Participant information and consent

Introduction

We are asking you to participate in a research study titled "Futures of Maths". The study is being led by Jon Roozenbeek, at the Department of Communication Science, Vrije Universiteit Amsterdam.

Who is involved

• Principal investigator: Jon Roozenbeek (VU Amsterdam and University of Cambridge).
• Research team: Francesco Fournier-Facio (Heriot-Watt University), Yara Kyrychenko (University of Cambridge) and Julian Wykowski (KIT). The research team wrote the launch statements and analyses the responses.
• Content moderators: mostly members of the research team. They check every statement that participants post before it appears in the feed, and may lightly edit its wording. They see the statement but never who wrote it, and they have no part in verification.
• Verification moderator: someone independent of the research team. They check manual review applications from people the automatic ORCID check misses, and nothing else. They never see consultation responses, and the anonymous pass means they cannot link anyone to their answers.

1. What the study is about

The purpose of this research is to learn about mathematicians' attitudes towards the use of artificial intelligence (AI) tools, including AI agents, in mathematical research and teaching, and towards how such use should be governed.

2. What we will ask you to do

First, you confirm that you are eligible: you need at least one article in a mathematics journal (a journal indexed by zbMATH Open) published in the last three years. You confirm this either by signing in with ORCID, or, if that does not work for you, by asking the independent verification moderator to check your eligibility. For manual review, an accepted article also counts, and you show the verification moderator both the article and that you are its author; this may take a few days or longer. You then receive an anonymous link to the consultation. In the consultation we will ask you to:

• scroll through a feed of short statements about AI in mathematics, written by the research team and by other participants, and rate each from "strongly disagree" to "strongly agree", or skip it. A few statements that the research team would like everyone to answer are pinned at the top;
• answer select-all questions in the feed, ticking every option that applies (or "None of these"), or skip them;
• answer a few ranking questions in the feed, putting options in order (for example, declarations on AI in mathematics by how closely they match your views), or skip them;
• tick any statement you consider to be of exceptionally high importance;
• if you wish, post your own statements for others to rate. A content moderator checks every statement before it appears;
• if you wish, answer a few optional questions about yourself: where you are based (a broad world region, or the country for some countries such as the United States), academic position, type of institution, department and research fields.

Once you answer a statement, you see how everyone has voted on it so far, as soon as at least 5 people have answered it. The order of the statements in your feed is chosen partly at random, favouring statements that have had few answers or on which opinions differ; it never depends on your own answers. You can take as little or as much time as you like answering statements. You can stop at any time and return later with your link until the consultation closes.

On the "My session" page you can see your karma points, earned by answering statements (skipping counts the same as rating), answering pinned statements and posting statements that are published. Karma is only a light-hearted way to show your contribution: no other participant can see it, there is no leaderboard, and it has no value outside the consultation. The research team sees it next to your participant code in its data files, to study how it relates to participation.

3. Risks and discomforts

We do not anticipate that you will feel uncomfortable during this consultation, nor do we anticipate risks that are uncommon in daily life. The main risk is that someone could recognise you from what you write or from a rare combination of answers about yourself. To reduce this risk, please do not include names or other identifying details in statements or "Other" answers, and skip any question about yourself if you think your answers could make you identifiable. Section 7 describes how we protect you.

4. Benefits

There are no direct benefits to you from taking part in this research. We cannot promise any benefits to others from your participation. However, we anticipate that this study will create knowledge about the use of AI in mathematical research and teaching that can inform the mathematical community, universities, journals and funders.

5. Compensation for participation

You will not be financially compensated for this research. The karma points shown in the consultation are not a reward and cannot be exchanged for anything.

6. What will happen to your responses

1. While the consultation is open, your ratings, importance ticks, answers to select-all questions, rankings, the statements you write and your optional answers about yourself are stored in the consultation database, on servers of Scaleway SAS in Amsterdam, the Netherlands. They are stored under a random code, not under your name, email address or ORCID iD, which the consultation database never receives.
2. Statements you write are read by the content moderators (mostly members of the research team), who may lightly edit them for clarity or to remove identifying details, and are then shown, without your name or code, to other participants to rate. In the research files, each statement you write is marked with your random participant code, so that the research team can study how authors rate statements, including their own. Statements may be quoted in publications and included in the published dataset.
3. What other participants see. After answering a statement, participants see how everyone has rated it so far, as percentages, and how many people have answered it. These summaries appear only once at least 5 people have answered the statement (and at least 5 have rated it, for the percentages), so that no one's answer can be read from them. The same applies to select-all questions, where participants see the share of people who ticked each option, and to ranking questions, where they see the combined ranking of everyone, in both cases only once at least 5 people have answered. Nobody sees another participant's individual ratings or rankings, and participants never see who wrote a statement.
4. The research team (Francesco Fournier-Facio, Yara Kyrychenko and Julian Wykowski) downloads the responses as spreadsheets for analysis and keeps them on password-protected, encrypted computers and university storage. In these files each participant is identified by a random participant code, which is not your name, email address or ORCID iD and cannot be traced back to them. The same code marks the statements you wrote, and the files also include your karma points. Answers about yourself are included only where at least 5 participants share the same combination of answers; otherwise they are replaced with "suppressed". Text you type under "Other" is not included; the research team may give it a short general category (for example "Philosophy"), and only that category is included.
5. Analysis. We calculate how much participants agree with each statement and how important they find it, combine participants' rankings with standard voting methods (such as the Schulze and Borda methods), and we use statistics to find groups of participants who rate statements in similar ways and the statements these groups agree or disagree on. This is done on the whole dataset; no decision is made about any individual.
6. Publication. Results are published in academic articles, reports and presentations as summaries, such as percentages and averages. A de-identified dataset (ratings, importance ticks, select-all answers, rankings, statements and the protected answers about participants) may be published openly on the Open Science Framework (OSF) or DataverseNL, so that other researchers can check and reuse it. The published dataset does not show which participant wrote which statement, and does not include karma points, from which this could be worked out.
7. After the consultation closes, the online databases are deleted within 30 days of the end of the consultation, once the research data has been copied to secure VU Amsterdam storage. The de-identified research data is kept for at least 10 years after the last publication based on it, in line with VU Amsterdam research data policy.
8. We will not sell your responses, use them for any purpose other than this research, or try to find out who you are.

7. Privacy, confidentiality and data security

The legal entity responsible for processing your personal data is Stichting VU. The data processed in this study are described in full in the privacy statement. In short:

• Verification. If you sign in with ORCID, we read your public ORCID record to check that it lists at least one article in a mathematics journal, published in the last three years and added by Crossref. We do not store your ORCID iD, your name or your record. If you ask for a manual review, your email address, the details of your article, your message and any proof you send (for example a forwarded acceptance email) go to the moderator, the verification moderator, who is not part of the research team and may be based in the United Kingdom, and are not stored on the website. For both methods we keep only a coded, one-way fingerprint (a keyed hash) of your ORCID iD or email address, to make sure each person takes part once. It is kept separately from the consultation, cannot be linked to your responses, and is deleted within 30 days of the end of the consultation.
• Your responses cannot be linked to you. The anonymous consultation link is created with a cryptographic technique called a blind signature: the verification website confirms you are eligible without ever seeing your link, so nobody, including the research team and the verification moderator, can connect your responses to your identity. We do not record IP addresses, and we store dates only to the month.
• Personal data in your responses. Your views on how AI should be governed may reveal opinions that some regard as political, and your answers about yourself are demographic data. These are processed exclusively for scientific research, specifically social science research on attitudes to AI in mathematics.
• Who has access. Only the research team, Francesco Fournier-Facio (Heriot-Watt University), Yara Kyrychenko (University of Cambridge) and Julian Wykowski (KIT), and the principal investigator, Jon Roozenbeek (VU Amsterdam and University of Cambridge), have access to the research data. Content moderators see participants' statements before they are published, but not who wrote them. The verification moderator never sees research data. The websites are hosted in the European Union, by Scaleway SAS in Amsterdam, and all connections are encrypted. Members of the research team and the verification moderator may access the data from the United Kingdom, which the European Union recognises as protecting personal data to an equivalent standard (see the privacy statement).
• Consent. By ticking the consent box at the bottom of this page, you consent to the processing of your research data as described here and in the privacy statement.

8. Data availability

De-identified data from this study may be shared through the Open Science Framework (OSF) or DataverseNL with the research community at large to advance science. We will remove or code any personal information that could identify you before files are shared with other researchers, to ensure that no one will be able to identify you from the information we share by current scientific standards and known methods. Despite these measures, we cannot guarantee anonymity of your personal data.

9. Taking part is voluntary

Your participation in this study is entirely voluntary. You may refuse to take part, stop at any time, and skip any statement or question, for any reason and without any penalty or loss of benefits to which you are otherwise entitled. Only the consent box is required.

Withdrawing your consent. While you still have your consultation link, you can withdraw at any time by opening "My session" and choosing "Withdraw my answers". This permanently deletes your ratings, select-all answers, rankings and answers about yourself, and removes your participant code from the statements you wrote. Those statements stay in the study, with nothing linking them to you. Because your responses are anonymous, once you no longer have your link nobody, including the research team, can find your responses, so we cannot delete them on request. To withdraw consent for your verification data, email futuresofmaths@tutamail.com.

10. Ethics review and complaints

This study complies with the ethical guidelines of the Faculty of Social Sciences at Vrije Universiteit Amsterdam. It has undergone an Ethics Self Check (number 2026-10-5-484) and has been approved as a low-risk study that does not pose undue risks to participants.

If you have a complaint, please contact the main researcher conducting this study first (Jon Roozenbeek, futuresofmaths@tutamail.com). If this does not lead to an acceptable outcome, you can write to the vice-dean for research of the Faculty of Social Sciences via the Research Support Office at onderzoeksbureau.fsw@vu.nl.

11. If you have questions

The main researcher conducting this study is Jon Roozenbeek, an associate professor at Vrije Universiteit Amsterdam. If you have questions, before or after taking part, you may contact Jon Roozenbeek at futuresofmaths@tutamail.com. Questions about your personal data can also be sent to the VU Amsterdam Data Protection Officer at functionarisgegevensbescherming@vu.nl.

12. Statement of consent

You give your consent by ticking the box below the information on the consultation website before you start. We record only that you consented and the month, under your anonymous code; we do not collect a signature or your name.