• Innovation

How AI Can Enhance Negotiation Skills

Professor Jared Curhan reveals the positive contribution AI is already making in supporting leaders with negotiation


By downloading this resource your information will be shared with its authors. Full privacy statement.

Conducting negotiations—both formal and informal—is central to all business activity, from enterprise-level mergers and acquisitions, right down to everyday budget approvals and service agreements. Our negotiating skills can always be improved, but how can AI enhance what is surely, essentially, a very human set of interactions?

Jared Curhan, Professor of Work and Organization Studies at MIT Sloan School of Management Executive Education points to two distinct areas—negotiation research and the practice of negotiation—which can benefit significantly from the deployment of AI tools. This is not just about generative AI tools such as ChatGPT and Microsoft’s Copilot. In this context a broader definition of AI applies: the ability to perform tasks normally associated with intelligent beings. This includes a wider range of technologies such as: machine learning, computer vision, affective computing, natural language processing, as well as generative AI. The groundbreaking ways these tools are now being deployed paints a remarkable picture of innovation for the near future of negotiation.

AI in Negotiation Research

The psychological complexity involved in how people negotiate is notoriously hard to penetrate. Understanding how some negotiators or negotiation strategies succeed and others fail to get past ‘no’ and to reach an agreement, can be a mystery—one that business leaders are ever keen to decipher. Where AI is helping is in research into negotiation, providing new levels of data, new types of analyses, and new kinds of consistent study partners.

Observations made through AI can detect fine nuances in conversation and facial expression that, plotted across multiple instances, can predict negotiation breakthroughs and when they are likely to stall. Professor Curhan’s research has shown that conversational dynamics during the first five minutes of a negotiation accounted for roughly 30% of the variance in outcome. The use of raised voice, the amount of speaking versus pausing and listening, and the different influences these and other dynamics have on outcomes, can be finely monitored. It can then be shown how results vary according to the status of the negotiators vis-à-vis each other—for example, the use of emphasis in conversation can lead to a more/less positive outcome depending of the relative status of the opposing negotiators.

Facial expressions can also be finely examined and used to identify responses, from disgust to engagement, that might otherwise go unnoticed. The poker face or the flinch can both tell stories. New kinds of analyses are also made available by AI. Observing patterns in negotiating dynamics, such as the use of silence and pauses in talking, can point to potential breakthroughs and help in recognizing opportunities for shared value creation. AI in the form of affective computing enables this new level of granularly detailed facial and conversational analysis, and the research that is now leading to better understanding of negotiating and ultimately better negotiating practice and outcomes.

Generative AI can be used to create malleable study partners, or confederates, that can be manipulated to act as particular negotiating personas. This allows participants to try out negotiating with, for example, an ‘Obama’ persona or a ‘Trump’ type. There are various approaches to how a GAI bot can be manipulated. You can set limits for it to operate within, or ask it to act like a specific stereotype. The most effective way is to use interpersonal psychological theory and set up the bot with emotional characteristics—warm/cold, dominant/submissive. Better still is to use a ‘mixed motive’ or ‘dual concern’ model, whereby a bot reacts in a warm, empathetic way when interacting with people, but is firm and cool when focused on the problem.

AI in Negotiation Practice

There are two ways AI can be deployed in real-time negotiations: as an agent, or as a teacher. Generative AI tools can be used to negotiate on your behalf—as your agent. Walmart is already doing this, using GAI to negotiate with its suppliers. Nibble, an open platform for small traders, deploys techniques based on behavioral science to act as an agent for e-commerce and B2B sales. The Nibble bot recognizes user behaviors and adapts its own behavior in response.

As a teacher or guide, AI can be used to trace negotiations and offer advice. A GAI bot can be employed to simulate an opposing negotiator’s style, allowing realistic testing of a strategy or approach against the bot. Following specific feedback on how well the negotiation with the bot has gone, the bot’s memory can then be wiped and the test repeated.

MIT Sloan Executive Education offers two new courses—Negotiation Essential Sprint, and Negotiation Strategies (coming soon)—led by Professor Curhan, that allow students to test their skills against bots that are prompted to act as human negotiators would, and to practice responding to difficult, but typical, negotiator tactics such as stonewalling, attacking, or playing tricks.

• •

While bots negotiating with bots may well happen in less complex situations, business negotiations will largely remain a highly human activity. The real benefit that AI offers business people is its ability to improve research into the psychology and operational minutiae of negotiations, and its capacity to assist negotiators in practice. With the help of AI, the essentially human input on both sides of a negotiation, will become far better informed and outcomes greatly improved.


This article is based on the MIT Sloan Executive Education webinar: ‘AI and the Future of Negotiation,’ with Jared Curhan, Professor of Work and Organization Studies at the MIT Sloan School of Management, hosted by Rob Dietel, Director of executive Programs at MIT Sloan.

MIT Sloan is uniquely positioned at the intersection of technology and business practice, and participants in our programs gain access to MIT’s distinctive blend of intellectual capital and practical, hands-on learning.

Google Analytics Alternative