AI agents combed public astronomy archives over a few weeks, discovering candidate asteroids and dwarf planets that institutional surveys missed. Will Depue and his collaborators targeted unexamined telescope data to search for Planet 9, uncovering a long tail of overlooked solar-system objects.
Hunting Asteroids With AI Agents
Government-funded telescopes capture nightly images of the sky, but most surveys focus on specific targets. That leaves massive archives of unexamined sky sitting online. Will Depue said on the MTS podcast that he and his team used Claude and Codex agents to comb through these public astronomy archives over a few weeks.
“There’s just tons of this data out there and a lot of it hasn’t been totally deeply finely combed through because people just don’t have the time to look for every possible object that appears in any of these data sets,” Depue told BigGo Finance.
The project started as a search for the hypothetical Planet 9 and drifted into cataloging other discoveries. Depue worked alongside collaborators Luke Farritor and Josh Wall, pulling night shifts until 5 a.m. to run the agent searches.
Why NASA Missed the Long Tail
NASA and institutional observatories possess the same cloud access to public data, but their priorities lie elsewhere. Astronomy talent focuses on higher-priority targets rather than the long tail of small solar-system objects. Depue expects major institutions to keep capturing headline discoveries while individuals pick off the rest of the data.
“OpenAI only published the top 400 most important results they got,” Depue said.
“There’s a massive number of unimportant results they didn’t publish or that people independently are discovering.”
This institutional blind spot forms the core business case for automated data mining. When compute is abundant and heavy software engineering costs drop, small teams can exploit public datasets that major agencies lack the staff to search.
Funding Whimsical Research Through Macro Grants
Depue directs his nonprofit, Macro Grants, toward whimsical research projects that struggle to secure conventional funding. Grants range from $500 to $25,000, with decisions returned in two to three days through a frictionless application process. The organization announced a $100,000 amateur astronomy grant alongside the podcast episode.
The nonprofit’s review board supports projects like Mecan’s LIDAR archaeology flights over Central American jungle.
Once sensors are mounted on small planes, the number of Mayan structures found becomes a direct function of flight time.
“The second any problem in the world is simply a function of compute, of dollars, of tokens, it’s really easy to find those inputs,” Depue said.
The Fear of AI Displacement
Depue offers a stark forecast for human labor over the next couple of decades. He compares human skill to Magnus Carlsen playing against Stockfish 14, where human mastery inevitably yields to superior machine capability.
“We’re headed towards a world in which I am like you know would be equivalent to what Magnus Carlson is to Stockfish 14,” Depue said.
Humans will retain economic value only in areas where human origin remains the primary point, such as art, politics, and creative work. Depue argues that natural selection will favor machines because self-replication and self-modification allow artificial intelligence to achieve massively higher fitness than biological humans.
Common Questions About AI Astronomy
What data sources did the team use?
The team combed public archives from American, European, and Japanese government-funded telescopes, utilizing nightly sky imaging from both ground-based and orbital instruments.
Who collaborated on the astronomy project?
Will Depue worked alongside collaborators Luke Farritor and Josh Wall to run the Claude and Codex agents over a few weeks.
How much do Macro Grants award?
Grants range from $500 to $25,000, with decisions delivered in two to three days.
Related reading