DeepMind's Gato Model Shows Promise Beyond Exaggerated AGI Claims
DeepMind's new Gato model handles 604 distinct tasks simultaneously, representing a notable step in AI versatility. However, experts warn that exaggerated hype about human-level artificial intelligence overshadows the model's actual achievements.
DeepMind unveils Gato, a new "generalist" AI model that performs 604 distinct tasks including playing Atari games, captioning images, chatting, and controlling a robot arm. The model attracts massive attention after a DeepMind researcher claims it provides a clear path to artificial general intelligence, sparking breathless media coverage about the imminent arrival of human-level AI.
This intense hype frustrates many AI experts who see it as a disservice to the broader field, similar to previous exaggerated reactions to models like GPT-3 and DALL-E. Critics point out that focusing on sensationalized claims about superhuman AI distracts from the actual technical progress and valuable research happening across the industry.
Despite the noise, Gato represents a genuinely interesting advance because it learns multiple tasks at the same time instead of forgetting previous skills to learn new ones. While it does not perform these varied tasks as well as specialized single-task models, this ability to seamlessly switch between 604 different skills without catastrophic forgetting marks a small but significant milestone in machine learning.