One thing that many proponents of the harmless / weak / uninteresting AGI view the to ignore is that computers already have many superhuman capabilities. Obviou

One thing that many proponents of the harmless / weak / uninteresting AGI view the to ignore is that computers already have many superhuman capabilities. Obvious examples are math, rode memorization and similar. Even a AGI thats "general" intelligence is comparable to a very dumb human would reasonably have these superhuman capabilities readily available. It's hard to imagine how powerful even that combination might turn out, even leaving the obvious possibility of self re-engineering out.