NASA will launch spacecraft made of strange components entirely designed with artificial intelligence. The revolutionary technology of the North American agency designs and manufactures spaceship parts using the same generative AI technology that ChatGPT, Dall-e or MusicLM use to create text, images and music from human indications. But, unlike the previous ones, NASA technology does not make mistakes.
Generative AI has been used for the design of mechanical and architectural components for a long time. Multiple developers, from traditional assisted design software companies like AutoCAD of Autodesk to new companies like Ntopology, which have been selling programs to do it for years. There are even startups, like Divergent3D, whose business model relies entirely on generative AI design and manufacturing to create things like the rear frame of Aston Martin’s DBR22. The company told us that it had reduced the weight of that part by 40% and, furthermore, outperforming traditionally designed parts on its performance in the event of a collision. Three of the pieces made with different metal alloys. Three of the pieces made with different metal alloys. But, according to McClelland, he soon discovered that the tools might be up to the task of designing mission-critical parts.
“Most people just didn’t believe these parts could be made with that process until they saw it happen in real life.” This is how McClelland got into what he calls Evolved Structures.
Design at the speed of light
It makes sense because despite the alien look of the parts, the result performs better and is more efficient than more traditionally created parts, saving a third of the weight but getting the same performance values or exceeding part specifications. designed by humans. “The generative AI tool compresses that whole process,” he says. The artificial intelligence algorithm does 30 or 40 iterations in just one hour. “At most, a human team can do a couple of iterations in a week,” McClelland says.
However, this technology cannot function without human input. The same goes for the generative AI solution. Place holder Ryan McClelland. Ryan McClelland.
One of his current problems, for example, is that tools have a tendency to make things too fine at times, and that’s something, McClelland says, to be careful about.
Prompt engineering
As with generative AI imaging programs like DALL-E, Stable Diffusion and MidJourney, you have to find the right ‘prompt’, the text that contains the instructions for the AI to do what you describe. “To get a good result you need to give it a text with detailed instructions,” McClelland tells me, and the same goes for industrial design. Placeholder One of the pieces designed by artificial intelligence for the Excite ship. One of the parts designed by artificial intelligence for the Excite ship.
When you use generative AI in engineering, the ‘prompt’ is a series of geometric boundaries and physical specifications. The generative AI came up with the design from those basic instructions. McClelland says this is the skill an engineer will need to develop to use this technology successfully. The results are so good that NASA is going to use parts designed with this technology in several missions. The technology is also being used in a large number of other mission proposals, McClelland says.
To infinity and beyond
For McClelland, this technology represents the future of the industry at all levels and, therefore, that of humanity itself. “You tell the AI the function of a circuit and it designs it autonomously,” says McClelland. Placeholder Tycho Station, one of the space stations that inspired McClelland to investigate generative artificial intelligence. Tycho Station, one of the space stations that inspired McClelland to investigate generative artificial intelligence.
“All these things are coming, some before others, but that’s the big vision of the future.” And no, we are not going to have a text field in which to write “build a spaceship to travel to Jupiter in 30 days” waiting for the AI to design and print it with gigantic 3D machines. But we’re headed there and it’s an irresistible drive, as McClelland says. Afterwards, he couldn’t leave it and humanity has come to this point in history.
Generative AI is the monolith that will elevate us to a new plane and blaze a hyper-fast path to a bright future for humanity. So it seems fitting that the first children of this technology to go into space have the appearance of being back on their alien mothership.
