An artificial intelligence (AI) coach that advises people who have been evicted from their homes on how to advocate for themselves in court. A contest in which individuals compete to see who can prompt ChatGPT to negotiate as effectively as possible. A “bargaining bot” that analyzes negotiation transcripts and offers counterparts rich feedback on how to improve their skills.
Claim your FREE copy: Negotiation Skills
Build powerful negotiation skills and become a better dealmaker and leader. Download our FREE special report, Negotiation Skills: Negotiation Strategies and Negotiation Techniques to Help You Become a Better Negotiator, from the Program on Negotiation at Harvard Law School.
These are just a few of the research projects presented at the Program on Negotiation’s 2025 AI Negotiation Summit, held March 8–9 at the Massachusetts Institute of Technology (MIT). In 22 presentations, the 200-plus attendees learned how AI can add value to negotiations by offering advice, assistance, training, and research support.
Many of the presenters described how AI can serve as a behind-the-scenes advisor to help people prepare for, conduct, and learn from their negotiations. Some explored how human negotiators perform against AI and how bots compete against each other. Others tested how well bots negotiate on behalf of humans, help us fairly divide tasks and resources, and mediate our contentious political disagreements (quite effectively, it turns out). And one panel covered how AI is contributing to negotiation research by coding transcripts, analyzing conversations, and serving as “virtual confederates” in experiments.
Throughout the weekend, the panels sparked spirited discussion and ideas for future collaboration between negotiation and computer science researchers, instructors, and practitioners.
Your Own Personal Negotiation Coach
AI can serve as a highly effective backstage negotiation coach, offering personalized negotiation training and feedback, according to multiple panelists.
For their courses on multiparty negotiations, Harvard Negotiation Project instructor Samuel Dinnar and MIT professor Lawrence Susskind created a tailored negotiation coaching bot to help their students learn more from complex dealmaking simulations. Rather than offering scripts or lessons for students to follow, the bot encourages deep thinking and consideration of context by asking questions, such as “What are you thinking about doing next?” and “Why do you think that group reached consensus and yours didn’t?” The bot reinforces the preparation and debriefing work that individuals do on their own and in groups.
The goal isn’t to produce cookie-cutter negotiators but to encourage each student to develop the negotiating style and approach that works best for them. As hoped, students who used the bot moved toward developing “their own personal theory of practice,” said Dinnar.
AI as Real-World Negotiation Trainer
A key takeaway from the AI Negotiation Summit: AI tools work particularly well when they account for the specific context of users’ negotiations. Panelist Zilin Ma of Harvard reported that in his research, humanitarian frontline negotiators—people navigating high-stakes conflicts such as ceasefire and hostage negotiations—found AI tools that factored in the context of their negotiations and leveraged their unique expertise to be more helpful than those that simply provided general negotiation advice.
Numerous presenters reported on how AI can level the playing field for those negotiating with more powerful or better-resourced counterparts. In eviction and truancy court, for example, litigants often have no access to lawyers and little understanding of their rights and options. Ohio State University professor Amy Schmitz described how her law school students consulted with such clients when designing an AI coach to give them legal and dispute-resolution advice. The JusticeTech project aims to reduce these litigants’ stress and improve their outcomes.
Similarly, Northwestern University professor Jeanne Brett and her team created an AI-based negotiation training bot, NegotiAge, to help family caregivers prepare for potentially stressful negotiations with health-care providers, insurance companies, family members, and others. In the research, participants who were caring for family members with dementia used the bot to role-play scenarios they might encounter, such as negotiating with a sibling over a parent’s care plan.
A large majority of participants said the training helped them be more productive and gave them more control over their lives. In addition, a month after the training, 74% reported using their newfound negotiation skills in their caregiving role—and 42% said they applied these skills to negotiations beyond caregiving.
Warmth and Dominance: A Winning Combination
According to conventional wisdom, people need to be aggressive and ruthless negotiators to compete with the quick, encyclopedic “brainpower” of AI. Yet several projects reinforced the past finding that it pays to be both assertive and warm in negotiations.
In MIT’s inaugural AI Negotiation Competition, for example, bots chose to exit negotiations with hypercompetitive opponents. “Bots walk away from you if they can,” said MIT professor and summit co-chair Jared Curhan. By contrast, bots whose style was dominant and warm performed well against bots in the competition.
And in another study, where pairs of humans engaged in a simple negotiation over the purchase of a table via chat, those coached by AI to be warm and dominant claimed the most value, according to MIT researcher Harang Ju. The coaching was most beneficial for those with a naturally frosty negotiating style.
AI as a Negotiator: The Risks Ahead
Although companies such as Walmart have successfully used chatbots to automate simple boilerplate negotiations, numerous summit panelists cautioned against giving AI negotiating authority over more complex negotiations with ongoing relationship partners.
One pitfall: When assigning AI to bargain on our behalf, we condone less ethical behavior, according to research by summit co-chair Jonathan Gratch (University of Southern California). This finding is in line with past research showing that we feel more morally detached from talks when negotiating through an agent.
In addition, AI could make our negotiations less creative. USC researcher Zhivar Sourati described how his team analyzed the linguistic features of online text (found on Reddit, Patch local news sites, and open-source research repository arXiv) before and after the public release of ChatGPT in 2022. As more and more site users appeared to run their writing through ChatGPT, the linguistic diversity of their writing plummeted.
Because large language models like ChatGPT are trained on the most dominant language patterns, they erase demographic and personality differences—our unique “psychological signature”—from our writing, says Sourati. This can make our conversations, including our negotiations, blander, less creative, and less productive.
In addition, bots can make mistakes they weren’t programmed to avoid. During MIT’s AI Negotiation Competition, for example, bots committed the negotiation no-no of revealing their bottom line when asked for it, reported Curhan.
At the conclusion of the AI Negotiation Summit, Harvard Business School professor James Sebenius highlighted recurring themes across presentations, including the “serious risks of replacing people versus the immense potential to help people.” Fortunately for us humans, researchers and instructors alike are intently focused on identifying how AI can best support—rather than supplant—us as negotiators, including by taking over routine tasks and improving our skills.
What topics related to AI negotiation do you think researchers should explore?
Claim your FREE copy: Negotiation Skills
Build powerful negotiation skills and become a better dealmaker and leader. Download our FREE special report, Negotiation Skills: Negotiation Strategies and Negotiation Techniques to Help You Become a Better Negotiator, from the Program on Negotiation at Harvard Law School.
