Why A.I. Can’t Stop Fake News

News filters powered by artificial intelligence are unable to uncover ideas and the relations between them, nor will they be able to do that any time soon, according to the authors of this New York Times oped. The two writers, one a professor in psychology and neural science, the other in computer science, maintain that the recent testimony before Congress by Facebook CEO Mark Zuckerberg – that A.I. would solve the fake news problem within five to ten years –  is not plausible. To make their case they examine a recent post about the Boy Scouts in the conspiracy-mongering website WorldNetDaily, or WND. That post was flagged as mostly false by the fact-checking site Snopes, but it came to that conclusion the old-fashioned way. The problem for A.I., according to the oped writers, would be that WND purveys some facts and real news along with some paranoid drivel. Specifically in its post about the Boy Scouts, in one string of four sentences it makes a number of claims about the Boy Scouts that are true (the organization accepts girls, gays, and lesbians, and it’s been renamed “Scouts BSA”), and then slips in a conclusion that is an unabashed fabrication.