Researchers (Including Google) are Betting on Virtual 'World Models' for Better AI (msn.com) 12
"Today's AIs are book smart," reports the Wall Street Journal. "Everything they know they learned from available language, images and videos. To evolve further, they have to get street smart."
And that requires "world models," which are "gaining momentum in frontier research and could allow technology to take on new roles in our lives." The key is enabling AI to learn from their environments and faithfully represent an abstract version of them in their "heads," the way humans and animals do. To do it, developers need to train AIs by using simulations of the world. Think of it like learning to drive by playing "Gran Turismo" or learning to fly from "Microsoft Flight Simulator." These world models include all the things required to plan, take actions and make predictions about the future, including physics and time... There's an almost unanimous belief among AI pioneers that world models are crucial to creating next-generation AI. And many say they will be critical to someday creating better-than-human "artificial general intelligence," or AGI. Stanford University professor and AI "godmother" Fei-Fei Li has raised $230 million to launch world-model startup World Labs...
Google DeepMind researchers set out to create a system that could generate real-world simulations with an unprecedented level of fidelity. The result, Genie 3 — which is still in research preview and not publicly available — can generate photo-realistic, open-world virtual landscapes from nothing more than a text prompt. You can think of Genie 3 as a way to quickly generate what's essentially an open-world videogame that can be as faithful to the real world as you like. It's a virtual space in which a baby AI can endlessly play, make mistakes and learn what it needs to do to achieve its goals, just as a baby animal or human does in the real world. That experimentation process is called reinforcement learning. Genie 3 is part of a system that could help train the AI that someday pilots robots, self-driving cars and other "embodied" AIs, says project co-lead Jack Parker-Holder. And the environments could be filled with people and obstacles: An AI could learn how to interact with humans by observing them moving around in that virtual space, he adds.
"It isn't clear whether all these bets will lead to the superintelligence that corporate leaders predict," the article concedes.
"But in the short term, world models could make AIs better at tasks at which they currently falter, especially in spatial reasoning."
And that requires "world models," which are "gaining momentum in frontier research and could allow technology to take on new roles in our lives." The key is enabling AI to learn from their environments and faithfully represent an abstract version of them in their "heads," the way humans and animals do. To do it, developers need to train AIs by using simulations of the world. Think of it like learning to drive by playing "Gran Turismo" or learning to fly from "Microsoft Flight Simulator." These world models include all the things required to plan, take actions and make predictions about the future, including physics and time... There's an almost unanimous belief among AI pioneers that world models are crucial to creating next-generation AI. And many say they will be critical to someday creating better-than-human "artificial general intelligence," or AGI. Stanford University professor and AI "godmother" Fei-Fei Li has raised $230 million to launch world-model startup World Labs...
Google DeepMind researchers set out to create a system that could generate real-world simulations with an unprecedented level of fidelity. The result, Genie 3 — which is still in research preview and not publicly available — can generate photo-realistic, open-world virtual landscapes from nothing more than a text prompt. You can think of Genie 3 as a way to quickly generate what's essentially an open-world videogame that can be as faithful to the real world as you like. It's a virtual space in which a baby AI can endlessly play, make mistakes and learn what it needs to do to achieve its goals, just as a baby animal or human does in the real world. That experimentation process is called reinforcement learning. Genie 3 is part of a system that could help train the AI that someday pilots robots, self-driving cars and other "embodied" AIs, says project co-lead Jack Parker-Holder. And the environments could be filled with people and obstacles: An AI could learn how to interact with humans by observing them moving around in that virtual space, he adds.
"It isn't clear whether all these bets will lead to the superintelligence that corporate leaders predict," the article concedes.
"But in the short term, world models could make AIs better at tasks at which they currently falter, especially in spatial reasoning."
AI needs... (Score:5, Interesting)
World models
The ability to constantly learn after initial training
Long term memory
Cabby Ride Violation (Score:1)
Re: (Score:2)
Those last two are the same thing.
The premise seems obvious (Score:2)
A bee has a simple model of reality constantly updated by its sense of its self and surroundings. If you constantly update a model on external observations of bees, you only get something like a first order approximation. It will never out-bee an actual bee that way.
Marketing (Score:2)
Look at Google's genie 3 page [deepmind.google]. Especially the credits. Half of those people are marketeers, people from agencies, and video editors. When you critically look at the video you basically see some poststamp size video generations that are OK looking, but no better than like open source models like hunyuan. They show two videos with impressive consistency but like 95% of the vids are just a couple seconds long. It is suggested that the generation is realtime, but there is no indication of what kind of hardware
There is already the real world to learn from... (Score:2)
how about the real world (Score:2)
World Labs could make super hats, like little google maps cars. Then 10,000 interns could go (both) on with their lives and on assignment ( 2 hrs in park, take a ride out and back on the Green line, shop at the farmer's market, take a Porsche out on track day).
There's some real-world data.
Starting Over.. (Score:2)
superintelligence? Yea right! (Score:2)
yum farts (Score:2)
You could have AI learn from a simulated world, but you don't want it to be AI generated. You can't develop genius by sniffing farts.