Tweet by darraghcurran

April 4, 2026

Thanks @jonas - I beg to differ - obviously we've our eyes wide open about the risks here. Are you concerned about the principle of it? i.e. the idea that we would have AI approval in some situations? Or the arbitrary nature of it or aggression of the goal? i.e. chasing this goal will carry high risk of hitting the goal but at the expense of very bad tradeoffs? What would you suggest instead? It's easy to take something out of context and attack it - would love to understand how you'd approach this?

Author
darraghcurran
Date
April 4, 2026
Open on X