The 11th annual Olympic for hagglebots was held earlier this month for artificial intelligence (AI) that has been trained to negotiate.
Automated Negotiating Agent Competition
This competition pits more than 100 participants from different countries. For instance, these countries include Japan, France, Israel, Turkey and the United States against one another, in five leagues.
This competition would have been held in person (or in silicon) in Japan, as part of the International Joint Conference on Artificial Intelligence. However, due to the coronavirus, the competition was part of a virtual conference.
Universities from Turkey and Japan were the big winners this year, as their AI can haggle with humans and each other. They stimulate a factory manager doing supply chain management, and the game Werewolf.
In other leagues, they haggle with real-life human subjects recruited online or their bots negotiate with other bots.
“In the first years, [the AI] were really easily outperformed by human beings,” says the hagglebot games’ co-founder, Tim Baarslag. He is from the Centrum Wiskunde & Informatica, the Dutch national research institute for mathematics and computer science.
But now they are “behaving closer to human form, sometimes even better than humans, but only in very artificial domains,” he says.
Before artificial intelligence was created for negotiation, negotiating software is designed to help in this field. They appear to be a few negotiating support system in the early 1980s. For example, Inspire and Negoisst is both negotiating tools.
These tools tried to help negotiators prepare offers and strategies. Then, they help both sides arrive at good deals that “don’t leave money on the table”.
Uses of negotiating AI in business
Walmart tested an AI system developed by Pactum. The chatbot contacts the suppliers and invites them to renegotiate together contracts over things such as price and payment terms. In this case, there are up to 30 different points to settle with some suppliers.
“We now see that some vendors prefer talking to a bot. If there are tens of thousands of vendors, it’s hard to get human attention sometimes,” says Martin Rand, chief executive of Pactum.
This artificial intelligence often proposes unrelated offers to a vendor. For example, it will ask the vendor if they would rather increase their payment time by a certain number of days, or keep items in their own warehouse. Then they will learn the vendor’s preferences based on their answers.
Much of the negotiation is finding out which subjects matter more to their counterpart, and at what point they’d walk away.
With the machine learning technology these days, it can help an AI predict the other side’s preferences based on their observations. Besides, it can also gain insights and experience in prior negotiations.
More Information About: