I think I found a planet nobody knew existed. I used Claude Code to find it
An amateur using Claude Code on data from NASA’s TESS telescope claims to have found a previously unknown exoplanet and even secured a follow-up observation program to verify the signal. Commenters are intrigued by the potential of AI-assisted research but stress the need for independent confirmation, rigorous statistics, and awareness of false positives when mining massive astronomical datasets. A parallel thread critiques the overblown, low-density prose often produced with AI tools and weighs the trade-offs between solo, AI-augmented discovery and participating in formal citizen-science programs run by agencies like NASA and ESA.
Scientific validity and verification
- Many are impressed but emphasize that without independent confirmation it remains a candidate, not a confirmed exoplanet.
- Some insist that peer review alone is insufficient; real verification requires an independent analysis of the data.
- Others push back against dismissing it solely for lacking academic review, arguing that early-stage, “hacker-style” work is in the spirit of the site.
- A TESS follow-up observing proposal for this star was reportedly approved (Program #100), but whether this counts as “NASA pointing a telescope at it” is debated; earlier claims were traced to an AI-generated summary, not the original post.
False positives and methodology
- Commenters worry about false positives when using coding agents on raw astronomical data; “hallucinations in data reduction” are a serious concern.
- Suggestions include: validating on data with known planets, injecting synthetic signals, and running the pipeline on shuffled/noise data to see spurious detections.
Use of AI tools
- Some view this as an excellent example of AI-assisted research and speculate about models trained specifically on scientific workflows and agentic loops.
- Others note that if humans can define clear statistical criteria, non-“intelligent” automation might suffice; LLMs are just one option.
Writing style and “AI slop”
- A large subthread criticizes the article’s prose as overdramatic, low–information-density, and “AI-ish.”
- Some readers appreciate warnings when text appears LLM-generated so they can avoid it; others are tired of constant “this is AI” complaints.
- Multiple comments describe cognitive fatigue from reading verbose, generic AI-style text and prefer raw bullet points or transcripts.
Citizen science vs collaboration
- Some find it inspiring that an individual with AI tools can do meaningful work from home.
- Others argue participating directly in NASA/ESA citizen-science programs would yield better, more efficient science and shared credit.
- Counterpoints stress that hobbies need not be collaborative from the outset and that solo exploration can still feed into institutional efforts later.
Why it wasn’t found earlier
- Explanations include enormous data volumes, computational limits, and that the signal may sit near existing noise thresholds.
- Several note that “obvious” low-hanging fruit is often missed because everyone assumes someone else already picked it.