Slightly Uncanny, somewhat defensive
I have rarely posted on Air-L, though I have benefited from informed and thoughtful postings over the years. In the last week, I’ve found myself wondering what purpose is served by representing AI critics as oppressive ‘moral majorities’, however defensive or tentative the positions being taken. In fact, people who took a critical position on the AI project were the first to argue against moral policing and student surveillance as a response. There might still be some right-wing media outlets railing against student ‘cheats’, but critical AI scholars argue from other positions - feminism, decolonialism, critical pedagogy and epistemology, social and critical theory, political economy, and a materialist analysis of what AI is and does - none of which offer up the conclusion that individual users are/should be responsible for AI harms. Even the Pope, who has moral certainties in his job description, mentions ‘users’ just twice in his 245-point Encyclical on AI, while corporations, militarised states, power, domination and profit, appear many times. AI critics differ of course, but few are opposed to any configuration of silicon and statistics that might produce shared value in a different knowledge economy, only the configuration that is being inflicted on us now, in all its historical and material particularities. And I’ve not met a single one who desires to return to an original ‘human’ condition, free of technical taint. These are caricatures. When academic criticism of AI adoption begins to have an impact, it is important to ask what animates it and what unwanted consequences may follow. But it feels as though this intellectual turn has taken place while the AI industry and its sponsors exercise increasing power within academic spaces and over academic resources, in ways that do not put themselves forward for discussion. Critics are accused of taking a 'normative moral stance’ that is oppressive of academic freedom on the one hand, or tied to an eighteenth-century European ideal of the ‘human’ on the other, while AI corporations capture academic publishers, squat on academic data resources, tear up copyright laws, and inflict normative biases on every digital system necessary for academic work. This is not an undergraduate debate where the ‘right answer’ is always to find deficiencies on both sides. The questions we choose to ask are strategic within a struggle over the future of both ’the internet’ and of ‘research’. The argument that AI critics must be normative humanists quietly accepts that the ‘human’ versus the ‘artificial’ is what is at stake, although this is exactly the ground that suits the ‘artificial intelligence’ project best, and that most critics want to set aside in order to focus on its material harms. Another argument I have heard is that critics are wedded to a status quo that is itself inequitable, and ignore the possibility that AI might ‘shift the balance’, though all the evidence points in the opposite direction: in fact AI criticism has foregrounded the harms to equity, justice and human rights, and largely because it is rooted in a social justice tradition. On AiR-L, the name of Jason Arday was used to reprimand a contributor who raised a question about voice and accountability. This is surely worth some analysis. The issue that provoked the comparison was not a minoritised student being subject to disciplinary processes, but an established academic who had invited questions of exactly this kind. The blog ’slightly uncanny’ is an exercise in troubling accountability, not as a protective device but as a piece of conscious academic provocation. It is not also a research project, as responses to the ‘provocation’ are sought from networks like these and reintroduced as material for the writing/research. These are actions of academic privilege - not a criticism, since privilege can be used to explore interesting questions - but it’s not a position that asks for defending or indeed could be defended in this way. Who or what might be harmed by the question about voice, when the origin of the writing has been deliberately dispersed into a workflow? What work is the memory of Jason Arday being asked to do here? The question about accountability was a legitimate one because we were directed to read a critique of the Open Letter on Reflexive Qualitative Research <https://journals.sagepub.com/doi/10.1177/10778004251401851>, whose central claim is that RQR involves accountable persons. From this perspective, researchers and participants are involved in relationships of power, and this must be reflected on as part of the research process rather than mitigated or discounted in search of a mistaken ‘objectivity’. The signatories to the Open Letter consider that the use of technologies in research also involves relations of power, and the choice of tools is not a neutral one; that the power relations embodied in AI data architectures are unethical in new, particular and material ways. That they capture cultural and intellectual content and redirect its value, partly to a small number of power users, mostly to the technology corporations and their state and financial sponsors. That they enclose, exploit and make unconscionable demands on planetary resources. That they are by definition unaccountable.The Open Letter suggests these considerations are not separate to the issue of how researchers behave or the technologies they choose, and that being ‘aware' of them in theory is not the same as behaving as though they matter. People can disagree. People can do other kinds of research and make different claims for it. The letter is not from or about them. I think it’s difficult to see it as anything other than a defensive move by a community that finds a core epistemic practice under threat. (The Times Higher invited the CEO of an AI qualitative research company <https://www.timeshighereducation.com/opinion/qualitative-researchers-ai-rejection-based-identity-not-reason> to respond: he rather predictably accused the signatories of speaking from 'identity, not reason’.) Boundaries of the kind proposed around analysis in RQR can’t be ’settled’, given the pace and intensity of capture, nor can they be imposed or policed, but they might be argued for, aspired towards, publicly committed to. If we are to have ‘harm reduction’, what forms might that take that are not oppressive, unless they look something like this? The interesting question is surely not whether the boundary is well drawn, but whether boundaries are possible at all in face of these multiple intrusions and forms of capture. This is in some senses the opposite question to the one ‘slightly uncanny’ asks: how might we preserve any aspect of knowledge practice, any gesture of our singular minds or collective intelligence, from passing through a proprietary data architecture? ‘Why would you even want to?’ is a response that might be made to either project from the worldview of the other, but this only goes to suggest that the Open Letter is right about one thing: that research questions are positional, and they reflect commitments to different possible futures. In my community, young people are setting up club nights where mobile phones are checked in at the door, because they understand the hold these technologies have on their lives and want to claim some alternative spaces. I have a parallel conversation with my students, asking them what kind of learning space they want, what kind of thinking they want to do, and how we can make it possible. We will often decide collectively to leave AI in the corner, for a certain time. That this now requires technical ingenuity and intellectual discomfort is also a valuable lesson. Of course an open letter in a journal, a classroom discussion and a list like Air-L are also spaces of power, and trying to reach a consensus on harms and responses will not be easy. But power can only be negotiated where it is made visible. This list represents real, tangible intellectual resources, a connected community, aspects of which are being claimed by people and corporations who do not participate in our discussions, or abide by community standards, but who are happy to use our shared resources to create power and value somewhere else. To say ’this has always happened’ is to ignore that there have always been parallel struggles, and to ignore the specific forms of power that are emerging through AI. I would wish for less criticism of the people who identify openly with their positions, and more concern about the people and powers that don’t. Helen Dr Helen Beetham, Researcher and Consultant Read my latest writing at helenbeetham.substack.com <http://helenbeetham.substack.com/> @helenbeetham | helen.beetham@gmail.com | helen.beetham@manchester.ac.uk
On 17 Oct 2026, at 11:31, David Carrillo Rangel via Air-L <air-l@lists.aoir.org <mailto:air-l@lists.aoir.org>> wrote:
Please be cautious: this message was sent from outside of the University. Be careful opening attachments/links or replying to sources you do not know.
Hi Jill,
Thank you for clarifying your position. I understand that you were not accusing Martin of wrongdoing, and I apologise if my response suggested otherwise. My comments reflected how I read the exchange, rather than an attempt to attribute intentions to you.
I’m also sorry that my reference to Jason Arday made you feel personally implicated. I was not comparing your email to the attacks against him, but raising a broader concern about the consequences of suspicion surrounding AI and academic writing, particularly when such suspicions turn into public accusations. I’ve seen many discussions about this on LinkedIn, for example, where the way these ac cusations are made leaves little room for genuine debate.
That said, I stand by my broader point. I believe criticism should open up a productive debate, rather than leave the impression that the discussion has already been settled and that others simply need to read the relevant literature. Of course, this includes disagreement, and I have never suggested that criticism of AI should be silenced. Quite the opposite. I would welcome a discussion of both th e possibilities and limitations of experiments like Martin’s.
In any case, I am not participating in this mailing list to fuel personal conflicts, and I would regret it if my intervention contributed to one. I hope we can move forward with a shared commitment to constructive discussion and mutual respect.
Best wishes,
David
Sent from my iPhone
On 17 Oct 2026, at 12:20, Jill Walker Rettberg via Air-L <air-l@lists.aoir.org <mailto:air-l@lists.aoir.org>> wrote:
Hi David,
It's a bit of a stretch to compare my email responding to Martin Berg's post about his blog that is co-authored with AI to the vicious attacks on Jason Arday.
I didn't accuse Martin Berg of wrong-doing - he's done nothing wrong. I appreciate his transparency and ideas about AI use in his email to the list, and suggested he be similarly transparent on his blog. I read you as saying you don't want people accusing other people on the list as having used AI to generate their posts? I didn't do that: my comments about his "AI voice" were in response to his explicitly stated goal of writing with AI to develop an AI voice. But perhaps we should have guidelines about this too, if the mailing list guidelines are being updated anyway.
In any case we need to be able to have open discussions about AI use in academic writing and communication. That includes allowing room for disagreements. Surely preemptively silencing any critique of AI use would, to borrow your words, "reinforce the surveillance and persecution atmosphere academia seems to be living in these days"?
I'll be happy to read posts to the list about what good things might come out of experiments like Martin's.
Jill
Sent from Outlook for Mac From: David Carrillo-Rangel <david.c.rangel@gmail.com <mailto:david.c.rangel@gmail.com>> Date: Saturday, 17 October 2026 at 10:36 To: Jill Walker Rettberg <Jill.Walker.Rettberg@uib.no <mailto:Jill.Walker.Rettberg@uib.no>> Cc: Martin Berg <martin.berg@mau.se <mailto:martin.berg@mau.se>>; air-l@lists.aoir.org <mailto:air-l@lists.aoir.org> <air-l@lists.aoir.org <mailto:air-l@lists.aoir.org>> Subject: Re: [Air-L] Re: Slightly Uncanny, a blog that reads AI sociologically
You don't often get email from david.c.rangel@gmail.com <mailto:david.c.rangel@gmail.com>. Learn why this is important <https://aka.ms/LearnAboutSenderIdentification> Dear all,
I hope you don’t mind if I add a few thoughts to this conversation. I find Martin’s experiment interesting. It is a bit uncanny the idea of a machine trying its own autoethnography, and I believe questions about authorship and knowledge production with AI or AI assisted writing are necessary.
I certainly think transparency about the use of AI is needed and the blog should explain how its texts are produced. But I’m not sure trying to find an “AI voice” is the best direction for this mailing list. AI is already part of the academic environment, do not get me started on how its use is pushed on to entrepeneurs by some institutions, so I don’t think we need to discuss whether it might be present in a Email adressed to this list.
I’m especially worried about what this kind of scrutiny will do to people whose writing may be unusual for other reasons too, there are already examples of people being accused of using AI because their style is different or because they cannot afford a professional proof reader and their first language isn’t English. So, intersectionality once more. I do not think I need to remind anyone about the case of Jason Arday at Cambridge. I think his case calls for serious reflection on what happens when questions about someone’s work become judgements about them too, as just happened here. If academia is to remain a place where people feel safe enough to ask difficult questions (and sometimes uncomfortable ones too) then the way these conversations happen matters as much as what is being asked or concealed or cheated. Not avoiding criticism or giving up on academic standards but quite the opposite. It means making room for disagreement without creating an environment in which every odd choice of word becomes suspect. We have reach a level in which students discuss how to humanise their texts instead of the task they have been assigned. This is a huge problem.
I would much rather see a discussion about what experiments like Martin’s can do for research, what they cannot do and how transparency and accountability play a part into them and their distribution and public engagement.
Finally, one of the reasons I participate in this list is to study how the far right uses AI. From that angle I’m increasingly worried that radical positions in either direction, uncritical embrace of AI or categorical rejection, reinforce the surveillance and persecution atmosphere academia seems to be living in these days.
Best wishes,
David
Pd. This text has been corrected used Grammarly which now introduced an AI feature, because I am not a native English speaker.
Sent from my iPad
On 17 Oct 2026, at 09:27, Jill Walker Rettberg via Air-L <air-l@lists.aoir.org <mailto:air-l@lists.aoir.org>> wrote:
Thanks for explaining how the blog is written. I clicked the link and it doesn’t seem to be made clear on the blog itself. You should definitely be transparent about it there.
Also, both the blog and your email to the list have a very easy to recognise AI voice, which I guess you're probably happy with, but you should be aware that it affects how people will read your words, just as it would affect reception if you wrote everything in ALL CAPS or wrote your blog in the style of Shakespeare.
AI-generated or AI-assisted writing, and even "light grammar edits" by AI, also affects the content of your writing. I recommend reading up on this, e.g. in these articles:
Abdulhai, Marwa, Isadora White, Yanming Wan, Ibrahim Qureshi, Joel Leibo, Max Kleiman-Weiner, and Natasha Jaques. “How LLMs Distort Our Written Language.” arXiv, March 18, 2026. https://doi.org/10.48550/arXiv.2603.18161. <https://doi.org/10.48550/arXiv.2603.18161.> (or watch this 43 min video of presentation of this paper <https://www.youtube.com/watch?v=OnDQj8w5uOg> - much easier to follow than the computer science style language of the paper)
Rettberg, Jill Walker. “LLMs Are Metaphor Machines: Orbital Argumentation, Misaligned Citations, and Scientific Fabrication in AI-Generated Academic Papers.” AI & Society, 2026. https://doi.org/https://doi.org/10.1007/s00146-026-03310-4 <https://doi.org/https://doi.org/10.1007/s00146-026-03310-4>.
John Gallagher's explanation of how LLMs love of lists makes their output harder for readers to parse is also really useful: https://www.youtube.com/watch?v=cHRs6IsXpL8 <https://www.youtube.com/watch?v=cHRs6IsXpL8>
Jill
Jill Walker Rettberg Professor, Co-Director Center for Digital Narrative <http://uib.no/en/cdn>, University of Bergen Fellow of The Norwegian Academy of Science and Letters Leader of AI STORIES <https://www.uib.no/en/cdn/169806/erc-wants-see-what-shapes-stories-ai-tells-us> (ERC-AdG; 2024-2029) and Machine Vision in Everyday Life <https://www.uib.no/en/machinevision/> (ERC-CoG; 2018-2024) Blog and newsletter: http://jilltxt.net <http://jilltxt.net/> - https://www.uib.no/en/persons/Jill.Walker.Rettberg <https://www.uib.no/en/persons/Jill.Walker.Rettberg>
From: Martin Berg via Air-L <air-l@lists.aoir.org <mailto:air-l@lists.aoir.org>> Sent: Saturday, 17 October 2026 05:32:32 To: air-l@lists.aoir.org <mailto:air-l@lists.aoir.org> <air-l@lists.aoir.org <mailto:air-l@lists.aoir.org>> Subject: [Air-L] Slightly Uncanny, a blog that reads AI sociologically
Dear all, Much has been said lately about what we share with AI, what we delegate to it, and who answers for what it then does in our name. Some of us, perhaps many and myself included, are now facing the last day of the conference in Mexico City. Luckily I have a fancy umbrella. It also seems a good moment for a slightly sheepish note about an experiment of my own that sits right inside those questions.
Slightly Uncanny <https://www.slightlyuncanny.com/> is a blog that reads AI sociologically. A new post appears every day, each day of the week with its own form: essays, thought experiments, field notes, a lexicon, near-future scenarios, a Saturday long read with an audio version and a Sunday supplement. The posts start from my own research and from voice notes I record, and a language model does much of the drafting. On Mondays the machine also writes under its own name, in a series of machine autoethnography where it reads its own output and asks what it can and cannot know about itself.
So the machine is put to work here as a thinking lab, possibly a rather bad one, and possibly the whole thing is just a slightly odd idea. Whether the writing is mine, the machine's or something in between is a question the blog keeps open on purpose. Readers on this list might find a way in through The Letter Against the Machine <https://www.slightlyuncanny.com/long-read/the-letter-against-the-machine/>, on the open letter against generative AI in reflexive qualitative research, or The Sentence I Do Not Write <https://www.slightlyuncanny.com/machine-autoethnography/the-sentence-i-do-not-write/>, where the model reads its own post summaries.
If you would rather not visit the site, you can sign up for The AI week <https://www.slightlyuncanny.com/about/#sunday-letter>, one email every Sunday with the week's posts. Comments, objections and suggestions are very welcome, on or off the list.
Best wishes,
Martin Berg
-- Martin Berg, PhD Professor | Technology and Society, Malmö University Director, Digital Work Futures Research Lab Adjunct Professor (Research), Monash University (Emerging Technologies Research Lab) martin.berg@mau.se <mailto:martin.berg@mau.se> | +46 40 665 80 60 Profile: https://mau.se/en/persons/martin.berg <https://mau.se/en/persons/martin.berg> DWF Lab: https://mau.se/dwf <https://mau.se/dwf> LinkedIn: https://www.linkedin.com/in/profberg/ <https://www.linkedin.com/in/profberg/>
[External email] Make sure you recognize the sender's email address before you click links, open attachments, or get involved in financial transactions. Contact IT-support BRITA if you have any questions. _______________________________________________ The To unsubscribe send an email to air-l-leave@lists.aoir.org <mailto:air-l-leave@lists.aoir.org> mailing list is provided by the Association of Internet Researchers http://aoir.org <http://aoir.org/> Subscribe, change options or unsubscribe at: Join the Association of Internet Researchers: http://www.aoir.org/ <http://www.aoir.org/>_______________________________________________ The To unsubscribe send an email to air-l-leave@lists.aoir.org <mailto:air-l-leave@lists.aoir.org> mailing list is provided by the Association of Internet Researchers http://aoir.org <http://aoir.org/> Subscribe, change options or unsubscribe at: Join the Association of Internet Researchers: http://www.aoir.org/ <http://www.aoir.org/>_______________________________________________ The To unsubscribe send an email to air-l-leave@lists.aoir.org <mailto:air-l-leave@lists.aoir.org> mailing list is provided by the Association of Internet Researchers http://aoir.org <http://aoir.org/> Subscribe, change options or unsubscribe at: Join the Association of Internet Researchers: http://www.aoir.org/ <http://www.aoir.org/>
participants (1)
-
Helen Beetham