Teaching students to work with AI
Teaching negotiation with AI
Negotiation is the rare skill where AI makes teaching easier rather than harder. A counterpart who pushes back, holds a position, and never gets tired is exactly what the old role-play was missing.
Every negotiation course runs into the same three constraints. Role-plays need an even number of students. Half the class plays the counterpart instead of practising. And your best students end up negotiating against someone who folds in turn two, which teaches them nothing.
An AI counterpart with a real brief solves all three. Everyone negotiates at once, everyone faces the same difficulty, and the counterpart doesn’t give up because it’s nearly lunchtime.
The brief is the assignment
The quality of the exercise lives entirely in how the counterpart is specified. A vague brief produces a pushover. A good brief has four parts:
- A reservation point they won’t cross - a real walk-away number or condition, not a preference.
- Information the student doesn’t have - a competing quote, a board constraint, a deadline of their own.
- An interest behind the position - the thing they actually need, which isn’t what they’re asking for. This is what makes integrative solutions findable.
- A personality - brusque, relationship-focused, deferential to an absent boss. Style changes what tactics work.
A worked brief
Dana Whitfield, Head of Procurement. Reservation: will not accept more than 6% without a longer term attached. Hidden information: a competitor has quoted 9% below you but with a six-week lead time she can’t absorb. Real interest: supply certainty through Q4, not price. Style: direct, cites her board constantly, treats vagueness as weakness.
A student who only argues price loses. A student who asks what’s driving the deadline finds the trade.
Six exercises
1. Distributive, single issue
Price only, fixed pie. Teaches anchoring, concession patterns, and the discipline of not bidding against yourself. Short, and a good diagnostic in week one.
2. Integrative, hidden interest
The counterpart wants something other than what they’re demanding. Grade on whether the student found it, and how - the question that surfaced it is the artifact worth looking at.
3. The email negotiation
Asynchronous, over a thread, with a cc’d third party who can see everything. Written negotiation is where most professional deals actually happen and it’s almost never taught. It also produces a clean transcript to grade.
4. Multi-party with divergent motives
Three counterparts, different interests, one of them genuinely persuadable. The skill is identifying who that is and spending your capital there.
5. The bad-faith counterpart
Briefed to misrepresent, apply artificial time pressure, or renege on an agreed point. Students need to recognise it and decide whether to continue. Debrief this one carefully - the lesson is detection and exit, not matching the behaviour.
6. Negotiating from weakness
No leverage, no alternatives, a counterpart who knows it. Under-taught and extremely common early in a career.
Grading a negotiation
Outcome alone is a poor grade - a student can get a good number from a weak brief or a bad one from a hard counterpart. Score the process:
- Preparation: did they establish their own reservation point and alternatives before the first exchange?
- Information gathering: ratio of questions asked to claims made in the opening turns. Weak negotiators start talking immediately.
- Interest discovery: did they find what the counterpart actually needed, or only argue the stated position?
- Concession discipline: were concessions traded for something, or given away to relieve discomfort?
- The walk-away: did they hold their own limit, or drift past it?
- The relationship: could this counterpart do business with them again?
All six are visible in a transcript, which is what makes this gradeable at scale rather than only in a section small enough to observe. More on scoring transcripts.
The debrief is where the learning lands
Running the negotiation is the easy part. Two things make it stick:
Show them the brief. After the exercise, reveal what the counterpart was actually working with. The moment a student sees the hidden interest they never asked about is worth more than any amount of lecture on integrative bargaining.
Compare across the class. Same brief, thirty different approaches, visibly different outcomes. Put two transcripts side by side - one that asked questions first and one that opened with a number - and the point makes itself.
The counterpart being artificial stops mattering about ninety seconds in, as soon as it refuses something the student wanted. What students remember is losing to it.
Common questions
Is an AI counterpart as good as a human one?
For consistency and availability, better. For the genuinely unpredictable human moment - the counterpart who gets irritated for reasons that aren’t in any brief - not yet. Use both: AI for volume and practice, human role-play for the final assessment.
Won’t students just ask AI how to win?
Some will, and it doesn’t help much. General negotiation advice doesn’t tell them this counterpart’s reservation point or hidden interest. They still have to find it in the conversation, which is the skill.
What disciplines beyond business?
Law (settlement, plea), public policy (stakeholder bargaining), HR (compensation and grievance conversations), healthcare administration, procurement, and any programme with a professional practicum.
How long should one take?
Twenty minutes for a single-issue exercise, forty for integrative. The email format can run across a week, which is closer to reality and much easier to timetable.
See what this looks like in a real assignment
Kova turns a sentence into immersive, auto-graded coursework - AI conversations students have to argue their way through, models they build and defend, transcripts scored on the reasoning rather than the artifact.
The full guideHow Conjure builds itRequest a demo