Publications

AI in Science: The 42 problem in AI-assisted research

Spitale G., Germani F. (2026) eLife Sciences 15:e113137 10.7554/eLife.113137

Read the paper ↗
Cite

In two lines

Like Douglas Adams’s 42, an AI summary can sound right and still be useless once it is cut off from the question and context behind it. We argue that AI literature tools should be judged on how easily readers can recover what they left out, and propose three ways to measure that.

Abstract

Large language models are very good at providing concise answers. But if we only see the answer without knowing the original question or the context behind it, that answer may not be very useful. The important issue is not simply how much information an AI summary removes but whether we can still recover the missing context. Thus, AI literature tools should also report how much information was removed and how recoverable their omissions are.

Cite

Spitale, G., & Germani, F. (2026). AI in Science: The 42 problem in AI-assisted research. eLife Sciences, 15, e113137. https://doi.org/10.7554/eLife.113137
@article{spitale2026ai,
  author  = {Spitale, G. and Germani, F.},
  title   = {{AI in Science: The 42 problem in AI-assisted research}},
  journal = {eLife Sciences},
  year    = {2026},
  volume  = {15},
  pages   = {e113137},
  doi     = {10.7554/eLife.113137}
}

Published in

eLife Sciences

Year

2026

DOI

10.7554/eLife.113137

Topics

,
← Earlier