Are you beuntamederd about synthetic ambiguous intelligence, or AGI? It’s that skinnyg OpenAI is obsessed with ultimately creating in a way that “advantages all of humanity.” You may want to get them solemnly since they equitable elevated $6.6 billion to get sealr to that goal.
But if you’re still wondering what the heck AGI even is, you’re not alone.
In a wide ranging converseion on Thursday at Credo AI’s reliable AI directership summit, Fei-Fei Li, a world-commemorated researcher standardly called the “godmother of AI,” said she doesn’t understand what AGI is either. At other points, Li converseed her role in the birth of up-to-date AI, how society should shield itself agetst evolved AI models, and why she skinnyks her new unicorn beginup World Labs is going to alter everyskinnyg.
But when asked what she thought about an “AI singularity,” Li was equitable as lost as the rest of us.
“I come from academic AI and have been directd in the more rigorous and evidence-based methods, so I don’t repartner understand what all these words unbenevolent,” said Li to a packed room in San Francisco, beside a big triumphdow disseeing the Gagederen Gate Bridge. “I frankly don’t even understand what AGI unbenevolents. Like people say you understand it when you see it, I guess I haven’t seen it. The truth is, I don’t spfinish much time skinnyking about these words becaemploy I skinnyk there’s so many more convey inant skinnygs to do…”
If anyone would understand what AGI is, it’s probably Fei-Fei Li. In 2006, she produced ImageNet, the world’s first big AI training and benchtaging dataset that was critical for catalyzing our current AI boom. From 2017 to 2018, she served as Chief Scientist of AI/ML at Google Cdeafening. Today, Li directs the Stanford Human-Caccessed AI Institute (HAI) and her beginup World Labs is erecting “big world models.” (That term is proximately as confusing as AGI, if you ask me.)
OpenAI CEO Sam Altman took a stab at defining AGI in a profile with The New Yorker last year. Altman portrayd AGI as the “equivalent of a median human that you could employ as a cotoiler.”
Meanwhile, OpenAI’s charter expounds AGI as “highly autonomous systems that outcarry out humans at most economicpartner precious toil.”
Evidently, these definitions weren’t quite excellent enough for a $157 billion company to be toiling towards. So OpenAI produced the five levels it internpartner employs to gauge its evolve towards AGI. The first level is chatbots (appreciate ChatGPT), then reasoners (apparently, OpenAI o1 was this level), agents (that’s coming next, presumedly), innovators (AI that can help conceive skinnygs), and the last level, organizational (AI that can do the toil of an entire organization).
Still beuntamederd? So am I, and so is Li. Also, this all sounds appreciate a lot more than a median human cotoiler could do.
Earlier in the talk, Li said she’s been intrigued by the idea of intelligence ever since she was a juvenileer girl. That direct her to studying AI lengthy before it was profitable to do so. In the timely 2000s, Li says her and a scant others were mutely laying the set upation for the field.
“In 2012, my ImageNet united with AlexNet and GPUs – many people call that the birth of up-to-date AI. It was driven by three key ingredients: big data, neural nettoils, and up-to-date GPU computing. And once that moment hit, I skinnyk life was never the same for the whole field of AI, as well as our world.”
When asked about California’s contentious AI bill, SB 1047, Li spoke nurturefilledy to not rehash a dispute that Governor Newsom equitable put to bed by vetoing the bill last week. (We recently spoke to the author of SB 1047, and he was more enthusiastic to reuncover his argument with Li.)
“Some of you might understand that I have been vocal about my troubles about this bill [SB 1047], which was vetoed, but right now I’m skinnyking convey inantly, and with a lot of excitement, to see forward,” said Li. “I was very flattered, or honored, that Governor Newsom askd me to participate in the next steps of post-SB 1047.”
California’s ruleor recently tapped Li, alengthy with other AI experts, to create a task force to help the state broaden protectrails for deploying AI. Li said she’s using an evidence-based approach in this role, and will do her best to aid for academic research and funding. However, she also wants to determine California doesn’t punish technologists.
“We insist to repartner see at potential impact on humans and our communities rather than putting the burden on technology itself… It wouldn’t produce sense if we penalize a car engineer – let’s say Ford or GM – if a car is misemployd purposefilledy or unintentionpartner and harms a person. Just penalizing the car engineer will not produce cars safer. What we insist to do is to persist to originate for safer meacertains, but also produce the regulatory summarizetoil better – whether it’s seatbelts or speed restricts – and the same is real for AI.”
That’s one of the better arguments I’ve heard agetst SB 1047, which would have punished tech companies for hazardous AI models.
Although Li is advising California on AI regulation, she’s also running her beginup, World Labs, in San Francisco. It’s the first time Li has set uped a beginup, and she’s one of the scant women directing an AI lab on the cutting edge.
“We are far away from a very diverse AI ecosystem,” said Li. “I do consent that diverse human intelligence will direct to diverse synthetic intelligence, and will equitable give us better technology.”
In the next couple years, she’s excited to convey “spatial intelligence” sealr to fact. Li says human language, which today’s big language models are based on, probably took a million years to broaden, whereas vision and perception anticipateed took 540 million years. That unbenevolents creating big world models is a much more complicated task.
“It’s not only making computers see, but repartner making computer understand the whole 3D world, which I call spatial intelligence,” said Li. “We’re not equitable seeing to name skinnygs… We’re repartner seeing to do skinnygs, to direct the world, to convey with each other, and closing that gap between seeing and doing insists spatial understandledge. As a technologist, I’m very excited about that.”
