A Harvard Law and computer science professor explains why recursive AI self-improvement could challenge human oversight and raise difficult questions about future safety.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results