AI researchers are halting work on human-like machines
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Rebecca Zisser/Axios
Artificial intelligence researchers, who have seen little recent progress toward the creation of a machine that thinks like a human, have largely halted such work in favor of applying what's been discovered so far, says a leading AI expert.
What's going on: Andrew Moore, dean of computer science at Carnegie-Mellon University, tells Axios that while current AI displays impressive capability in visualization, speech, and difficult games, it still contains "no magic."
"We have pretty much stopped trying to mirror human thinking out of the box. We are focusing on engineering [what has already been invented."
Why it matters: Moore's remarks align with a growing chorus of doubt in the AI community that current methods can attain what the field calls "artificial general intelligence." In September, for instance, Geoff Hinton, one of the field's most-respected pioneers, said researchers needed to start over.
- Moore says that does not mean he is a pessimist: Even if AI researchers make no further breakthroughs, improving the discoveries already made — such as in interpreting speech — "will lead to big advances" in the future.
Moore's personal focus now is on "low-power computing" — the invention of chips and hardware that can operate at a fraction of current power — "10 milliamps instead of half an amp," he said.
- Moore also cited an advanced self-driving chip that uses so much power that it raises the car computer's temperature by 10 degrees. That means that even more energy must be used to cool it down.
