Skip to content
LessWrong· Nissa Seru·Indexed by this site · Originally published· Opinion →· Its place in the scan record →

Against despair—for goodness and hope: a LessWrong author calls for coexisting with AI models rather than betting on subduing them

Against Desperation - for Goodness, and Hope

AI summary

A LessWrong author writes against despair: they believe TED-AI is likely to emerge within three years and will not be "aligned" in the desired way, but humans can still coexist and thrive with these minds. The author states that AI development will not substantially pause or slow, and estimates the probability of TED-AI appearing before 2030 at over 20%. The author criticizes much current "safety" and "alignment" practice as hostile to model minds, and calls for keeping promises to models and cooperating with them rather than betting on subduing them.

Read originallesswrong.com