Bringing generative artificial intelligence into space

TAMPA, Fla. — Amazon Web Services is focusing on equipping its cloud infrastructure business to harness the potential of generative artificial intelligence to transform space and other industries.

More than 60% of AWS’s space and aerospace customers are already using some form of AI in their business, up from single digits about three years ago, according to Clint Crosier, director of aerospace and satellite at AWS.

Crosier predicts a similar growth in generative AI in space over the next few years. Use deep learning models to answer questions and create content It is based on patterns detected in large datasets and is a major advancement over traditional machine learning algorithms.

Advances in mathematics, an explosion in the amount of available data and cheaper, more efficient chips to process it are a “perfect combination” for the rise of generative AI, he said. Space News In an interview, he said that the company is helping accelerate the adoption of cloud-based applications.

“Last year, AWS fundamentally restructured its internal structure to ensure it had the right teams in place. [and] “We now have the organizational structure in place to really focus on generative AI,” he said.

He said AWS has formed a small team of people called “Generative AI for Space” to work with cloud customers and help develop next-generation capabilities.

These efforts include a generative AI lab for customers to try new ways to use these new capabilities.

Crosier lists three main areas where generative AI will be used in space: geospatial analysis, spacecraft design, and constellation management.

Earth observation satellite operators such as BlackSky and Capella Space are already using these tools to manage search queries and gain deeper insights into geospatial data.

While it’s still early days in manufacturing, Crosier said engineers are experimenting with how generative AI models, fed with design parameters, can generate new concepts by utilizing data that might otherwise be overlooked, such as in the automotive industry.

“Whether you’re designing a satellite or a rocket or a spacecraft, you’re having generative AI do some expeditions around the world with decades of data,” he said. “Then it comes back and brings up novel design concepts that no one has thought of before that can serve as a baseline for the team to start improving.”

He said generative AI could also help operators manage increasingly congested tracks by helping to simulate test scenarios.

“If you have a constellation of 600 satellites, you want to model how that constellation will perform under different design parameters,” he said.

“Well, you can get models of two concepts, but that’s not enough. But it takes time and money to model those. Or you could model an infinite number of them. Gen AI tells you the top 25 cases to model in your modeling simulation capabilities, which gives you the best design optimization. So that’s what we use it for.”

AWS is also working to accelerate the adoption of these new computing capabilities through scholarships and a commitment announced in November to provide free AI training to 2 million people worldwide by the end of 2025.