EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Six principles for evaluating cognitive capabilities in AI models

Melanie Mitchell

AI Magazine · 2026-06 · 47 (2)

DOI: 10.1002/aaai.70061 ↗

Kurz gefasst

Fachartikel in der Zeitschrift AI Magazine, online erschienen am 12.05.2026. Verfasst von Melanie Mitchell. Die Arbeit nennt 33 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Modern AI systems have exceeded human performance on many benchmarks meant to evaluate general cognitive capacities. However, it is often the case that benchmark performance does a poor job of predicting general capacities in real‐world settings. In this article I describe several issues related to evaluation that can cause this mismatch, and propose six principles, inspired by developmental and comparative psychology, that need to be adopted to enable rigorous evaluation for AI systems. These principles are illustrated by case studies from the psychology and AI literature.

Zitieren

APA

Mitchell, M. (2026). Six principles for evaluating cognitive capabilities in AI models. AI Magazine, 47(2). https://doi.org/10.1002/aaai.70061

Vancouver

Mitchell M. Six principles for evaluating cognitive capabilities in AI models. AI Magazine. 2026;47(2). doi:10.1002/aaai.70061

BibTeX
@article{Mitchell2026,
  author = {Mitchell, Melanie},
  title = {Six principles for evaluating cognitive capabilities in AI models},
  journal = {AI Magazine},
  year = {2026},
  volume = {47},
  number = {2},
  doi = {10.1002/aaai.70061}
}
Kostenlos starten →