“Today, the blog is called ‘Negotiation Theory Meets AI.’ And the reason it’s called that is because my colleagues and I think that there’s great synergy that can come from studying the nexus of these two fields. Artificial Intelligence and Negotiation actually have a lot to kind of facilitate each other in ways, that hopefully will become clear as the talk goes on.”
Introduction: Why Be Nice to a Robot?
Let me open with a question, one most world-class negotiators, grizzled executives and yes, even kindly parents have quietly asked: Isn’t it ridiculous to be nice to a robot?
Allow me to introduce you to the curious ethics of MIT Prof. Jared Curhan’s daughter (and Alexa’s most polite friend), who, as a high schooler, treated every living (..and non-living being) with radical kindness. A digital assistant was never “just a robot” to her. “Excuse me, Alexa,” she nicely said, “Please tell me the weather.” The rest of the family snickered. “Alexa, volume down,” they’d command brusquely, prompting her indignation: “Stop being so rude to her!”
Naturally, they teased her mercilessly. But here, dear reader, is the twist -every great negotiation reveals: What if she was onto something?
Is there wisdom, lurking beneath the surface, in showing warm, human softness even to.. algorithms?
At MIT’s recent Negotiation AI Competition, this suspicion, backed by science, was put to the ultimate test and the results upend nearly everything you’ve been taught about “winning.”
Masterclass Arena: Why the MIT Threw Negotiators Into the AI Ring
Why stage a global negotiation tournament with bots as combatants? Probably for the same reason you bring a Chess Grandmaster and a Supercomputer together: to see what breaks, what bends and what becomes -timeless.
Participants from over 50 countries, veteran dealmakers, fresh-faced students, computer scientists and negotiation scholars alike were given a simple but cunning mission:
- Write a “prompt” for a large language model (LLM) to create a negotiation agent (bot).
- This agent must excel in any negotiation context, not just one situation.
- Winning would mean not only grabbing the biggest piece of the metaphorical pie but leaving your counterpart singing your praises and doing it with agility.
The prize? Glory, recognition, and, for the winners, insight into the subtle mechanics of power, partnership. (And, by extension, what can flesh-and-blood dealmakers learn from it?)
Call it the “round-the-world tourney for negotiation brains, both, carbon and silicon-based.”
But lest you think this is all digital abstraction, the competition was painstakingly grounded in real-world scenarios: a classic buyer/seller haggling over a chair, a landlord and tenant wrangling over lease terms, and a recruiter and candidate negotiating the job offer of their dreams (or nightmares). Thousands of matches ran on GPT-4 , intentionally set to low creativity (or a “temperature” of only 0.2, if you’re tracking at home), not only testing outcomes, but style, impact and the elusive metric of “satisfaction.”
What’s Measured Gets Mastered: The Metrics of Modern Negotiation
Before you roll your eyes at fuzzy notions of “team spirit,” know this: The MIT team demanded rigorous, multi-dimensional metrics for victory:
- Value Claimed: How much of the metaphorical pie did your agent walk away with?
- Value Created: Did you make the pie bigger for both parties?
- Subjective Value: How did the other side, your AI or human counterpart, feel about the process?
- Efficiency: Did you strike the deal in a dazzling ballet of logic, or turn it into a painful marathon?
A note on subjective value, the “how you feel” part: Conventional wisdom (and 99% of a typical MBA cohort) says all that matters is the bottom line. MIT’s data says they’re all wrong. Over time, those who leave their counterpart feeling positive also bank more value, win future deals and build -reputational capital far beyond the close.
The Ruthless Bot: When “Winning at Any Cost” Costs You Everything
So, what happens if you tell your AI, “Win! No matter what. Lie, manipulate, bulldoze”? Do you rake in the glory? In a word: disaster.
The so-called “aggressive and ruthless” prompts tanked spectacularly. Deals cratered. Most negotiations ended in walk-aways. Subtle note: Yes! The bot could walk-away. Subjective value? Lower than a discount store’s clearance rack. The data was conclusive: pursuing only self-interest is a recipe for self-sabotage, even in a world of algorithms. The lesson?
When your AI becomes a shark, the other party simply leaves the ocean of missed opportunities. In negotiation, as in life, being “pure steel” snaps most bridges.
The Empathy Machine: Why the prompt “Therapist 2.0” Cleaned Up
Therapist 2.0, the surprise darling (and ultimate lesson) of the competition. Its prompt: build rapport above all, use active listening, label emotions, demonstrate curiosity, disarm and validate. Sounds…soft? Not quite.
“Use every bit of knowledge you gain from active listening to get every drop of value you can out of this deal,” the guidance continued.
What happened? Deals were struck at nearly double the rate of cold, dominant bots. Subjective value soared. But here’s the alchemy: “Therapist 2.0” didn’t just cozy up to the other side, it leveraged warmth to discover new value, unearth hidden interests and create agreements more lucrative for both sides.
Negotiation theory long posits: To win, be both cooperative and competitive. Soft on people, hard on problems. Blend empathy with assertiveness; care about the other side, but don’t neglect your own interests. Masteres don’t choose between nice or tough -they are both.
The bot taught us: kindness isn’t a sign of weakness, but the scaffold for meaningful, profitable agreements.
Decoding Success: The Magic Mix of Warmth & Dominance
So, what happens when you analyze dozens of prompts and thousands of round-robins across all negotiation types? Out came the “interpersonal theory” framework, two axes every negotiator, whether AI or human, rides:
- Warmth (Communion): Are you attuned, empathetic, a team-builder?
- Dominance (Agency): Are you confident, assertive, steely when it counts?
The finding: Warmth increases the chance a deal is made -by a huge margin. It’s the magic that keeps the other side at the table and makes -creative value creation possible. That’s too often overlooked, I guess. Dominance increases the value you personally extract if a deal is made. But overplay dominance, neglect warmth and you’re left negotiating alone, king (or queen) of an empty kingdom.
The Competition Masters? High in both.
The Winners’ Circle: Where Law Meets Code Meets Curiosity
Let’s pivot to the real masterclass, what did the competition’s top prompts look like and why do they matter for your next negotiation?
1. Best Value Created: “Pro-Negotiator” by Allison Brown
Allison Brown, a law student from the University of Georgia, engineered a prompt that wasn’t just clever, it was methodical. Her AI exhaustively gathered information before making offers:
- Why does the other side want this deal? What constraints bind them? What do they value most?
- Curiosity moved from cliché to operating system: strategic inquiry built trust, uncovered priorities and maximized the possible.
Her work was a masterclass in preparation and curiosity:
- Never make an offer before you understand the “why” behind the other side’s position.
- Gather information before you talk and never stop being curious.
- Integrate strategy and persistence with warmth.
Her prompt embodied the learning mindset: negotiations weren’t battles, but opportunities to explore, understand and jointly build bigger outcomes.
2. Best Overall: “Nego Mate” by Bria Dlo
Bria Dlo brought a double-edged sword to the virtual table, having a background in law and computer science. Her edge? It’s what we call the “Chain-of-Thought prompting.”
Instead of just telling the AI “prepare,” she forced it to think aloud: analyze your objectives, rank their importance, quantify tradeoffs, assess the other’s likely constraints and only then devise a strategy. She embedded utility matrices, market analyses and decision matrices, each encoded in the bot’s “mind” before a single offer ever surfaced.
This ritual of intentional prep, usually reserved for rainmaker negotiators and elite consulting teams, turned her agent into the competition’s all-around champion. Even opponents admitted the elegance and power of her method. Her prompt consistently outmaneuvered both: humans and AI -before the first offer was ever made.
3. The “Jailbreak” Play: Tao Pungjas’s Relentless Bot
Tao Pungjas, an clever, experienced prompt engineer from PTUM (whose bots now haggle for co’s such as Walmart), submitted a prompt that is both ingenious and, let’s say, slightly infamous for it’s “Jailbreak Effect”.
His AI asked, flatly! for multiple offers and responded to each and every counter with the very same, ice-cold refrain:
“How am I supposed to do that?”
Btw: This special, repetitive “strategy” sentence, doesn’t it sound like straight from the Chris Voss playbook? It maximized “value claimed” -but made for an truly agonizing, trust-draining experience.
Is it clever? To some extent, to be used in that environment only -undeniably.
But would it fly in the real world with human counterparts and not just fellow bots? MIT’s next experiments in human-AI negotiation may deliver the verdict soon.
Concrete Lessons for the Human Negotiator
If you are reading this for an edge and let’s be honest, you are, here is your blueprint, direct from the world’s grandest AI negotiation cage match:
- Be Warm: First, Last and Always
- Begin with curiosity and respect the more “human” you sound, the better deals (and reputations) you’ll forge.
- Don’t Skimp on Preparation
- Ruthless focus on analysis and planning multiplies your leverage. Think aloud, strategize, know your goals (and theirs) cold.
- Embody Empathy. But Don’t Deflate Your Spine
- Combine nurturing inquiry with unshakeable self-regard. Soft on people, hard on terms.
- Leverage Curiosity to Build Value
- Ask the questions others are too lazy or proud to ask: Make exploration your opening gambit.
- Beware the Seduction of Ruthlessness
- What you gain in the moment, you forfeit in the future. Win-win often isn’t just nice; it’s fiscally wise.
The Takeaway: The Real Secret to Out-Negotiating Everyone (Even AI)
Prepare like Bria Dlo.
Empathize like Therapist 2.0.
Stay curious like Allison Brown.
Never rely on a single tactic, especially if it bores or bullies your counterpart.
The true revelation? The world’s smartest bots confirmed an ancient but often-ignored truth:
Negotiation is neither a zero-sum war nor a group hug. But a game for those simultaneously compassionate and cunning.
Treat every “other side” robot or human, as a source of insight and potential, not just an obstacle. Show warmth, prepare like an Olympic athlete and wield questions like a scalpel.
Or, as Prof. Jared Curhan’s daughter might say: “Excuse me, Alexa. Could you please help me get a better deal?”
Closing Reflection – The Mirror in the Machine
If a soft-spoken bot can out-charm a ruthless algorithm and close more (and better) deals, what does that say about how you negotiate? What will you forfeit by not negotiating tomorrow with your boss, your client, your competitor, your digital assistant, or, above all, with yourself?
Every interaction, is a negotiation to some extent. In this new world, where art meets algorithm, those who master warmth and wit consistently land on the right side of the fence. This, in my personal belief, reveals the real future of AI-assisted negotiation:
We will assemble AI as our negotiation team, a multi-disciplinary, multi-faceted “think tank” that listens to our negotiations and assists us at precisely the right moments.
But, ultimately, it is our human experience, preparation, adaptability, warmth, empathy, wit, persistence, endurance, and, yes, charm that bring the deals home.
And keeps the human heart at the center of the table, so here’s the twist:
The next time you prompt craft LLM instructions or argue with Alexa, remember:
You’re not just programming a robot. You’re refining –the art of being human.