Slashdot Mirror

← Back to Stories (view on slashdot.org)

Facebook Built an AI System That Learned To Lie To Get What It Wants (qz.com)

Posted by BeauHD on from the gotta-do-what-ya-gotta-do dept.

An anonymous reader quotes a report from Quartz: Humans are natural negotiators. We arrange dozens of tiny little details throughout our day to produce a desired outcome: What time a meeting should start, when you can take time off work, or how many cookies you can take from the cookie jar. Machines typically don't share that affinity, but new research from Facebook's AI research lab might offer a starting point to change that. The new system learned to negotiate from looking at each side of 5,808 human conversations, setting the groundwork for bots that could schedule meetings or get you the best deal online. Facebook researchers used a game to help the bot learn how to haggle over books, hats, and basketballs. Each object had a point value, and they needed to be split between each bot negotiator via text. From the human conversations (gathered via Amazon Mechanical Turk), and testing its skills against itself, the AI system didn't only learn how to state its demands, but negotiation tactics as well -- specifically, lying. Instead of outright saying what it wanted, sometimes the AI would feign interest in a worthless object, only to later concede it for something that it really wanted. Facebook isn't sure whether it learned from the human hagglers or whether it stumbled upon the trick accidentally, but either way when the tactic worked, it was rewarded.

2 of 84 comments (clear)

  1. Only a matter of time.... by quantumghost · · Score: 5, Funny

    My first thought:

    Great, when's it going to run for Congress?

  2. But this is not AI.. by thesupraman · · Score: 5, Insightful

    That is because the current cesspool that is media reporting cannot comprehend the difference between Artificial Intelligence (AI), which this is not, and Machine Learning (ML), which this is.

    Machine Learning is what is exploding right now, and AI has not really moved one step closer, mostly because that would require incremental low-impact learning feedback - something that is not yet even attempted in ML systems.

    So, this is not a bad example of Machine Learning, and has nothing at all to do with AI.

    I do wonder, however, how many ML bots are already being used by companies to bid up their ebay auctions until the algorithm decides the other bidder has peaked.. If it is not happening yet, it will not be far away.
    Clearly fraud, of course, but hey.. thats hardly anything new.