Just around an hour ago, OpenAI released its o3 model. The model blew existing AI benchmarks out of the water, getting record scores in math and coding benchmarks. Crucially, its strongest version it managed to get a score of 87% on the ARC-AGI benchmark, while humans normally score around 85%. This has led many to speculate whether the o3 model represented humans achieving AGI, or Artificial General Intelligence, which means creating an AI model that’s better than humans at most tasks.
“o3 has scored over 85% on ARC-AGI. Human performance is at 85% In other words: AGI has been achieved in 2024 2025 is going to be wild!” wrote Kraftful CEO Yana Welinder.
Others said that AGI was here much before expected. “AGI was achieved today I tell it like it is: AGI has been achieved. Artificial general intelligence is here – and even faster than expected, in 2024. Arc-AGI was achieved and solved with 87.7% (threshold 85%). What’s more: new benchmarks are needed because almost all of them are saturated,” wrote X user Chubby.
Some said that the momentous occassion wouldn’t be understood by most outside the tech world. “AGI was just announced and not many people know,” wrote X user adonis_singh.
Others speculated on what this model’s results mean. “AGI next year. e/acc won. solving fusion. curing cancer. “fixing the climate, establishing a space colony, and the discovery of all of physics”. all of it. we live in the best timeline. strap in, folks,” wrote X user sporadicalia. “AGI has to be built. There is simply nothing more important, more necessary, for humankind to do. It is our final calling,” they added.
Some people marvelled at how quickly AI had developed. “This is incredible but also probably the first time we’ve seen up to date SOTA announced. to put a finer point on it, 2 or 3 years ago these numbers would have represented essentially consensus achievement of AGI,” wrote X user _xjdr.
There were some philosophical comments as well. “AGI is part of the evolutionary path of the cell to fully understand itself,” wrote X user @Plinz.
Others said that today’s date would become a significant turning point in history. “AD / BC no longer holds any real meaning to humans. life will simply be categorized as before AGI & after AGI. this is a monumental development & 99% of the world has no understanding of its significance & the underlying implications,” wrote X user @signulll. Their sentiments were echoed by X user VictorTaelin. “I feel like today will be historically regarded as the day AGI was soft-announced to the world. Yes, too many valid concerns, but, at least as far as my own intuition goes, unless they cheated somehow, this provides overwhelming evidence that we’re headed towards it, quickly,” they said.
People began predicting how this new development will change everyone activities. “When you have these godlike models on your glasses seeing the world with you and helping you make decisions, you will eventually handover most of the cognitive load to the AI and just do what it says. It will run you,” wrote X influencer Varun Mayya.
Former Stability AI CEO Emad Mostaque spoke about how the world was likely to change in the coming years. “My take on o3: the global economy is cooked, we need a new economic and societal framework. Any work that can be done on the other side of a computer screen, AI will be able to do at a fraction of the price,” he said.
But one of the most prominent voices in the field didn’t quite agree that OpenAI’s o3 model was AGI. Francois Chollet, who is the co-creator of the ARC-AGI benchmark that o3 scored higher than humans on, said that while the achievement was undoubtedly impressive, he didn’t think it was AGI. “So, is this AGI? While the new model is very impressive and represents a big milestone on the way towards AGI, I don’t believe this is AGI — there’s still a fair number of very easy ARC-AGI-1 tasks that o3 can’t solve, and we have early indications that ARC-AGI-2 will remain extremely challenging for o3. This shows that it’s still feasible to create unsaturated, interesting benchmarks that are easy for humans, yet impossible for AI — without involving specialist knowledge. We will have AGI when creating such evals becomes outright impossible,” he said.
While it might take a while for it to be clear whether OpenAI’s o3 model is really AGI, it does definitely seem to represent a marked rise in AI capabilities. It could be a while before the model is released to the public — OpenAI says it is currently under safety testing — but if the speculation by many on the internet is correct, 20th December 2024 might go down as the day when humanity managed to achieve AGI.