Campaign to Stop Killer Robots Warns U.N. of AI Threat ‘a Few Years Away’

The Campaign to Stop Killer Robots gathered experts in artificial intelligence, lawyers and activists at the United Nations on Tuesday to warn against a growing reliance on “stupid AI” that can be disastrous in the real world.

The Guardian reports:

… there are concerning technologies “only a few years, at best, away”, [said Toby Walsh, a professor of artificial intelligence at the University of New South Wales], and with semi-autonomous systems, such as drones, “it would take very little to remove the human from that loop and replace them with a computer”.

Walsh and other members of the campaign painted a dire picture of uninhibited artificial intelligence taking root on battlefields and national borders. Unlike those needed for nuclear weapons, they said, the resources to create “killer robots” will only become cheaper and more available over time. With an online-bought drone, a smartphone and the right software, anyone could create “a little killer robot”, Walsh added. …

“If we don’t [create rules], we will end up in an arms race,” Walsh said, “and the endpoint of that arms race is going to be a very undesirable place.” …

“This will seduce us into warfare,” Walsh said. “It will be too easy, we’ll think that we can fight these wars cleanly, and as we have seen I mentioned with the drone papers, that is a deception because we actually aren’t able to make that kind of technical distinctions between combatant and civilian.” …

The US and UK have policies on military AI, but the activists, who want the UN to formalize talks on “meaningful human control”, said that both countries’ rules were too ambiguous.

Read more here.

— Posted by Alexander Reed Kelly.