“Nature is trying to tell us something here, that is, this isn’t really working, but the field is so trusting of its own press clippings that it can’t see what’s going on. there,” he added.
Even de Freitas’ DeepMind colleagues Jackie Kay and Scott Reed, who have worked with him on Gato, were more cautious when I asked them directly about his claims. When asked about whether Gato is headed for AGI, they weren’t intrigued. “I really don’t think it’s possible to make predictions with things like this. I try to avoid that. It’s like predicting the stock market,” Kay said.
Reed said the question is a difficult one. “I think most machine learners will try to avoid answering. It’s very difficult to predict, but you know, hopefully one day we’ll get there. “
In a way, DeepMind’s calling Gato a “generalist” may have made it a victim of the AI exaggeration surrounding AGI. Today’s AI systems are called “narrow” AI, which means they can only perform a limited number of specific tasks, such as generating text.
Some technologists, including at Deepmind, think that humans will one day develop “broader” AI systems that can work as well or even better than humans. Some call this “generic” artificial intelligence. Other say it’s like “belief in magic.” Many top researchers, such as Meta’s chief AI scientist, Yann LeCun question whether it is even possible at all.
Gato is a “generalist” in the sense that it can do many different things at the same time. But it’s a world beyond a “generic” AI that can meaningfully adapt to new tasks different from what the model was trained on, says MIT’s Andreas. “We are still a long way from being able to do that.”
Making the models bigger also won’t solve the problem that the models don’t have “lifelong learning”, meaning they can be taught things once and they will understand all the implications and use it to inform all the other decisions they will make, he said.
Emmanuel Kahembwe, an AI/robotics researcher and part of the Black in AI organization co-founded by Timnit Gebru, thinks the hype around tools like Gato is bad for the overall development of AI. “There are a lot of interesting topics that are left open, which are in short supply, that deserve more attention, but that is not something that the big tech companies and the majority of researchers in the companies are interested in,” he said. This tech company cares.
Vilas Dhar, president of the Patrick J. McGovern Foundation, a charity that funds AI projects “for good,” tech companies should take a step back and consider why they’re doing it. build what they’re building.
“AGI speaks to something deeply human – the idea that we can become more than what we are, by building the tools that bring us to greatness,” he said. . “And that’s really good, except it’s also a way of distracting us from the fact that we have real problems facing today that we should be trying to solve by using use AI.”
