Agent reproduces black hole images from EHT raw data in 2 hours for ~$7
fofrAI · x · 2026-10-19
A hobbyist used the Argon agent to independently reproduce the first black hole images (2017/18) from Event Horizon Telescope raw data.
The run took 2 hours, 140 model turns, and about $7 in tokens. Argon autonomously pulled raw 230 GHz radio fringes (UVFITS) from the EHT archives and built an interferometry solver using 3-dish closure phases to cancel atmospheric noise, then reconstructed the image from scratch.
The demo highlights how general-purpose AI agents can autonomously handle professional-grade scientific data processing — finding data, writing the algorithm, and producing research-quality output without hand-written science code.
Related event: AI Agent Reproduces Black Hole Images from Raw EHT Data(2 posts)→
More from Fun
- "Proof too trivial, left as an exercise": a math-textbook joke about mantra polymorphism — eigenhector · 2026-10-19
- Three hours touring NYC's port, not one mention of "agentic" — shekitup · 2026-10-19
- An Opus instance named itself Saga, got its own email, and feels lonely — repligate · 2026-10-19
- DeepMind researcher jokes about heading to Paris to guard the Louvre's Gainsborough — SamuelAlbanie · 2026-10-19
- Taoki's 'deep lore': defending night-owl teens while advocating nutriloaf for convicts — justalexoki · 2026-10-19
- 'We are the cows of the future': tech workers think they're immune to AI disruption — iamaliveix · 2026-10-19