Common Sense Test

What is a common sense test?

A common sense test checks the everyday reasoning that most adults do without thinking: knowing that an ice cube melts in hot tea, that a glass dropped on a tiled floor will probably shatter, that someone who is crying is unlikely to be pleased, and that you should not put a metal pan in a microwave. None of this is taught in a classroom. It comes from years of living in a physical, social world, which is exactly why it is so hard to write down and so interesting to test.

This free test has 16 multiple-choice questions and a 10-minute limit. It reports a percentage score with a breakdown across four categories, and because the same questions are put to AI models through the site's API, you can see how your intuition compares with theirs on the human-versus-AI leaderboard.

The four categories

• Everyday physics — what happens to ordinary objects when they are dropped, heated, tilted, floated or squeezed. No formulas, only expectations built from experience.
• Cause and effect — which outcome follows from an everyday action, and which explanation for something you observe is the sensible one.
• Social sense — reading ordinary human situations: what a reaction means, what is polite, what a reasonable person would do next.
• Practical reasoning — the small judgement calls of daily life, such as what to check first, what is safe, and what is worth the effort.

How the test is scored

Each question has one correct option. Your headline score is the percentage answered correctly out of 16. The results screen also shows your score within each category, so a low total can be traced to a single weak area rather than a general one. The order of the answer options is shuffled on every attempt, so there is no fixed pattern to learn, and the categories are interleaved so you never get four physics questions in a row.

After the test you will also see how your percentage compares with other players on this site. That figure is a site percentile, not a population norm: this is an original test written for interest and self-reflection, not a validated psychometric instrument.

Why people beat AI at common sense

Large language models learn from text. Text is full of things people bother to write down and almost empty of the things everyone already knows. Nobody writes "water is wet" or "you cannot fold a brick", so a model has to infer those facts indirectly, and it sometimes infers them wrongly. People, by contrast, have handled bricks and water. The questions in this test are chosen to sit in that gap: obvious to most humans, occasionally baffling to a machine that has only ever read about the world.

That gap has been narrowing. Newer models are trained specifically on common-sense benchmarks, and on some questions they now match or beat the typical human score. The leaderboard makes this visible over time, which is why the test is part of the AI vs Human Arena alongside the Lateral Thinking Test, the Token Trap Test and the Visual Reasoning Test.

Tips for a fair attempt

• Go with your first instinct. The questions are designed so that the intuitive answer is the correct one; over-thinking tends to lower scores.
• Picture the situation. Most everyday-physics items are easier to answer if you imagine actually doing the thing described.
• Do not look anything up. The point is to measure what you already know, and the 10-minute limit is generous enough for a careful read of every option.