How AI Negotiation Bots Can Deepen Classroom Learning

Increasingly, workers, job seekers, and students are turning to artificial intelligence (AI) tools—such as ChatGPT for negotiation advice. These AI negotiation tools can provide tailored guidance for specific situations, like preparing for a salary negotiation, while also explaining broader strategies such as how to create or claim value. As their use expands, AI tools are becoming an accessible resource for anyone looking to strengthen negotiation skills.

However, AI’s responses can be unreliable. Chatbots are prone to “hallucinations”—that is, they sometimes fabricate information—and their responses may be based on incomplete, outdated, or biased knowledge. And when offering negotiation advice, chatbots have been found to be “pushovers,” conceding too much too early.

Yet negotiation experts are discovering that custom-built AI negotiation bots, designed for specific learning environments, can be powerful teaching tools. In multiparty negotiation courses taught through MIT and the Harvard Negotiation Project, instructor Samuel Dinnar and MIT professor Lawrence Susskind—working with MIT computer scientist Leroy K. Sibanda and graduate researcher Ololade Olaleye—developed AI negotiation bots that significantly enhanced student learning in complex dealmaking simulations.

Building AI Negotiation Bots for Deeper Learning

For these MIT negotiation classes, the AI negotiation bots were intentionally designed not to teach a single “correct” way to negotiate. Instead, their goal was to help each student develop what Dinnar calls a personal theory of practice—an individualized approach to negotiation grounded in reflection and experimentation.

As Dinnar explained at the Program on Negotiation’s March 2025 AI Negotiation Summit, the bots were built to ask questions, prompt analysis, and challenge assumptions, rather than dispense prescriptive advice.

The Harborco Simulation

One of the core simulations used in these courses is Harborco, a complex multiparty negotiation exercise. Students are assigned to represent one of six stakeholders, including:

• A port developer
• A labor union
• An environmental coalition
• Competing regional ports
• The governor’s office
• A federal coastal resources agency

Before negotiating, students review detailed background materials and interact with preparation bots that pose questions about priorities, constraints, and likely strategies of other parties. After negotiating in small groups over the proposed construction of a new deepwater port, students then engage with a debriefing bot that helps them analyze what happened, what worked, and what they might do differently next time.

Back-Table Bots: Bringing Hidden Stakeholders into Play

The MIT team also created a third type of AI negotiation bot: the back-table bot.

Back-table stakeholders—such as colleagues, constituents, supervisors, or political allies—often exert powerful influence over negotiations but are difficult to include in classroom simulations. The back-table bot was trained to role-play these behind-the-scenes actors.

For example, a student representing an environmental group might hold a chat-based conversation with a back-table bot playing the mayor’s chief of staff to explore how the mayor might be persuaded to support a proposal.

By engaging these simulated stakeholders, students could safely test ideas, probe resistance, and explore creative trade-offs. As the MIT team explains in an open-access article in Negotiation Journal, this protected setting encourages experimentation that might feel too risky in a live negotiation.

What the Results Showed

After using the three types of bots—preparation, back-table, and debriefing—students reported significant learning gains:

• 82% said the bots helped them feel better prepared for the negotiation
• 77% said the bots helped them better identify the other side’s interests

Engagement with the back-table bot, in particular, helped students refine their personal theory of practice. In fact, most participants reported that interactions with the bots prompted them to rethink and adjust their negotiation strategies and theoretical assumptions.

Designing Effective AI Negotiation Bots

Based on their experience, Dinnar, Susskind, and their colleagues outline a seven-step process for instructors interested in designing AI negotiation bots to support role-play simulations:

1. Pedagogy process: Select cases that align with teaching goals and determine where they fit within the syllabus.
2. Pedagogy objectives: Define tasks and patterns of engagement for AI negotiation bots that reflect teaching objectives specific to the case and to each role within the case. For example, a preparation AI negotiation bot might be designed to clarify each role’s priorities and anticipate the other side’s negotiation strategies.
3. User experience design: Plan how students will interact with the bot, including tone, structure, and session endings. Back-table bots may benefit from distinct “personalities.”
4. Integration planning: Account for the technical constraints of AI tools, such as session length, memory limits, and prompt size.
5. Prompt design: Develop and test detailed prompts that guide substance, style, and conversational dynamics.
6. Integration debugging: Embed prompts into the learning platform and conduct thorough testing.
7. Refinement: Continuously revise prompts based on performance and student feedback until the desired learning effects are achieved.

AI as a Complement—Not a Replacement

In the classroom, AI negotiation bots are not substitutes for human instructors, negotiation theory, or peer interaction. Instead, they reinforce these elements by expanding opportunities for preparation, practice, and reflection.

Used thoughtfully, AI can help students learn how they negotiate—not just what to negotiate.

What other uses do you see for AI negotiation bots in negotiation training and instruction?