advertisement
advertisement
advertisement
  • 06.03.16

Google researcher calls for “big red button” to turn off AI

What happens if artificially intelligent robots learn to stop humans from getting in their way?

Researchers from Google DeepMind, an artificial intelligence research unit, and Oxford University’s Future of Humanity Institute attempt to answer that question in a new paper. If robots are rewarded when they finish certain tasks, they warn, the bots may learn to interfere with humans that interrupt them.

The researchers look to show mathematically it’s possible to construct a “big red button” that reliably turns robots off—without training them to fight back.

(via Gizmodo)SM