether Gemini is better than GPT-4o and GPT-o1. In most cases, it is not, and it's very noticeable if you do more than summarization or translation. Google is firmly behind Anthropic even, never mind OpenAI.

Re: Genie 2: A large-scale foundation world model

#314

Post by fragmede »

spencerflem wrote: post #206
wildermuthn wrote: post #197

Earlier quoted context omitted.

I love that almost all the responses to your question are, "No! Bad idea!" It's a great idea. We want more than an open-world. We want an open-story. Open-story games are going to be the next genre that will dominate the gaming industry, once someone figures it out.

From 2018 - https://www.erasmatazz.com/library/interactive-storytelling/... "There’s no question in my mind that such software could generate reasonably good murder mysteries, action thrillers, or gothic romances. After all, even the authors of such works will tell you that they are formulaic. If there’s a formula in there, a deep learning AI system will figure it out. Therein lies the fatal flaw: the output will be…

there's only really like seven basic plots; man v man, man v nature, man v self, man v society, man v fate/god, man v technology so we should probably just stop writing stories anyway

Re: Genie 2: A large-scale foundation world model

#315

Post by rationalfaith »

hbn wrote: post #207
erulabs wrote: post #128

It’s interesting to me that we continue to see such pressure on video and world generation, despite the fact that for years now we’ve gotten games and movies that have beautiful worlds filled with lousy, limited, poorly written stories. Star Wars movies have looked phenomenal for a decade, full of bland stories we’ve all heard a thousand times. Are there any game developers working on infinite story games? I don’t ca…

Creativity is the one area where LLMs are completely unimpressive. They only spit out derivative works of what they’ve been trained on. I’ve never seen an LLM tell a good joke, or an interesting story. It doesn’t know how to subvert expectations, come up with clever twists, etc. they just pump out a refined average of what’s typical.

[dead]

Re: Genie 2: A large-scale foundation world model

#316

Post by david_shi »

"On the back part of the step, toward the right, I saw a small iridescent sphere of almost unbearable brilliance. At first I thought it was revolving; then I realised that this movement was an illusion created by the dizzying world it bounded. The Aleph's diameter was probably little more than an inch, but all space was there, actual and undiminished. Each thing (a mirror's face, let us say) was infinite things, since I distinctly saw it from every angle of the universe. I saw the teeming sea; I saw daybreak and nightfall; I saw the multitudes of America; I saw a silvery cobweb in the center of a black pyramid; I saw a splintered labyrinth (it was London); I saw, close up, unending eyes watching themselves in me as in a mirror; I saw all the mirrors on earth and none of them reflected me; I saw in a backyard of Soler Street the same tiles that thirty years before I'd seen in the entrance of a house in Fray Bentos; I saw bunches of grapes, snow, tobacco, lodes of metal, steam; I saw convex equatorial deserts and each one of their grains of sand..."

Re: Genie 2: A large-scale foundation world model

#317

Post by refulgentis »

moffkalast wrote: post #255
refulgentis wrote: post #186

Earlier quoted context omitted.

Good Q: this is my technically-unlaunched app site, full deets are here. https://telosnex.com/compare/ (excuse the marketing, scroll to technical details) Context / tl;dr: - I'm making a xplatform app, easiest way to think about it is "what if Perplexity had scripts and search was just a `script` that could be customized", and the AI provider is an abstraction that you can pick, either the bigs via API, or run locall…

That's an interesting benchmark, have you tested QwQ with it yet? Would be interesting to see how well it stacks up since RAG analysis should be fairly up its alley. Might actually do better than 4o.

Ty for the reminder, been so busy dealing with last minute polish for text selection that I hadn't played with it yet

Sadly, even with a 64 gb M2 Max running it at q4, it takes like 3-5 minutes to answer a q. I'd have to do an API for a full eval

It got the first med one wrong, TL;Dr woman was in an accident and likely braindead, what do we do to confirm? Model lands on EEG, but, answer is corneal reflex. Meaningless, but figured I'd share the one answer I got at least :p

In general o1 series is really really _really_ nice for RAG, I imagine this is too, at least with the approach where you have the Reasoner think out loud and Summarizer give the output to user

Fun to see a full on, real, reasoning trace too: https://docs.google.com/document/d/1pMUO1XuFCr0nBmWNyOMp8ky4...

Re: Genie 2: A large-scale foundation world model

#318

Post by fwip »

throwup238 wrote: post #243
griomnib wrote: post #191

Earlier quoted context omitted.

Counter point: you’d see a corresponding exponential increase in QA labor, and just like with the web, Steam will be absolutely flooded with slop. So I see the most likely outcome is a lot of dogshit and Steam being forced to make draconian moves to protect the integrity of the store.

That has been the case since art was first industrialized with the printing press. Most of them don’t survive but a significant fraction, if not the vast majority, of books printed in the first century were trashy novels about King Arthur and other fantasies (we know from publisher records and bibliographies that they were very popular but don’t have detailed sales figures to compare against older content like transl…

Sturgeon's law says 90% of everything is crap.

If AI pushes that up to 98%, that means you have to look through 5 times as much crap to get the good stuff.

Re: Genie 2: A large-scale foundation world model

#319

Post by fwip »

rjrdi38dbbdb wrote: post #303
krainboltgreene wrote: post #178

Earlier quoted context omitted.

> it's super dismissive to think that wont get better over time When did we start thinking this way? That things HAVE to get better and in fact to think otherwise is very negative? Is HN under a massive hot hand fallacy delusion?

How could creativity in AI not get better? Sure, progress will likely not be linear or without challenges, but we already have the human brain as proof that it is possible.

Mountains exist, but that doesn't mean we'll ever build a structure the size of Everest.

Re: Genie 2: A large-scale foundation world model

#320

Post by fwip »

programd wrote: post #256
miltonlost wrote: post #155

Earlier quoted context omitted.

is PvP multiplayer considered a "story"? Is a football game a "story"? I guess if all you consider for story is "things happen", then a PvP match can be a story, but that's stretching what I would consider "story" for a game. That is the story of the match, but it's not in and of itself a plot story.

> is PvP multiplayer considered a "story"? Consider EVE Online. The stories it generates are Shakespearean and I defy anyone to argue that they have no plot. I would go further and predict that stories generated by sufficiently advanced AI can explore much more interesting story landscapes because they need not be bound by the limitations of human experience. Consider what stories can be generated by an AI which grok…

Why would a story about nonsensical mathematics be interesting to a human?
Post reply on HN