AI on Trial: How Well Do LLMs Classify Astronomical Images?

This project has been built using the Zooniverse Project Builder but is not yet an official Zooniverse project. Queries and issues relating to this project directed at the Zooniverse Team may not receive any response.

AI on Trial: How Well Do LLMs Classify Astronomical Images?

Help judge how well an LLM explains astronomical images.

Learn more
Get Started!
  1. Look at the three images
  2. Read the explanation given by the LLM
  3. Decide if the description is coherent with the images
  4. Decide if the "interest" level assigned by the LLM is coherent with its own description of the image (even if the description of the image is wrong, we want to know if the LLM is self-consistent)

Zooniverse Talk

Chat with the research team and other volunteers!

Join in

AI on Trial: How Well Do LLMs Classify Astronomical Images? Statistics

View more stats

Keep track of the progress you and your fellow volunteers have made on this project.

Every click counts! Join AI on Trial: How Well Do LLMs Classify Astronomical Images?'s community to complete this project and help researchers produce important results. Click "View more stats" to see even more stats.

0%
Percent complete

By the numbers

0
Volunteers
0
Classifications
0
Subjects
0
Completed subjects

Message from the researcher

Connect with the research team on Talk to learn more about this project!

About AI on Trial: How Well Do LLMs Classify Astronomical Images?

For years, machines have learned from your input. Now, you’re the one testing them. You’ll receive three astronomical images—New, Reference, and Difference—and a prediction generated by an LLM. Judge how coherent and accurate that prediction really is.