6 ms·
Pretty sure any superhuman AGI that decides to scare humans can figure out how, whether by direct experiment or by, you know, reading stuff on the Internet. Tra
by throwanem 10y ago
Pretty sure any superhuman AGI that decides to scare humans can figure out how, whether by direct experiment or by, you know, reading stuff on the Internet. Training a neural net doesn't strike me as much of an increase in that risk, assuming you think that's a risk.