5/Been Kim discusses AI explainability. AI has taken an engineering-centric approach, where researchers devise techniques via trial and error, and she urges developing fundamental scientific principles that make explanations more trustworthy and accurate.

→ View original post on X — @andrewyng