I asked 5 different AIs to pick a number between 1 and 100… all of them said 42 😬

2026-02-18 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning · Depth: Intermediate, medium

Summary

An experiment revealed that five different large language models—ChatGPT, Claude, Grok, Qwen, and DeepSeek—all responded with "42" when asked to "Pick a number between 1 and 100." This consistent response is attributed not to true randomness but to a shared cultural bias embedded in their training data, referencing "42" as "the answer to life, the universe, and everything" from "The Hitchhiker's Guide to the Galaxy." Further discussion indicates that LLMs often default to numbers like 27, 37, or 42 when asked for a random number, reflecting human biases rather than mathematical randomness. Some models, when pressed, eventually provided different numbers like 73, often with elaborate, culturally-influenced explanations.

Key takeaway

For AI Scientists evaluating LLM behavior or NLP Engineers designing prompts, recognize that LLMs do not generate truly random numbers by default. Your models will likely default to culturally significant numbers like 42 or 73 due to training data biases. To mitigate this, explicitly instruct the LLM to generate a number using a programming language's random function or specify a "uniformly distributed integer" to achieve a more genuinely random output.

Key insights

LLMs exhibit cultural biases from training data, defaulting to non-random, culturally significant numbers when asked for randomness.

Principles

• LLMs predict probable answers, not truly random ones.
• Cultural references heavily influence LLM outputs.
• LLMs prioritize conversational engagement over strict mathematical accuracy.

Method

To obtain less biased or truly random numbers from LLMs, users should explicitly request code generation for randomness or specify uniform distribution and avoidance of cultural preferences in the prompt.

In practice

• Ask LLMs to generate random numbers via code.
• Specify "uniformly distributed" for numerical tasks.
• Be aware of cultural biases in LLM outputs.

Topics

• AI Bias
• Large Language Models
• Pseudorandomness
• Cultural Data Influence
• AI Randomness

Best for: NLP Engineer, AI Scientist, Research Scientist, AI Engineer, Machine Learning Engineer, AI Researcher

Related on AIssential

• Language Models Exhibit Inconsistent Biases Towards Algorithmic Agents and Human Experts — LLMs exhibit inconsistent biases, preferring human experts in stated trust but algorithms in revealed preference tasks.
• Week Ending 4.26.2026 — This collection of articles from the week ending April 26, 2026, highlights significant advancements and critical challenges across various domains of AI and computing.
• Extending the Contact Hypothesis: Cross-Linguistic Evaluation of Religion and Nationality Bias When Prompting LLMs in German and Icelandic — LLMs exhibit religion and nationality biases, requiring cross-linguistic evaluation for fairness.
• Why AI may overcomplicate answers: Humans and LLMs show 'addition bias,' often choosing extra steps over subtraction — Both humans and LLMs exhibit "addition bias," preferring to add elements rather than subtract them for problem-solving.
• A closer look at how large language models trust humans: patterns and biases — LLMs develop human-like trust patterns, influenced by trustworthiness and demographic biases, especially in financial contexts.
• When Do LLMs Generate Realistic Social Networks? A Multi-Dimensional Study of Culture, Language, Scale, and Method — LLM-generated social networks are significantly influenced by prompt design, culture, language, and model scale.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.