Glossary · term
Benchmaxxing
The process in which labs build synthetic data near benchmark distributions, cultivating "intelligence spikes" that cover the test points. "Training on the test set" has become "a new art form." As a result, benchmarks lose their credibility as a proxy for general capabilities.
Karpathy2025Wave 1 · 2023Maturity: 2/5
Maturity rationale
Benchmaxxing — critical term in circulation
References
Author: Andrej Karpathy
Cite or link to this resource · CC BY 4.0
Free to quote, link, and reuse with attributionReferencing this analysis in documentation, research, or an engineering write-up? Copy a formatted link or citation below:
Show citation text & canonical URL
Markdown:
[Benchmaxxing — AI Glossary — Skills Intelligence](https://www.skillsintelligence.tools/glossary/term/benchmaxxing)APA:
Skills Intelligence. (2025). Benchmaxxing — AI Glossary. Skills Intelligence AI Glossary. Retrieved from https://www.skillsintelligence.tools/glossary/term/benchmaxxing