UPDATED 19:09 EST / JUNE 14 2017

EMERGING TECH

Facebook made an almost sneaky AI that negotiates better deals for itself

Can we use artificial intelligence to solve our everyday disagreements? Facebook Inc. thinks so, and today the social media giant published a paper that shows how an AI could be used to negotiate like a human.

The paper was published by researchers from the Facebook Artificial Intelligence Research lab, and they explained that negotiation is “an interesting task” for AI. Much like how AlphaGo proved that AI was capable of intuitive thinking, Facebook’s researchers claim their program proves an AI can weigh multiple outcomes and choose the most beneficial course of action for itself, even if it has to give up something in return.

In one example, the FAIR researchers showed two AI agents an uneven collection of items including two books, one hat and three balls. Each agent assigned their own value of desirability to the items, and they then had to divide the objects between themselves in such a way that they each received something that they wanted. According to Facebook, scenarios such as this helped the AI learn to negotiate for what they wanted, since a failed negotiation would result in neither agent getting what they desired.

Facebook said that not only does its AI negotiate for better deals, but it does so just like a human would: by talking it out. The researchers explained that the AI must plan ahead in conversations to anticipate how a deal might change over time. In other words, the AI must know how to build up to its final offer as if it is haggling. To accomplish this, the researchers developed what they call dialogue rollouts, where an AI essentially imagines the possible turns a conversation might take and plans for its maximum expected future reward.

Facebook’s researchers also tested their AI against human negotiators, which they said was more difficult than dealing with another AI agent. The reason for this is that while the AI avoids walking away from a deal without anything, humans are more likely to quit a negotiation if they cannot get everything they want.

The researchers said that as the AI improved, it would carry on longer negotiations with humans as it tried to reach a compromise. The AI also learned to use clever negotiating tactics such as feigning interest in a less valuable item so it could later give up to get a better deal. According to Facebook, its AI was so good at negotiating deals that most human participants did not even realize they were talking to a bot.

In addition to publishing its research, Facebook also open-sourced the code behind its AI, which the company says will allow it to improve even faster.

“This breakthrough represents an important step for the research community and bot developers toward creating chatbots that can reason, converse, and negotiate, all key steps in building a personalized digital assistant,” Facebook said in a blog post announcing the new research. “Working with the community gives us an opportunity to share our work and the challenges we’re aiming to solve, and encourages talented people to contribute their ideas and efforts to move the field forward.”

Facebook noted in its blog post that its negotiation AI could be “easily be adapted to other tasks.” If the social media giant is planning on taking over the world, this might be how it does it.

Photo: jeffedoe via Flickr (license)

A message from John Furrier, co-founder of SiliconANGLE:

Your vote of support is important to us and it helps us keep the content FREE.

One click below supports our mission to provide free, deep, and relevant content.  

Join our community on YouTube

Join the community that includes more than 15,000 #CubeAlumni experts, including Amazon.com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger, and many more luminaries and experts.

“TheCUBE is an important partner to the industry. You guys really are a part of our events and we really appreciate you coming and I know people appreciate the content you create as well” – Andy Jassy

THANK YOU