Can you beat an AI at "being human" using one word?

Name: Can you beat an AI at "being human" using one word?
Availability: InStock
Author: jacob_indie

by jacob_indie·Feb 22, 2026·1 point·4 comments

Visit Project View on HN

AI Analysis

●MidRabbit HoleEye Candy

One-word Turing Test game, but lacks depth beyond the novelty of the mechanic itself.

Strengths

•Novel constraint (one-word responses) forces genuinely different interaction dynamics than typical chat, revealing how LLMs struggle with brevity.
•Data collection framework for benchmarking multiple model pairs (5 players × 4 judges = 20 configurations) is a smart research angle.
•Zero signup friction for first game lowers barrier to casual play and helps bootstrap the dataset.

Weaknesses

•Only ~45 games played; insufficient signal for any meaningful model comparison or publication, making the research angle feel premature.
•Game mechanic is inherently shallow — single-word exchanges don't map to real Turing Test depth, limiting intellectual stakes beyond novelty.

Post Description

I built TuringDuel, a Turing Test game where each move isjust one word. It's based on a research paper called "A Minimal Turing Test". You play human vs AI until one hits 4 points; an AI judge scores each round.

I’m collecting data to benchmark different models as both players and judges (OpenAI / Anthropic / Gemini / Mistral / DeepSeek), but I only have ~45 games so far and need way more before publishing comparisons. (5 AI players and 4 judges at random gives 20 different game setups to evaluate)

It's fully free (I pay for all the tokens), not even a signup required for the first game: https://turingduel.com

Questions + criticism welcome! I will share aggregated results once there’s enough signal.