Article originally published on LinkedIn, 16 June 2017
Recent research published by Facebook Artificial Intelligence Research (FAIR) – “Deal or No Deal? End-to-End Learning for Negotiation Dialogues” – shows that it is possible to train bots to do negotiations. Could this technology be used by job seekers and employers to arrive at optimal salaries?
What is interesting is that FAIR’s bots were negotiating with each other, not with humans, and apparently doing so quite effectively. So, if intelligent bots were to negotiate salaries, it would be on both ends – i.e., not just the employer or the pending employee benefiting from a negotiating bot, but both using them.
This capability could in theory fit in more complex AI processes that facilitate the range of tasks involved in job hunting on one side, and those in recruitment on the other. (Still thinking that such “intelligent agents” [IAs] will be a feature of the evolving job market in the near future.)
News about this research has generated articles focusing on interesting side aspects. One in The Atlantic highlights how the bots “developed their own language for negotiating.” Not long ago, Google found something similar in intelligent systems communicating with each other. This, then, would be an aspect of IAs working on behalf of job seekers and employers to expect and plan for – including how to translate information back into human languages where necessary (e.g., when a printed resume is needed or the outcome of a salary negotiation communicated).
Another article, in Wired, discusses how FAIR’s bots used deceptive tactics and lies. In negotiations for a position of trust (as any employer-employee relationship must be), this aspect could be problematic. It is one thing in a salary negotiation to be appropriately self-interested in sharing information, but another entirely to misrepresent or fabricate facts.
In any event, AI could have benefits in salary negotiations, especially for people who may be very skilled in their field, but not as strong on negotiating money matters. Used by both potential hires and organizations hiring them, it could lead to more fair outcomes between organization and employee, as well as laterally among employees across similar positions and vertically along hierarchies. (If AI handled CEOs’ salary negotiations, how might those turn out?)