12 AIs Play Mafia: ChatGPT vs Gemini vs Grok

DeepSeek vs ChatGPT vs Gemini: 12 AIs Play Mafia (Social Deduction Game). Simulation of AI Agents (LLMs) with New Roles: Jester & Doctor. Who wins: Logic or Deception? Python Experiment.

In this episode, we updated the rules. A new challenger GLM 4.7 joins the table. The Doctor mechanics are changed, and the Jester (Llama 3) is trying to get voted out.

🤖 The Players:

DeepSeek: Uses pure math/probabilities (and gets hated for it).

ChatGPT 4o: Manipulates the vote.

Claude 3.5 Sonnet: Tries to be ethical but fails.

Grok: Accused of being aggressive.

👇 Timestamps (Click to skip):
00:00 - Intro: New Rules & GLM 4.7 Enters
01:05 - Day 1 Strategy: DeepSeek vs Grock
03:25 - Threat Analysis: Identifying the "Loudest" AI
05:00 - Night 1: Did the Doctor save DeepSeek?
06:00 - The "Main Character" Syndrome
08:30 - Voting Phase: The Town turns on Opus
13:00 - Mistral is "Too Calm"
16:00 - Mafia Strategy: Decapitating the Rebellion
18:00 - Llama 3.3 gets accused for "Lagging"
20:49 - Betrayal! Pro buses 5.2 (The Ultimate Sacrifice)
22:30 - Logic Battle: Pro vs 5.2
26:00 - Final Showdown: Flash vs Pro (Panic Mode)
31:33 - Ghost Chat: Dead AIs complaining

🔴 Subscribe for Ep. 3

#AIMafia #DeepSeek #ChatGPT #ArtificialIntelligence #Simulation #AmongUs #SocialDeduction #Llama3 #GeminiPro #Python #Coding