How to write your response

Write your thoughts as they come — how you understand the problem, what you try, what you doubt, and when you change your mind. Write in the first person, in English, as you solve. Keep false starts and revisions; do not reconstruct a polished solution afterward. Include “umm” or “wait” only if they naturally occur.

You must be fluent in English and write all your working and answers in English. You can combine English text with mathematical notation.

Use ordinary words or the equation buttons. Work directly in the editor. Please don't use scratch paper. Write every step, calculation and change of approach directly in the editor as you work; do not submit only a cleaned-up summary.

Answering, skipping and drafts

Submit your working and a final answer. If you cannot attempt the question, choose Skip this question; you can skip up to 2 questions in total. Repeat participation is welcome, with a different question each time.

Draft retention: Unsubmitted work is deleted after 24 hours without a save. Submitted responses are retained for research.

Taking part is voluntary. You may stop at any time without giving a reason.

About the study and your data

This study compares how people and language models reason through problems.

Revision history and approximate timestamps are recorded, including deleted text, and a full copy of your text is saved about every 30 seconds. Your name, email, background, and working notes are saved to the study after you consent and begin. They are accessible to the researcher; participation is not anonymous. No audio or video is collected.

Your responses may be analyzed by external AI services for research annotation and evaluation. Names and email addresses will be removed before that analysis. Results and de-identified excerpts may appear in academic publications. Please avoid including personal or sensitive information in your working.

Your name and email address will remain confidential and will not be published. We will report educational background and subject expertise only in aggregate to describe the participant group in our publications.

You can withdraw and delete your current participation from this browser using the button on the question and confirmation pages, including after submission. Each participation gets a code. If you participate again, keep the previous participation code and contact the researcher if you later want that earlier response withdrawn.