while i'm not a fan of ai regulation stifling other approaches, i DO think it's important that we build a version of ai assistant that is oriented towards supporting human agency rather than just smoothing away all friction between humans and their hedonic satisfaction