A recent experiment in which an artificially intelligent chatbot became virulently racist highlights the challenges we could face if machines ever become superintelligent. As difficult as developing artificial intelligence might be, teaching our creations to be ethical is likely to be even more daunting.
