Teun van der Weij comments on [Paper] AI Sandbagging: Language Models can Strategically Underperform on Evaluations

[Teun van der Weij]

Ha, you’re clearly right. We will fix it.