DeepMind Launches ‘Gato’ an AI that can perform hundreds of Tasks
The great challenge in the field of AI is developing a system that integrates an artificial general intelligence, or AGI. Such a system must be able to understand and master any task that a human being would be capable of. However, this week the DeepMind research lab just announced the release of Gato, an AI capable of learning and managing hundreds of tasks. AI is more versatile than ever.
To be precise, Gato is capable of 604 very different missions: lettering on a picture, engaging in dialogue, stacking boxes with a robotic arm, playing old Atari games. The device would be capable of performing very different actions, while current examples, like that of Ithaca, focused on a very specific type of task.
Although general artificial intelligence has long remained science fiction, Gato is not the first of its kind. Google has started using a Unified Multitasking (or MUM) model in its search engine, which interprets text, images, as well as videos to perform a task more efficiently. From a software architecture point of view, Gato is not very different from its predecessors. The difference, as Jack Hessel, a researcher at the Allen AI Institute, points out is instead in the variety of inputs Gato can interpret and tasks it can perform. Like other artificial intelligence systems before it, it learned based on billions of words, images of real or simulated environments, but also keystrokes or even symbols. A property that gives greater versatility to the system, since the interactions between these different types of inputs multiply the number of possible services.
Gato would outperform an expert more than half the time. For his part, assistant professor of information at the University of Alberta, Matthew Guzdial, remains skeptical. To say that this is a big step towards AGI is a bit of an exaggeration for them because we haven’t gotten to human intelligence yet and I don’t think we’re likely to get there any time soon. . Personally, I’m more in the camp of many smaller models, but these general models definitely have advantages in terms of performance on tasks outside of their training data.
Newer Articles
- Starbucks Plans to invest in Web 3.0.
- System Debugging: Advances in Software Vulnerability
- New Technology guides Autonomous Vehicles safely through a highly uncertain environment
