Reaction
Alignment fears over a hypothetical Mythos-class AI model that might ‘crash out’
A reposted comment hopes DeepMind’s alignment work is strong, citing what it calls Gemini’s tendency to ‘crash out.’
TLDR
A reposted comment says a ‘Mythos class’ model with what the commenter calls Gemini’s tendency to ‘crash out’ would be ‘pretty scary.’ The commenter also hopes DeepMind’s alignment work is strong.
Combined views
42
1 Source, first seen ago