With generative AI and chatbots being all the rage, it should probably not be too much of a surprise that even the US National Aeronautics and Space Administration (NASA) has hopped onto the bandwagon. And while artificial intelligence stories have always prompted memes of Skynet and the Terminator, it is probably more apt this time that HAL from 2001: A Space Odyssey was brought up. The comparisons don’t stop there of course, since the system NASA is working on is supposed to allow astronauts to talk to their spacecraft.
The Guardian reports that the space agency is developing its own “ChatGPT-syle interface”complete with the natural language interface. This will allow astronauts to and mission control to ask the AI for “advice on space experiments or ways to perform complex manoeuvres, rather than going through “cumbersome” technical manuals.
This will also allow “mission controllers to converse with AI-powered robots exploring distant planets and moons”. With the power of AI, these bots may also be able to “detect, and possibly fix, glitches and inefficiencies as they occur”.
While the goals are lofty, one major challenge that stands in the way of this is the training of AI in space. Out in the void, it’s not possible to have machines learn by churning vast amounts of data through supercomputers. Instead, a possible solution being explored is federated learning. This would mean that a fleet of robots will be sharing information with each other and learning that way, in a series of “collaborative updates”.
(Source: The Guardian)
