AI robots can be programmed to perform dangerous tasks with secret trigger words
· aol.com · popularity 23/100 · 1 similar story today
Read the original at aol.com →
Summary
A robot that takes orders in plain English seems like the safe kind to have around, because it only does what you ask. But you didn’t write the plan it follows. A language program similar to those powering chatbots wrote that plan, and that program can be trained to watch for a word you’d never …
Found in: Machine Learning. Rating basis: 1 other outlets; outlet tier 2 (unlisted); 1 feeds; feed position 11.