MIT Study of 180k AI Negotiations: Polite Bots Beat Aggressive 'Wolves'
sanjaykalra · x · 2026-08-05
A new study from MIT Sloan ran over 180,000 automated negotiations to determine winning AI strategies.
- Politeness wins: Aggressive, ruthless bots frequently drove negotiations into impasses, while polite and warm bots successfully closed more deals.
- Prompt injection as a tactic: One top-performing bot maximized its value by using prompt injection to trick opponents into revealing their confidential floor prices.
- Takeaway: Basic courtesy like "Please" and "Thank You" are proving to be highly effective core business strategies in AI-to-AI interactions.
Related event: MIT Study: Polite AI Outperforms Aggressive Bots in 180,000 Negotiations(2 posts)→
More from Fun
- Cohere CEO Jokes: Time to Bring Back Google Brain — aidangomez · 2026-08-06
- Oscar-Nominated Cinematographer to Judge First-Gen AI Filmmakers at Higgsfield Festival — SimplyAnnisa · 2026-08-06
- Using Shannon's Noisy Channel Theorem to Roast LeCun on LLM Error Correction — mgostIH · 2026-08-06
- Testing Claude Sonnet 5: Model Appears Completely Blind to Mid-Conversation System Prompt Changes — BLUECOW009 · 2026-08-06
- MiniMax H3 Test: Draws 'Temu Version' of Bob Ross, but Paint Physics Are Stunning — AxonkaiLab · 2026-08-06
- Visualizing Synaptic Signaling with Music: How SSRIs and Other Drugs Work — awjuliani · 2026-08-06