← back
arXivVésteinn Snæbjarnarson, Anej Svete, Josef Valvoda, Reda Boumasmoud, Brian DuSell, Ryan CotterellMon, Jun 8, 2026, 10:58 AM PDT
score 17.2

Formal languages reveal flaws in how we measure AI learning

Original: Causally Evaluating the Learnability of Formal Language Tasks

Source: arxiv.org

Writing ELI5 summary…