Press
Source:
MIT Technology Review
A new public database lists all the ways AI could go wrong
"What our database is saying is, the range of risks is substantial, not all of which can be checked ahead of time."