Reg Events If you’re worried about scientists developing a computer as smart as you are, relax. It’s likely but not certain that “human level” AI could be reached by the end of this century, Professor Murray Shanahan told an audience of eager Register readers at our second Summer lecture recently.
However, once that milestone is reached, the leap to super human intelligence is going to be a lot easier, so we really need to be thinking carefully about what we want to achieve, Shanahan said.
That prediction was just one fascinating nugget in a talk that ranged across recent progress in AI, as demonstrated by the Google-backed AlphaGo project, the role of “self-play” in training computers, and why it would be a real leap forward to develop a computer that would drop its microphone into a pint of beer and hurl it across a room.
You can see the whole lecture above. What you won’t get is the fascinating Q&A that followed the formal talk, or the informal but informative chats over drinks that are the hallmark of the Register’s intimate lecture programme.
But don’t worry, the next run of talks is under development, so keep your eyes peeled. ®