Terms of Service

Version v2.0-2026-07-01 · StudentBench (https://studentbench.org)

1. What StudentBench is — a research study

StudentBench is research. It is a free research platform run by a research team studying how well AI (large-language-model) tutors teach. By creating an account and checking the research-consent box — “I consent to my data being used for research to evaluate and improve AI tutoring.” — you agree to these Terms, to the Privacy Policy, and to take part in that research. The purpose of the research is to measure and improve how effectively AI tutoring raises real test performance.

2. What participation involves

A study session is: a short set of intake questions, a timed pre-test, for most participants a tutoring session with an AI tutor (about an hour), and a timed post-test — in one continuous sitting of about 2–2.5 hours (the timers keep running once you start; there are short breaks between parts). You may work through concepts at your own pace, ask the tutor questions, request extra practice problems, and finish the tutoring session early.

Study conditions are assigned by the platform. Because this is a benchmarking study, you may be randomly assigned a condition — for example which AI model tutors you, which version of the tutor’s instructions is used, which order of matched test forms you receive, a difficulty tier based on a short screener, or (in some study designs) a session without tutoring between the two tests. These assignments are part of the experiment and are recorded with your session.

3. Who may participate (18+; minors need parental permission)

The signup research-consent checkbox is consent by an adult (18 or older) for their own participation. If you are under 18 you may create an account, but you cannot begin a study until a parent or guardian completes our parental-permission workflow (we email a parent or guardian a permission form, they sign it in StudentBench, and then you review and sign the youth assent form). The checkbox is never treated as parental permission. Questions about the minor workflow: inaara.hasmani-ctr@joinhandshake.com,trevor.khangi@joinhandshake.com.

4. Participation is voluntary; you may stop at any time

Taking part is voluntary. You may decline to participate, and you may stop at any time, for any reason, without penalty or loss of anything you are otherwise entitled to — simply close the tab, and ignore any remaining steps. (Compensation does require completing the study — see §6.) To withdraw from the research after participating, or to request deletion of your account, contact inaara.hasmani-ctr@joinhandshake.com,trevor.khangi@joinhandshake.com; note that data already de-identified or included in aggregate analyses may not be retrievable (see the Privacy Policy).

5. Risks and benefits

Risks. We expect no more than minimal risk — the kinds of effort and discomfort ordinary test practice involves: the session is long (about 2–2.5 hours), timed tests can be tiring or frustrating, and the AI tutor’s explanations are generated by a language model and may contain errors. The principal data risk is a breach of confidentiality of the information we collect; the Privacy Policy describes what we collect and how we protect it. Because this research is no more than minimal risk, no compensation or treatment for research-related injury is provided.

Benefits. You receive free AI tutoring that may improve your test skills, and you can see your pre-test results; improvement is not guaranteed. The research aims to produce generalizable knowledge about how well AI tutors teach.

Alternatives. Participation is not required for anything — the alternative is simply not to take part (for test preparation itself, many non-research alternatives exist, such as books, classes, and free practice resources).

6. Compensation and integrity review

Eligible adult (18+) participants can be paid through the Handshake platform after completing the full study. At the end you receive a completion code; pasting it into your Handshake task links your Handshake identity to your StudentBench session so we can verify completion and pay you. Eligible SAT participants under 18 can receive a $50 gift card after completion, subject to parent permission, youth assent, completion, and the integrity review.

Payment is conditioned on completion and an integrity review. To qualify you must complete the pre-test, tutoring, and post-test in full, in one sitting, giving your honest best effort. Each session is reviewed using the session records described in the Privacy Policy — including checks for signs of outside help, use of LLMs or other AI tools on the tests, switching to other tabs/windows or apps during a timed test, and implausibly fast correct answers on hard questions. Sessions are flagged for human review, and payment may be withheld if the review concludes the integrity rules were broken. Gift-card eligibility may likewise be withheld for minors if the review concludes the integrity rules were broken. The review is performed by people; flags alone never automatically disqualify you.

7. Honest-effort rules (what you agree to during tests)

During the pre-test and post-test you agree to work alone: no LLMs or other AI tools, no other people, and no outside help of any kind. You also agree not to copy, photograph, share, or republish test questions or tutoring content — the test items must stay confidential for the research to remain valid for future participants.

8. AI-generated content

The tutoring plan, explanations, practice problems, and chat replies are generated by a third-party AI model (see the Privacy Policy for the provider). AI output can be wrong, incomplete, or misleading, and we deliberately do not correct the tutor’s teaching (measuring it is the point of the study — though AI-generated practice-problem answer keys may be programmatically verified for grading accuracy). The service is provided “as is,” for research and practice — it is not professional educational advice, and no particular score outcome is promised.

9. Your account

Provide accurate information (your real name, a working email address, and your real age — age determines the consent workflow that applies to you), keep your password confidential, and use one account per person. We may suspend accounts that violate these Terms or compromise the research (for example, sharing test content). One study session is designed to be completed per sitting; retakes, where offered, are recorded as separate attempts.

10. Changes to these Terms

If we materially change these Terms or the consent language, we update the version identifier shown at the top (and recorded with your consent). Significant changes are posted on this page; continued use after a change is acceptance of the new version.

11. Contact — questions, concerns, or your rights as a participant

For any question about the study, your participation, payment, withdrawal, or your rights as a research participant, contact the research team at inaara.hasmani-ctr@joinhandshake.com,trevor.khangi@joinhandshake.com.

See also the Privacy Policy — what we collect, how it is processed, who sees it, and how long we keep it.