# Peer review in the age of artificial intelligence

An episode of [Physics World Weekly Podcast](https://invisiblenetwork.ai/podcast/1400686733), 18 Sep 2025, 28 min.

Audio: https://media.blubrry.com/physicsworldweeklypodcast/content.blubrry.com/physicsworldweeklypodcast/PWWeekly18Sept2025AIPeerReview.mp3

Global survey of reviewers reveals a growing polarization about the use of AI

## Transcript

[0:07] Hello, and welcome to the Physics World weekly
[0:10] podcast. I'm Hamish Johnston.
[0:13] It's the September 18, and we're in the
[0:15] middle of peer review week.
[0:18] And this year's theme is rethinking
[0:20] peer review in the AI era.
[0:24] My guest in this episode is Laura Fiethom
[0:27] Walker, who is reviewer
[0:29] engagement manager at IOP Publishing.
[0:32] As well as publishing Physics World,
[0:35] IOP
[0:36] produces over 100
[0:38] scholarly journals,
[0:39] and it's just reported the results of a
[0:43] new worldwide
[0:44] survey of reviewers' attitudes
[0:46] to the use of artificial intelligence
[0:49] in the peer review process.
[0:52] That report is called AI and peer review
[0:56] 2025,
[0:57] and Laura is the lead author.
[1:00] Here's our conversation.
[1:10] Hi, Laura. Welcome to the podcast.
[1:13] Thanks, Hamish. It's great to be here.
[1:16] So, Laura, before we talk about, this survey,
[1:20] can you just give us,
[1:22] an idea of what IOP Publishing's
[1:24] current policy is on the use of artificial
[1:27] intelligence
[1:28] in the peer review process?
[1:31] And is is is the policy in line
[1:33] with most other scholarly publishers?
[1:37] Yeah. So it's a good question. Our current
[1:39] policy
[1:40] is prohibitive. So we
[1:43] do not accept or condone the use of,
[1:46] large language models to
[1:48] write peer review reports at all or to
[1:51] edit peer review reports.
[1:53] And
[1:54] that's because we have a lot of concerns,
[1:57] ethical concerns about uploading confidential manuscripts into these
[2:01] AI chatbots. We don't know,
[2:04] where that information is being used.
[2:06] It's also because, as we'll talk about in
[2:08] a moment,
[2:09] views of the scientific community are actually really
[2:12] polarized,
[2:13] and there's a large proportion
[2:15] of researchers
[2:16] who do not feel comfortable
[2:18] with AI being used to review their manuscripts
[2:21] in any way.
[2:24] And so we went with,
[2:26] a prohibitive
[2:28] policy.
[2:29] There are lots of publishers who have the
[2:31] same policy,
[2:32] but, actually, when you look across the industry,
[2:34] there's a an enormous amount of variation
[2:37] in what different publishers are mandating.
[2:40] Another common approach is to say that,
[2:44] some publishers accept the use of,
[2:47] generative AI to kind of do light editing,
[2:50] language editing, grammar editing, things like that, but
[2:53] they emphasize that the reviewer is
[2:56] ultimately responsible for the content of that review.
[2:59] Obviously, it's very difficult to police that. It's
[3:01] difficult to police
[3:03] the extent to which,
[3:05] a reviewer might have used a large language
[3:07] model.
[3:09] And there's another another
[3:11] large proportion of,
[3:13] publishers whose policy is that
[3:16] reviewers can use
[3:19] large language models
[3:21] to write or edit their reviews, but they
[3:23] have to,
[3:24] be honest with
[3:26] the
[3:27] publisher and with the authors about how they
[3:29] were used. So they have to
[3:31] usually tick a box or make a statement
[3:33] to say to say how they were used.
[3:36] I think
[3:37] what we really need as an industry is
[3:39] some kind of harmonization of these policies
[3:42] because, you know, often reviewers aren't aware of
[3:46] the specific policy of the publisher that they're
[3:48] reviewing for. And,
[3:50] there are so many publishers, they might be
[3:52] getting lots of lot lots and lots of
[3:53] different requests for different journals.
[3:56] I think we need to work together a
[3:58] bit more so that we're all on the
[3:59] same page, and that will help reviewers, and
[4:01] it will also help authors.
[4:03] And, I mean, it sounds to me that
[4:06] surveys are really needed, aren't they? Because,
[4:09] you know, as you said,
[4:12] reviewers
[4:13] are polarized.
[4:14] I'm guessing that authors are polarized as well
[4:16] in terms of whether they want their,
[4:19] papers to be reviewed,
[4:22] using AI.
[4:24] So
[4:25] with this survey that, that you've just concluded,
[4:29] who did you speak to?
[4:31] How many respondents
[4:32] did you have, and what kind of questions
[4:35] did you ask?
[4:37] Sure. So we put this out to our
[4:39] peer review community. So that's people who have,
[4:42] been invited to review for us or reviewed
[4:44] for us in the past.
[4:46] And we got about 350
[4:49] responses.
[4:50] It was a really diverse group of respondents,
[4:52] really geographically diverse, a good mix of gender,
[4:55] a good mix
[4:57] of, career levels as well and subject areas.
[5:00] So we were quite pleased with the diversity
[5:02] of the responses we got.
[5:04] And
[5:05] it was quite recently that we did another
[5:08] survey. Essentially, in 2024,
[5:10] we did the state of peer review report,
[5:12] which asked
[5:13] much broader questions of the peer review community,
[5:17] and one of the questions we asked was
[5:18] about the use of AI.
[5:21] And this is such a fast moving field
[5:24] that we really felt the need to go
[5:25] out and ask more detailed question
[5:27] questions about this a year later. And what
[5:29] we have found is that things have shifted
[5:31] even in that quite short space of time.
[5:34] And I think we, as publishers, need to
[5:37] understand what our communities are feeling
[5:39] and to kind of
[5:41] stick with them throughout this process and adapt
[5:43] our policies accordingly.
[5:45] So,
[5:46] so we we asked a whole range of
[5:48] questions of these 350
[5:50] people.
[5:51] One of the the main insights was we
[5:54] asked them the same question we'd asked them
[5:56] in 2024,
[5:57] which was,
[5:59] what do you think the impact of generative
[6:01] AI will be on peer review?
[6:04] And we did see a shift. So far
[6:07] fewer people in 2025
[6:09] were neutral.
[6:11] In the 2024
[6:13] survey, 36%
[6:14] of people said, well, I don't think it's
[6:17] gonna have much of an impact.
[6:19] That had reduced to 22%
[6:22] in 2025.
[6:24] So a 14%
[6:25] reduction, and
[6:27] those,
[6:28] respondents had kind of gone to the either
[6:30] end of the spectrum.
[6:31] So there was a 2% increase in respondents
[6:34] who thought that,
[6:36] AI would have a negative impact overall, but
[6:38] a 12% increase in respondents who said that
[6:41] they thought AI would have a positive impact
[6:43] overall.
[6:44] And it's important to bear in mind this
[6:46] is a different,
[6:48] sample. It's a different population of people, and
[6:50] it's a smaller sample size. So it might
[6:52] just be that we we kind of got
[6:54] a different mix of people.
[6:56] But I do have a feeling that
[6:59] things are becoming more polarized, and it might
[7:02] be a a case of more people are
[7:04] aware
[7:05] of these tools now in the last year.
[7:08] And,
[7:11] you know, they've thought a bit more carefully
[7:13] about the impact that it might have on
[7:15] peer review.
[7:17] I see. And, I mean, I suppose,
[7:20] it it does make sense, really,
[7:23] that this polarization
[7:24] will increase
[7:26] maybe simply because people are becoming more aware,
[7:29] as you say,
[7:31] you know, because I I suppose AI has
[7:33] become normalized
[7:35] almost in in everyday life,
[7:38] for a lot of people. Do do do
[7:39] you think it's it's that and maybe a
[7:42] combination of of the fact that people are
[7:44] be have used AI, and so they know
[7:48] what it's capable of doing and that they
[7:50] sort of project that on
[7:52] on how they could use it in a
[7:54] review or how they wouldn't want it to
[7:56] be used in a review.
[7:59] It's a really good question. The question of
[8:01] to what extent are the research community
[8:03] truly aware of
[8:05] what these large language models are capable of
[8:09] and what is a maybe a good use
[8:11] for them in the peer review process and
[8:13] where they shouldn't be used. We asked,
[8:16] lots of free text questions, and we analyzed
[8:19] those responses, which were really interesting.
[8:22] And, again, they were very polarized. There were
[8:24] a lot of people who raised serious ethical
[8:27] concerns.
[8:28] But there were a lot of people who
[8:29] said,
[8:31] well, you know, I use large language models
[8:33] all the time, and, actually, I use them
[8:35] for analysis, and I use them to
[8:39] analyze the manuscript under review.
[8:42] And, really, when you understand how LLMs work
[8:46] kind of under the hood,
[8:47] they really shouldn't be used for scientific analysis
[8:51] and critique. It's not something that they're capable
[8:53] of.
[8:54] What large language models do is they predict
[8:57] what is the most likely next word, and
[8:59] they're capable of producing text which is very
[9:02] convincing
[9:03] but not necessarily very accurate. And that's something
[9:06] we see a lot when we look at
[9:08] fully
[9:09] LLM produced
[9:11] peer reviews.
[9:13] So
[9:14] I I'm not sure that the re the
[9:18] physical science research community, certainly the,
[9:21] some of the respondents to our survey,
[9:24] I'm not sure they are fully aware of
[9:26] how these tools work and what their drawbacks
[9:29] are, you know, where their blind spots are.
[9:33] Yeah. Well, that I mean, that sounds like
[9:34] a problem, but, I mean, I'm guessing, you
[9:37] know, you're you're dealing with, you know, with
[9:39] some, I suppose, smart tech savvy people. So
[9:43] at some point, I I the community will
[9:46] probably have a a better
[9:48] realization,
[9:50] of the issues involved and what AI can
[9:54] and can't do. Mhmm. Another interesting thing about
[9:57] the survey is that it it reveals a
[9:59] split in the views of early and later
[10:02] career
[10:04] researchers.
[10:04] And, you know, perhaps not surprisingly,
[10:07] early career researchers tend to be more positive
[10:11] about the impact
[10:12] of AI
[10:14] than their senior
[10:16] colleagues,
[10:17] whereas later career respondents tend to be more
[10:20] neutral
[10:21] about the possible
[10:23] impacts.
[10:24] I I mean, is that
[10:26] I mean, I know this is sort of
[10:28] stereotypical
[10:30] that, you know, maybe the early career
[10:32] researchers are are more savvy when it becomes
[10:34] to AI and and actually use it, where
[10:37] maybe
[10:38] people further on in their career aren't
[10:42] picking up
[10:43] new tools, you know, they're happy with the
[10:45] the way that they've done things in the
[10:47] past. Is is that the reason why, or
[10:50] is is that just a horrible stereotype that
[10:52] I've,
[10:54] that I've un unveiled?
[10:56] I think it's a very good question. I
[10:58] mean, we did this sub analysis, and, actually,
[11:01] we analyzed the question
[11:03] about what do you think,
[11:05] the impact will be. So that's where we
[11:07] looked at the the generational divide.
[11:10] And, yeah, you're completely right. So early career
[11:12] researchers were much more likely to have a
[11:14] positive view of the impact the future impact
[11:17] of generative AI.
[11:19] And it may well be because they're digital
[11:22] natives. They're generally just more comfortable with these
[11:24] kinds of online tools.
[11:27] But I think it's also
[11:29] important to consider that early career researchers now
[11:32] are starting their careers in a very different
[11:35] landscape.
[11:36] You know, academia,
[11:38] those early career researcher jobs look very different
[11:42] to how they did
[11:44] thirty, twenty, even ten
[11:46] years ago. The volume of,
[11:49] administrative
[11:50] tasks is is is much higher that lots
[11:53] more is expected of their time.
[11:57] There's a lot more precarity
[11:59] in their work, and and so
[12:02] it might
[12:03] be a product of the fact that early
[12:05] care researchers
[12:06] are less likely to have experienced
[12:10] peer review as it traditionally should be, which
[12:12] is a positive experience where,
[12:16] the authors receive really useful and supportive comments,
[12:18] which help them to improve their manuscript, which
[12:20] help them to improve as a researcher.
[12:23] And,
[12:25] you know, and then they can go on
[12:26] and get their paper published. And and,
[12:29] this is purely anecdotal. This is just my
[12:32] opinion, but I wonder if that's also a
[12:34] factor. They haven't experienced
[12:37] peer review in the same way that their
[12:39] senior colleagues have.
[12:42] I I think there are some very interesting
[12:44] dynamics
[12:45] in,
[12:46] you know, how people,
[12:49] adopt AI and why they adopt it. So,
[12:52] yeah, I suppose more more surveys are needed
[12:55] there as well.
[12:57] Absolutely. I mean, we you could just do
[13:00] you could ask so many questions. I think
[13:02] this is so fascinating, and things are moving
[13:04] so quickly.
[13:06] What I really enjoyed
[13:08] reading
[13:09] was the kind of, free text comments,
[13:12] because
[13:14] it was very clear looking at those. In
[13:17] fact, I can I can read you some
[13:18] of them,
[13:20] here?
[13:21] Oh, go ahead.
[13:23] So we asked
[13:24] respondents
[13:25] to tell us whether they thought there were
[13:28] any ethical issues around the use of generative
[13:31] AI in peer review, and
[13:34] there was a real range of responses. So
[13:37] I'm just gonna read you a couple.
[13:39] The theft of the corpus of data
[13:42] used to train AI models, the replacement
[13:45] of human labor, and the wasteful energy usage.
[13:48] That's one comment.
[13:50] Another comment was
[13:51] the main ethical issue is the transfer of
[13:54] responsibility over knowledge from a human intelligence to
[13:56] a nonbiological
[13:57] intelligence with unknown administrators
[14:00] or proprietries.
[14:03] And so these are quite
[14:06] existential ethical
[14:07] concerns, actually. They're not necessarily concerns about the
[14:10] capabilities
[14:12] of the software.
[14:13] Often, people
[14:14] who,
[14:17] are anti AI and don't like the idea
[14:19] of AI being used in peer review have
[14:21] quite deep seated,
[14:23] broader concerns than the fact that it might
[14:26] not produce very good reports.
[14:29] We also ask people,
[14:32] do you think you would be able to
[14:33] detect
[14:34] an
[14:35] AI authored,
[14:37] peer review reports if, you were an author
[14:40] and you received one on your manuscript.
[14:42] And and we'd ask them, what do you
[14:44] think the hallmarks are of these reports?
[14:47] And
[14:48] what came up again and again and again
[14:50] was
[14:52] these tools just do not have the depth
[14:54] of knowledge of an expert
[14:56] peer reviewer.
[14:57] And that was really the most common responses.
[15:00] They've seen these reports that are written by
[15:02] generous of AI tools, and they they're very
[15:05] high level.
[15:06] They
[15:07] make very broad generic statements, and they're just
[15:11] not useful. And it's it is quite obvious
[15:13] that they've been written
[15:15] by Generative AI. We see, unfortunately, we see
[15:18] quite a few fully
[15:21] generative AI written peer review reports, and we
[15:24] find the same thing. You know, they're just
[15:26] not up to scratch. They're not helpful, and,
[15:28] actually,
[15:29] we can spot them a mile off. That
[15:31] is not the same
[15:33] necessarily
[15:35] as reports
[15:37] that have been edited or augmented
[15:39] by AI. And that's where we get into
[15:41] well, there's different uses, and it's quite difficult
[15:43] to police how people are using them. Because
[15:47] when we ask people,
[15:48] in this survey,
[15:50] have you used generative AI tools, and if
[15:53] so, how?
[15:55] The most common response was, well, I wrote
[15:58] my review. Maybe they wrote it in kind
[16:00] of bullet points or in note form, and
[16:02] then they put the review into an a
[16:04] AI tool to improve flow and grammar.
[16:07] And so that was, you know, by far
[16:09] the most common,
[16:10] response, and
[16:13] that's a completely different use case to uploading
[16:16] a manuscript into chat GPT or whatever and
[16:19] asking it to to summarize to to
[16:22] write a review.
[16:23] Although,
[16:25] that was the second most reported use of
[16:27] AI tools was to,
[16:30] so 46 people said that they had they
[16:32] had used generative
[16:33] AI to
[16:34] digest or summarize an article under review.
[16:40] And that,
[16:41] for me, is a little bit concerning because
[16:44] that is not
[16:45] what
[16:46] current large language models are good at.
[16:50] In fact, there was a really interesting
[16:53] study that was published last week by,
[16:56] and colleagues at Sheffield University in Tokyo in
[16:58] Finland.
[17:00] And
[17:01] they asked they looked specifically at ChatGPT,
[17:04] and they asked ChatGPT
[17:06] to kind of summarize and critique
[17:08] a large number of manuscripts that had been
[17:11] retracted. So they'd been retracted for serious ethical
[17:14] concerns.
[17:15] And on the whole, chat GPT was very
[17:17] positive about these papers. It very rarely picked
[17:20] up on the issues that had led to
[17:22] the retractions.
[17:24] Generally speaking,
[17:27] it it tends
[17:29] to be very positive when you put
[17:31] a manuscript into one of these chatbots.
[17:34] You're gonna get a pretty positive,
[17:37] gentle peer review, which isn't very accurate.
[17:40] So do do you think there that,
[17:43] is the
[17:44] is chat GBT or the large language model
[17:46] trying to please,
[17:48] the person
[17:49] who's who's using it and,
[17:52] you know, saying,
[17:54] sort of couching any sort of review in
[17:56] positive language,
[17:59] because maybe that's more acceptable to,
[18:02] to the person who's making the request.
[18:04] Well, it's interesting you ask that because there
[18:07] was recently,
[18:09] in the news,
[18:10] reports that
[18:12] the new versions, particularly of ChatGPT,
[18:16] were to be made less sycophantic.
[18:18] So previous
[18:20] versions were very sycophantic, very, you know, positive
[18:23] about everything the user was saying. Maybe. Maybe
[18:26] it's that.
[18:29] I would imagine that in the study of
[18:31] retracted,
[18:34] manuscripts, though, where they asked ChatGPT to summarize
[18:37] the manuscripts, I would imagine the prompts
[18:40] the prompts would have been quite neutral.
[18:43] Who knows?
[18:44] Yeah. Well, that that's the problem, isn't it,
[18:46] that we don't know?
[18:48] And I I just wanted to check something
[18:50] with you, Laura. So
[18:52] if a reviewer were to upload,
[18:55] a manuscript,
[18:57] and use, let's say, chat GBT to provide
[19:00] them with a summary, that's something that we
[19:02] don't
[19:03] want people to be doing. Is that is
[19:05] that correct?
[19:07] So that summary
[19:09] being submitted
[19:10] as a peer review,
[19:12] no. That's not. And, actually, as I say,
[19:15] our policy also, doesn't prohibit the use of
[19:18] these tools
[19:19] to to edit reports either.
[19:24] Perhaps it's worth talking about
[19:27] people's feelings
[19:29] when we ask them to put their author
[19:32] hats on,
[19:33] and we ask them how would you feel
[19:35] if your manuscript
[19:37] was reviewed in full or in part
[19:40] by a large language model? And the responses
[19:42] were quite telling.
[19:44] So,
[19:48] 57%
[19:49] of people said they would be unhappy
[19:51] if a reviewer used generative AI to write
[19:54] an entire pair of your report on a
[19:56] manuscript that they had coauthored.
[19:57] And when we ask people, okay,
[19:59] so that's writing a full report, how would
[20:02] you feel if,
[20:03] AI was used to augment or edit
[20:06] a pair of your report? And the response
[20:08] there was still quite high. It was in
[20:10] the forties of people saying,
[20:12] roughly 40% of people saying they wouldn't be
[20:14] comfortable with that.
[20:16] So, no,
[20:17] we definitely don't want fully AI authored peer
[20:20] review reports. There is a real quality issue,
[20:23] and there are lots of ethical issues.
[20:25] And we know that
[20:26] a majority of our authors
[20:29] wouldn't be happy with that.
[20:31] There's a whole other practical
[20:35] concern here as well, which is
[20:37] these
[20:38] tools
[20:39] are currently
[20:40] free, and they're available to anyone with an
[20:43] Internet connection.
[20:46] If an author wanted an AI authored peer
[20:48] review report,
[20:50] they can do that themselves, right, in seconds.
[20:54] That's not what
[20:56] they submit their manuscripts
[20:57] to journals for. That's not what they expect
[20:59] from peer review.
[21:02] And,
[21:03] again,
[21:04] reading into the free text comments,
[21:07] we have these two words which describe the
[21:10] process, peer review.
[21:12] And I think often people focus on the
[21:14] review, like this is about
[21:16] assessing scientific rigor.
[21:17] But the peer part is really important to
[21:21] people, and that came across loud and clear
[21:23] when people responded to to our questions because
[21:27] they really do feel strongly about
[21:31] this manuscript, this research that they've worked really
[21:33] hard on being assessed by
[21:36] someone who's an expert, a real person, perhaps
[21:38] someone that they know and have met at
[21:40] conferences,
[21:41] but certainly someone with the the depth of
[21:44] knowledge
[21:45] and passion about the field to give
[21:48] a good and useful review.
[21:51] And, Laura, I wanted to ask you about,
[21:54] well, I suppose the survey in general, but,
[21:57] you know, perhaps more about the written comments
[21:59] where I suppose you you're probably
[22:01] you're probably,
[22:03] learning about people's passions
[22:05] about AI.
[22:07] Do do you think that,
[22:10] these two different views,
[22:12] you know, this polarization,
[22:14] is it going to be difficult
[22:16] to reconcile
[22:18] as,
[22:19] as we go forward? Because,
[22:21] well, AI is it's here to stay. It's
[22:23] a useful tool.
[22:25] People are using it more and more every
[22:27] day and learning how to use it effectively.
[22:31] Is it gonna be difficult to reconcile the
[22:33] views of the sort of never AI people
[22:37] with the, well, I can use AI responsibly
[22:40] when I do a peer review people. Is
[22:42] that gonna be tough?
[22:44] So
[22:46] I actually think the opposite.
[22:47] I actually think that there's a lot of
[22:50] noise at the moment, and there's been a
[22:52] lot of rapid change,
[22:53] and that will lead to divisions
[22:56] and differences of opinion.
[22:59] But as much as I sound like I've
[23:01] been quite hard on these tools and and
[23:03] quite negative about them, I am an optimist.
[23:06] And I do believe in the power of
[23:08] communities
[23:09] to kind of reach a consensus and reach
[23:11] some equilibrium.
[23:14] The way that these tools have been presented
[23:19] from the top down has often been quite
[23:21] fatalistic.
[23:22] So,
[23:24] the line seems to have been in a
[23:26] lot of cases.
[23:28] These tools are here now, and if you
[23:29] don't get with the program, if you don't
[23:31] use them, you're gonna be left behind.
[23:35] And a lot of people have got on
[23:36] board with that, and they're happy with that.
[23:38] They think, well, you you know, it's true.
[23:39] You can't put the genie back in the
[23:41] bottle. Let's just start using them. But there's,
[23:43] you know, there's a backlash against that view
[23:45] as well with communities and people saying, well,
[23:47] no. We do have a choice, and we
[23:49] we
[23:51] can think in a more nuanced way about
[23:52] the ethics
[23:54] and how we want these tools to be
[23:55] used. And I think, fundamentally,
[23:58] throughout human history,
[24:01] the way that tools are adopted and used
[24:03] and whether they become widespread
[24:06] is not something that's dictated from the top
[24:08] down.
[24:09] It's something that's dictated from the bottom up
[24:11] through communities
[24:13] who
[24:14] get together, test out tools,
[24:17] and decide whether these tools are gonna help
[24:19] them achieve their goals.
[24:21] And I think that that is what's going
[24:23] to happen
[24:24] with a little bit more time
[24:26] in the physical science research community. You know?
[24:29] Ultimately, the community is going to decide,
[24:31] and and we, as publishers,
[24:33] need to follow their lead.
[24:36] In the short term, when we are seeing
[24:38] a divergence of views, what we need to
[24:41] be careful to do is make sure we
[24:42] bring everyone along with us.
[24:45] And that includes, you know, different groups. We've
[24:47] discussed how there's a generational divide. Okay. How
[24:49] do we do this? How do we have
[24:51] policies
[24:52] that are inclusive for early career researchers and
[24:54] senior researchers?
[24:56] And
[24:58] the ability to opt in and out will
[24:59] be important,
[25:01] but the key thing will be transparency.
[25:04] So,
[25:06] real transparency
[25:08] around
[25:09] how these AI models are being used,
[25:13] and how we as publishers are using AI
[25:16] in our processes as well.
[25:18] I see. Yeah. Because, I mean, it's, you
[25:20] know, it's not just us, is it? It's
[25:22] not just the scientific community that's grappling
[25:25] with AI and sort of an overabundance
[25:28] of
[25:29] information,
[25:30] some of it very poor. This is, you
[25:33] know, this is the issue, isn't it, that's
[25:35] facing society
[25:37] at the moment?
[25:38] So, you know, really, we're not alone
[25:41] in this.
[25:42] But I I on the other hand, I
[25:44] may I suppose that, you know, we could
[25:46] lead the way
[25:47] as scientific
[25:48] publishers providing
[25:50] a
[25:52] a a a way of using AI in
[25:54] a responsible way to to to process information
[25:57] and and make the world a better place.
[26:00] Absolutely.
[26:01] I think
[26:02] what the future might hold is
[26:05] tools that are data safe, that are ethical,
[26:08] that protect research integrity, and that are kind
[26:11] of embedded within our systems. So there's, you
[26:13] know, complete transparency
[26:15] and safety in the way that they are
[26:18] used.
[26:19] And as I
[26:20] say,
[26:21] allowing,
[26:23] researchers to opt in or opt out of
[26:25] their use. So there will still be people
[26:27] with strong views, and they should be able
[26:29] to to bypass this if they want to.
[26:34] So exciting times as usual,
[26:37] I suppose, in the publishing industry. Thanks, Laura.
[26:40] Thanks so much for, coming on the Physics
[26:43] World podcast,
[26:44] Physics World weekly podcast,
[26:59] That was Laura Fietham Walker, who is reviewer
[27:02] engagement manager at IOP publishing.
[27:06] The reviewer survey is called AI and peer
[27:10] review 2025,
[27:12] and it can be found on the IOP
[27:14] publishing website.
[27:16] I'll put a link to the report in
[27:18] the podcast notes.
[27:20] I'm afraid that's all the time we have
[27:21] for this week's podcast.
[27:23] Thanks to Laura for a fascinating
[27:25] discussion, and a special thanks to our producer,
[27:29] Fred Isles.
[27:30] We'll be back again next week. See you
[27:33] then.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.