Proofs and Prompts

Notes from the common room

Informal questionnaire for researchers

Mark Hagen, Associate Professor at University of Bristol

Take part Short survey (6 questions) or long survey (14 questions)

The 100+ reactions post, and the ensuing discussion, made me think it would be helpful for the community to have a rough qualitative snapshot of not only how we’re feeling and what we’re thinking, but also how that depends on some material factors that intuitively seem relevant. I therefore made a questionnaire aimed at mathematics researchers (broadly construed).

This isn’t intended as a rigorous survey, and definitely won’t be. The point is, rather, that it can be easier to formulate and express one’s thoughts about something large and confusing by answering specific questions, even if they just serve to stimulate thinking or structure pre-existing thoughts. The survey responses are free-response and public, and designed to create a structured discussion to which people with significant stakes in the AI/maths research developments can contribute, and from which we can learn about what other people actually think, and why they think it, in some detail. Crucially, some of the questions are about where one is situated in the research community and in the economy, since it would be good to surface the influences of material factors on how we’re approaching some quite significant (for us) questions.

I am posting answers here as I receive them. The answers so far are thoughtful and informative, and I highly recommend reading them.

I would like to have as many responses as possible, so please share this survey with others. You can choose between a long version of the survey (14 questions) and a short version (6 questions). They are free-form questions, intended as a vehicle for expressing your thoughts. I think you’d be adding something important by taking one of the surveys.

There is a bit more information here. Without further ado:

Take the long survey.

Take the short survey.

Responses.

I am updating the responses page periodically; it may take a day or two for yours to appear there.

You can specify whether you want your name on the responses I post, or if they should be anonymous. In either case, the form does ask for your name and a small piece of information (an institutional email or a website URL, for instance) so I can more easily verify your identity. This identifying information will not be posted (except your name if you so specify) and will not be retained after I’ve posted your answers.

I will post your responses, verbatim, unless I am unable to verify that you’re involved in mathematics research in some way that makes the AI developments significant for you; I may also reject submissions that are obviously spam or made in bad faith. Thus far, I have not rejected any responses.

Received 20 October 2026.

Filed under ·
·
More by →

New to Proofs and Prompts? Start here →

10 responses to “Informal questionnaire for researchers”

  1. Anonymous Avatar
    Anonymous

    I think it would be more useful to have an anonymous multiple-choice survey, so that we could get actual statistics on opinions in the community rather than read through individual essays about people’s experiences and views. Part of the uncertainty we’re facing comes precisely from not knowing what people in our community think about AI and its use in research *on a larger scale*. These broader trends will shape what happens in the future.

    I’d be particularly interested in seeing charts showing how opinions on AI differ depending on things like career stage, employment status, homeownership, gender, or country of residence etc.

    Of course, there would need to be some way of checking that the respondents actually belong to the community, rather than being open to random people on the internet.

    Reply
    1. Mark Hagen Avatar
      Mark Hagen

      My understanding is that other people, with the assistance of actual social scientists, are going to do a survey more along the lines you’re suggesting, which is also very valuable and can and should exist in parallel to an effort like this one.

      This is not intended to be that. The point of this exercise is to allow us to read what others think, in detail, while knowing something about who’s talking and what their stake is in the situation. The other point is to help members of the community collect their thoughts (by asking specific questions) and display them in a place where others can read them. So, it’s an attempt to have a more structured collective discussion, rather than a scientific survey (which, again, I hope will soon exist in parallel to this).

      Reply
      1. Anonymous Avatar
        Anonymous

        Sorry, I didn’t mean to criticize your survey. It just made me think about how I’d like to see some real data, because right now I’m confused about what people actually think, beyond the few who are vocal about their opinions.

        I haven’t heard of anyone planning such a large-scale survey – good to know it’s coming.

        Reply
        1. Mark Hagen Avatar
          Mark Hagen

          No worries! I think you made a good point.

          I do think that the actual answers to “what do people think?” are likely to be inherently confusing and ambiguous, and hard to decoct from aggregated summaries.

          This is also a situation where, as you say, narratives are going to be created by whoever’s loudest, and those narratives are going to create self-fulfilling prophecies and dictate events. Outside the maths community, in the media, I would imagine the “maths community take” is being represented by a few big names, who may represent a certain spectrum of views on AI, but who are all subject to a specific set of pressures (or who have been shaped ideologically by a few decades of playing by institutional rules). That’s not a personal criticism of those people, that’s a general fact about how cultural narratives get created and propagated. So part of the idea here was thinking: “What if a journalist interviewed a large number of mathematicians other than the highest-profile people? What would they say?”

          Reply
  2. anonymous Avatar
    anonymous

    If I may highlight an answer by Anne Thomas, I like it very much:

    “Maybe a month ago I felt some institutional pressure to use AI to be more “productive”.

    I think the recent OpenAI dump shows the futility of entering such a race, so perhaps this pressure will desist?”

    Reply
  3. Anonymous T Avatar
    Anonymous T

    I think a survey like this is a great idea, but the current “anonymous answers” implementation is not great. If the 14 personal questions are answered honestly, reidentifying the responders is a trivial matter. If they are not answered honestly, the results will be misleading (not to mention you ask for personal information to “verify who they are”!). I am happy to express my opinions on AI, but not on the same page where I talk about my housing situation. I doubt I’m the only one, which means the results of your survey will be quite biased.

    Other Anonymous is right, it would be much more useful to have an anonymous multiple-choice survey.

    Reply
    1. Mark Hagen Avatar
      Mark Hagen

      Well, yes, people have to assess risks for themselves and act accordingly.

      If I didn’t ask for (and then delete) some information corroborating people’s membership in the research community, then the survey would become not only biased, but actually full of untrustworthy answers, due to bad-faith participation of various kinds. I am actually not sure what the plan would be for making an anonymous multiple-choice survey resistant to various forms of wholesale fake responses.

      Real survey design of the type you’re discussing is actually hard, and it’s especially hard in a discussion where it’s naive to uniformly assume good faith.

      (I still haven’t rejected any responses, by the way.)

      If someone isn’t comfortable discussing their ideas in conjunction with some information about the material circumstances under which those ideas are being formed, then they don’t have to, and they can take a different survey.

      As I said to the other anonymous commenter, this is not intended as a scientific survey and there is no claim that it is one. It’s intended as a structured discussion and as an opportunity for people to see what others are thinking and roughly from what position they are thinking it.

      Speaking about something in public does indeed entail some social risks, and those are in no way hidden by the survey; there aren’t promises made that a motivated actor couldn’t infer identities from answers; of course they could.

      I actually see no problem with highlighting the views of people who are willing to take the (small) risks involved in speaking in a public way — if people don’t speak, someone else will speak for them. Democracy does in fact make some demands of the participants.

      As I said, other people are planning a more statistically rigorous survey and I think that’s important, but serves a different goal; the two efforts are not in conflict.

      As to the question about housing: in a lot of the world, housing tenure is a very good rough indicator of the level of economic precarity someone faces, and it is in turn a reasonable guess that someone’s perception of economic precarity is going to influence their thinking on AI. The same is true for job security. I don’t really think we should be that interested in a given mathematician’s view on AI if we don’t know, in very general terms, the material conditions for the formation of those views.

      Reply
  4. Anonymous Avatar
    Anonymous

    I would like to answer such a survey, but not if it is hosted on a Google service. Even if I trust you to keep things anonymous, Google knows.

    Reply
    1. Mark Hagen Avatar
      Mark Hagen

      That’s a fair point; if you like, you can find my email on the site where the responses are hosted, and email me your responses, and I can post them with the others. The site where the responses are posted is my personal website, which is not hosted by google, and my email provider is not google; I’m just using a google form to collect the responses. If you do the email thing to submit responses, I won’t need to interact with any google services in order to post them. I don’t think it would be sustainable (from an inbox POV) to do this in every case where people have anonymity concerns because of the form, but if it really is the difference between answering and not answering, please do email me your responses.

      If you know a way to reproduce the functionality of a google form using a service that’s more protective of privacy than google, please let me know, and I’ll look into + consider relocating the form.

      Reply
  5. Anonymous Avatar
    Anonymous

    There are so many interesting responses to the survey. None of them ever appear in the mainstream press, which always interviews the same five people and gets the desired doom or nirvana responses, both of which benefit the AI industry.

    Reply

Comments are moderated. Read our comment policy.

Add to the discussion

Subscribe

New posts by email.

Prefer a feed reader? Subscribe by RSS.

Also on Mathstodon.

Join the discussion

Latest comments across the site.

  1. I actually don’t really understand the argument about “further progress”, which progress are further, which progress are closer, who defines...

  3. A $5000 prize awarded once every three years at the JMM is a step in the right direction, but a...

Read next

Discover more from Proofs and Prompts

Subscribe now to keep reading and get access to the full archive.

Continue reading