The author of the linked article seems to contradict himself. He writes: >Gato’s ability to perform multiple tasks is more like a video game console that can s

The author of the linked article seems to contradict himself. He writes: >Gato’s ability to perform multiple tasks is more like a video game console that can store 600 different games, than it’s like a game you can play 600 different ways. It’s not a general AI, it’s a bunch of pre-trained, narrow models bundled neatly. But the article also quotes a DeepMind blog post, which reads: >The same network with the same weights can play Atari, caption images, chat, stack blocks with a real robot arm and much more, deciding based on its context whether to output text, joint torques, button presses, or other tokens. "The same network with the same weights" isn't a bundle of narrow models, it's one generalist model. I think that's more impressive than the article's author gives it credit for. Really feels like the beginning of C3PO type droids -- they don't have to be as smart or creative as a human, just adequate at a wide enough set of useful tasks