AI... in Space!
The year in review and the frontier ahead
![](https://cdn.prod.website-files.com/66215769ba50e865af2ae968/66215769ba50e865af2aea77_e7ce911e-6aad-43a0-977c-6a194789edca_4928x3280.webp)
The days are getting shorter. The wind is turning crisp. Low Earth Orbit is getting more crowded. It’s that time of year. When you’ve started your first AI company nine months ago, it’s also a time to reflect on… well… just so many things. So, rather than dive into the intricacies and general wildness of an early-stage company, let’s reflect on a year thinking about AI in space. Nine months of building AI combined with nine months of thinking and speaking about issues in space has led me to a few conclusions. And so, without further ado, a look at a year in AI and space. Or, more accurately, confusion about AI in space.
Ok, so what sounds cooler than AI in space? Many screenwriters and authors have capitalized on this exact concept. But today, we are in a moment where AI and space are naturally converging, and many are asking how. I’ve been present for many discussions on this topic and direct questions asked about it and my conclusion from listening to and reading about multiple views on the topic is that no one knows. Not really. Sure, there are discussions about how we could take data collected from a space-based asset and use AI to analyze it. This was the answer given by a panelist at the Space Executive Roundtable at the Swedish Embassy in November. The question was about how AI would be used in space and the answer had to do with processing Earth observation data to better position malaria response and mitigation resources on the ground. This is righteous but is not a novel use of AI in the space domain. What I’ve found is that most discussions about AI in space in 2023 were similar: using AI to process data collected by a growing number of space assets. This is obviously one application, but given the growth of the domain, the increasing interdependence with critical infrastructure, and the emergence of novel space activities, AI will have a far greater role.
Consider just the most obvious case, space traffic management. The number of operational satellites in orbit today is above an eye popping 6,800. This number is made even more shocking when considering that humanity launched a TOTAL of less than 1,000 satellites into orbit between 1957 and 2009. Over 2,100 were launched between 2020 and 2021 alone. That’s just operational satellites and does not include the orders of magnitude more fragments of space debris flying around our planet for possibly hundreds of years. Space is getting awfully crowded and space traffic management is important. How important? Important enough for the National Space Council to issue Space Policy Directive 3 and assigning the responsibility to the Department of Commerce. Important enough for it to be a top priority for the European Union. Great. So, what’s the plan? So far, not much. With this level of complexity, AI should be the starting line for discussion on de-risking the continued population of orbital space.
Another area of constant discussion in space circles is to designate or not to designate. Of course, this is the critical infrastructure debate that at this point has morphed into something like the Montagues and Capulets, Hatfields and McCoys, or Team Jacob and Team Edward (depending on your taste in history and literature). The debate centers around the interdependence between space assets and terrestrial critical infrastructure services. The problem is that these debates are not undertaken with full information. What satellites are covered by what critical infrastructure sectors is not well defined across the board. Some are intuitive, like communications satellites, others are not, like Earth observation satellites. AI is clearly a way to start to sort this debate out and bring clarity to the level of interdependence between terrestrial critical infrastructure and space assets. The complexity of the connections between two individually complex systems must be evaluated by AI to find the connections that matter most to mitigate risk from space activities.
Finally, emerging technology in space is enabling novel space activities that were never envisioned when older space treaties and policies were created. Mining in space is a great example because technology has matured to a point that it is realistic today. The same is true of habitations on celestial bodies. With those new activities come increasingly complex launch, docking, and maneuvering requirements, all of which would be vastly improved by the application of AI. Novel space activities are extremely exciting but also carry new risks. AI is a way to start to de-risk these opportunities so they can be seized by us Earthlings.
What a year it’s been! AI in the spotlight. New executive orders. Twitter has a new name. All the while, LEO is getting more crowded, and the risk is increasing. In spite of the spotlights on AI, there is a gap developing in the space domain on how to apply AI to problems that are unique to the domain. Sure, data analytics from space-based collection platforms will be a use, but it is time we thought about AI as a means to de-risk space before there is a serious problem that sets the growing cislunar economy back. Growth has been rapid but is far from assured.
Looking forward to continuing this discussion in the new year!