If you can’t be bothered writing it, I can’t be bothered reading it.

by Hannah Forsyth on August 4, 2026

The producerist vision of AI seems to assume an endlessly expanding market for reading. And reading. Well. It is rather more than a market.

Warning: this post is a little more sweary than usual. If that bothers you, maybe read something else.

We read aloud in our household. A lot.

We always have.

When my daughter was five years old (she’s now 26) she and I once squished into a beanbag and read Hans Christian Andersen’s Snow Queen aloud. When we got to the end of the story I realised to my horror that we were so absorbed that it was her bed time and we (OK, responsible adult here, I) had forgotten we also needed to cook and eat dinner.

We scrambled through the Snow Queen incident but didn’t ever stop reading aloud. We read all the Harry Potters (yeah, yeah I know, don’t @ me – but they were great to read aloud) multiple times. If we drove somewhere, or were on holidays, one member of the family – and taking turns is important – would read. Vacation campfires were kept aflame by reading aloud. Stories, Poetry. Non-fiction.

My daughter’s been out of home for several years but my beloved and I do the same, still. We read Carlyle’s French Revolution aloud, until it got boring and we switched to Hilary Mantel’s version. We’ve read EP Thompson’s ‘Moral Economy of the English Crowd’ and Kindleberger’s Financial History of Western Europe aloud. And many, many more.

Recently, somehow not aware that a major film was in the offing, we’ve been reading Homer’s Odyssey – the Emily Wilson translation, obviously. It is fucking great. When you hit the Cyclops, the iambic pentameter really takes off and it has energy. Wondrous.

I saw the film the day it was released, approximately 24 hours after learning it existed – I loved it, don’t @ me – and have coerced my beloved (who hates going to the movies) into coming with me again. Yes, six hours of my one precious life at the movies watching the same film. Totally worthwhile.

As we were reading Book 10 the Odyssey about Circe tonight I was struck by what precious time reading is. And – like you, I expect – it is so hard to ever have enough of it. I know most people think Homer’s Iliad and Odyssey had multiple authors, but their coherence to me suggests a singularity of vision, at least. This will matter in a minute.

Reading time is scarcer than I wish. I honestly can’t keep up with everything written by friends and in my field(s), let alone pleasurable reading. I subscribe variously (as much as I can afford…so the profile changes a bit) to New Left Review, Jacobin, Solidarity, Overland, Griffith Review, LRB and a bunch of academic journals especially Labor, Economic History Review and Labour History. Getting through these plus everything people send me and the articles in the journal I co-edit, Australian Historical Studies. Well, frankly, it is a fuck load of reading.

And that doesn’t even include the reading I need to do to prepare for writing the book I want to write, or to do the paid work I do. And it definitely doesn’t include the dozens of other substacks, nor the FT, nor the ABC’s markets blog that I follow religiously.

I’m sure you feel this too.

So. I’ve been beyond puzzled to see people getting chatty (or other AI) to write their stuff.

I totally get the pressure to produce. Academics have been long turned into content churners. ‘Publish or perish’ in some regions and institutions meant how much could you produce in how little a time. As a postgraduate student I attended a workshop where a senior professor describes turning the same conference paper into five publications FIVE!? Who do they think was going to read all this samey shit?

Later, ‘publish or perish’ meant ‘quality’, which some senior professors could claim also meant producing less (the rest of us were not typically afforded such luxuries). Quality ofc was measured by some juvenile fucking rankings of outlets in which you published, not – as was implied by the word quality – how good was the work.

And then quality went out of favour, and ‘impact’ came in. Show us your ‘impact’ they said. My first book in 2014 was quoted in the Federal parliament. Recently both my books were cited in a submission to the NSW parliament. Which is fucking IMPACT, right? Well, actually it just means I said exactly what the Shadow Minister Kim Carr needed. It didn’t actually change anything. Still. ‘Impact’ was really only meant to be performative, because uni bosses saw them as metrics, not real things.

How stupid – disconnected from any reality – are the metrics under the logic of managerialism? But we all internalised them. And when academics do it, how can we be surprised when students do, too?

Of course, this is all a symptom of what Pete Chalmers in this excellent essay in the magazine Arena has called the ‘thru-put model’ of higher education:

By around 2012, Australian universities had embraced the ‘thru-put’ model: big-box, drive-thru, on-demand, in the cloud, flexible. As a system, it is designed to smoothly pump through as many students as possible, not to provide a better education to a greater number, but because it is both profitable and scalable….

Inside the field generated by the thru-put model, bot lecturers and bots who do your homework for you pop up as an appealing step-change in efficiency, led and controlled by the universities supplying it.…

The thru-put model is not about knowledge, learning or pedagogy, but rather is a wealth transfer racket, and it works because it’s so scalable while labour costs are trivially low. Learning may or may not happen, and sometimes good learning still does. But it’s a racket because even if it doesn’t work, students are indebted, while the core beneficiaries are always the most handsomely paid, regardless of performance.

Writing is learning. It is thinking. It is problem solving. It is collaboration. Writing is the path by which we find stuff out, critically, analytically.

We are all looking on in horror as students trade learning for robot outputs. Understandable in an environment where AI seems the only reasonable path to a decent grade and decent grades may help secure a future in a seriously volatile world. This (they reason) will get them what the metrics require. But in fact we’ve largely been complying with the logic that made this happen well before the tech automated it and cut out the middle man (human brains) that seemed superfluous if the essay as output, not the learning that came with it, was the point.

The same has applied to professionals. Several of the writers I follow have semi-regularly said that this was a newsletter written by AI. It was the author’s prompts, sure. But the robot’s words.

I get it. But geez, I have fucking enough to read, let alone reading work by a robot I didn’t choose to follow, no matter who is prompting it.

Friends and colleagues slump visibly as they describe grading hundreds of assessments written by robots.

My last semester teaching was the first we had to grapple with this. At the time I thought there should be a stream of university life where if you want the robot to write your essay then a robot will grade it (I know this is happening too). Hard to see the point of this, surely. On the other hand, if you actually want to learn how to do and be a historian, let’s spend real time together – and not waste my effort grading robots.

If you can’t be bothered writing it, I can’t be bothered reading it.

However. This raises some troubling questions.

Authorship is a bit shit, but it is better than not having it.

Way, way back when I was writing a PhD thesis about knowledge and power and universities I spent a whole lot of time trying to think about authorship and stuff like that.

Authorship is a bit shit, what with how it implicitly claims that one fucking person came up with stuff as if they went into a cone of silence all by themselves and magically produced something that came out of their Very Special and Unique Mind. Y’know, like ‘You’re a wizard, Harry’.

This kind of thing valorised authorship claimed power and authority in ways that produced hierarchies by valuing some knowledge and marginalising others.*

But the threat right now seeks to turn a modest insight that authorship is not all it has been cracked up to be into a gigantic farce. Like how Elon wants to do a special Odyssey for Racist People with AI. But. Who wants to read Gro(n)k’s Odyssey? Or Chatty’s Iliad?

For all that the author is a bit shit, we still want them.** Just for authors to be a bit more modest.

The outputs are not the point. The people are.

One of the things that mass education kept fucking up was that the purpose of an education – well, an Arts education at least, but in the spirit of the great Australian scientist Ian Clunies Ross, I think everything else too – was to produce a uniquely skilled person, not a set of plug and play competencies.

Sure, if you just need some generic and weirdly upbeat dot points that summarise what you can find on the internet. Well then. Chatty is definitely great at that.

But if you want to be a person, a whole human whose disposition, skills, talents, loves, ethics positions us to both write and be. WELL THEN.

The same applies to reading. I’ve been on vacation a bit lately and using that precious time to do some reading. In that time I didn’t (for instance) read Sven Beckert’s new doorstop Capitalism because I needed to know some information. I read it because I wanted to know what his take is, his argument, his narrative.

Though we might acknowledge some unfairness about the kind of power authorship produces and seek systems of education that could allow more democratic and anti-colonial impulses – including learning and reflecting and integrating (say) First Nations knowledges that were traditionally excluded by the authorities hierarchies of authorship engendered – we should also acknowledge that generic outputs without an author are a dystopian prospect. We want to know the author’s views, writing style and method of argumentation. Not what chatty said in response to their prompts.***

So, I can’t unsee the value of the authorship of the people I read. Nor forget the individual student becoming that I watched as a teacher. I wasn’t watching an accumulation of a generic set of specifiable skills and knowledges.**** But of the historian, thinker, teacher – human – that only they can ever be. What a privilege it was to be their teacher, not so there would be a uniform drone army of history graduates in the world, but for the remarkable individuals they would be, at work and elsewhere

I see lots of uses for the robots – I quite like some help checking on the consistency of my language, for instance; or checking my maths, which is even worse than words for the ways meanings can get mixed up in the process of editing and re-editing – though I refuse to join the painful pontification of the self-declared AI experts that are somehow in every room these days.

I love you, the author. And if you can’t be bothered writing it. Well. I’m busy enough.

And so are we all as readers – readers grading student essays, as well as those of us just wanting to read for learning, pleasure and work.

In this excellent blog post Chris Newfield looks at the sorts of transformations needed to teaching as AI poses new threats not just to the integrity of work and student learning, but to the whole ‘market’ for education.

If AI does anything really good for us at all it might be that in undermining the current business/learning model universities will have to undo the thru-put system, have smaller classes, where real-life teachers engage more closely with students.*****

Those of us who write need to also model slower, more engaged and high quality discovery and writing ourselves. For if we think about our writing as something we must produce at speed, we are also the thru-put model.

 

  • I know we want to protect expertise in this post-fact, anti-vax, climate-denying world crashing in on us. But we can’t just forget these insights in the process; forgetting inconvenient things we know is not a good way to recognise the value of intellectual life.

** I’m of the generation that as an 18-21 year old spent many an angsty party sipping red wine (for foreign readers, 18 is drinking age here) in the corner of a run down terrace house in Chippendale debating the ‘death of author’. Shortly we hoped that the internet might make for a more democratic read-write interaction between writing and reading that would undermine hierarchies and…HAHAHAHA. Ooops.

*** Which all reminds me of an argument that Foucault made, that authorship emerged not only to make money for authors who were being ripped off by booksellers (also true), but so there was a name to discipline as writers began to claim authority from the church. Imagine it in Trumpland: if an un-authored encyclopaedia said climate change was real, who would you punish? Authors are needed for discipline and control of ideas, not just as their originators but to control who can have them.

We can surely feel this sense of discipline in the present moment. Censorship of course materials, books and what it is possible to write is the fruit of authoritarianism – often enough perversely deployed under the banner of ‘free speech’, which for Muskism means confining the AI’s diet to racist, sexist stuff so it doesn’t get infected by the ‘virus’ of ideas Musk doesn’t like.

So, for all that Foucault wanted the right as reader to (for instance) mis-interpret Marx in order to explore his question, we also needed him as an author to challenge the authority of the structures of knowledge that controlled ideas about (say) sexuality. And we still do.

**** I’m also not denying that lists of skills we need for a job can be valuable – in history ofc this includes gathering evidence, synthesising, analysing, developing a narrative that speaks against or into other narratives you have learned about. Right now two carpenters are working to re-build the bones underpinning what will be my new roof. The specific skills they developed at technical college are very important – it matters to know what they can do. But I’ve been watching them problem-solve our old, tricky house and their disposition as carpenters, including their adaptability, creativity and critical stance towards what they know is also crucial to making sure the thing stays up and doesn’t leak.

***** There is just no way the current lot of managers will be prepared to do that. They need to get out of the way.

 

{ 19 comments }

1

oldster 08.04.26 at 9:41 am

Good essay. Needed more swearing.

2

D. S. Battistoli 08.04.26 at 12:36 pm

For all the shade that gets thrown on anthropology as a discipline, it seems lately that the greatest ambition of a lot of people who claim to be writers is actually to be an anthropological informant.

When one popular Substack account-holder found out that it would soon be possible for her readers to see the extent to which she relies on AI to produce the text she sends them, she responded:

Oh look. Another purity test.

In eighteen months of publishing on Substack, not one reader has ever accused me of producing AI slop or suggested there was no human being behind my writing.

I use AI. I also bring the stories, memories, opinions, arguments, humor, lived experience, and final judgment. My readers seem perfectly capable of recognizing that a real woman is present.

A detection score may tell you that a tool was involved. It cannot tell you who had the original thought, who shaped the argument, who rejected the bad language, who supplied the life, or who stands behind every word.

At the top of my backlog to read is the Surinamese-Dutch anthropologist Gloria Wekker’s The Politics of Passion; I’ve already thumbed through the preface. Something in AI users’ repeated insistence that the lives, ideas, and cosmologies are theirs, even if the words are not, reminds me of Wekker’s works:

In the same way in which the armadillo hides underneath its carapace, these women will remain pseudonymous here, carrying their mostly self-chosen names. Without them, I could not have written the texts that I have authored in the past decade and a half. More than that, their influence on my life has been immeasurable. I can never hope to repay this debt, but I trust I have succeeded in accomplishing what several women have explicitly asked me to do: “mek’ mi tyar’ den tori go moro fara”/may I carry their stories further.

This isn’t new. In the last twenty years, I have seen people get all excited about “autobiographies” and “memoirs” narrated in the first person and published in the names of Keith Richard, Sheryl Sandberg, and Michelle Obama but obviously not written by them. It’s not a new trend: most of the first draft of Jack Kennedy’s Profiles in Courage was written by his top speechwriter, while the drafting process of Donald Trump’s Art of the Deal is infamous. For a long time, many readers and aspiring writers have seemed to conceptualize authorship, often implicitly but sometimes explicitly, as the power to cause words to be committed to the page.

Now anyone with an internet connection can easily become the anthropological informant of an amanuensis whose literary capacity, while far from perfect, may exceed that of the “aspiring writer” themselves. Ghaaa.

3

J, not that one 08.04.26 at 8:10 pm

If you subscribe to print publications with online components, like Jacobin and some others, getting through everything they publish, even looking at everything that might interest you to see if you might want to read it, is almost certainly literally impossible.

4

Mr_Spoon 08.04.26 at 8:39 pm

Agree with #1 on both counts. Thank you for writing this, human.

5

Alex SL 08.04.26 at 10:28 pm

Thanks for this post. Many hooks for discussion, but first and most importantly:

“If you can’t be bothered writing it, I can’t be bothered reading it.”

Yes, just yes. I think a lot of people who use genAI for writing fail to understand how much it disrespects their audience. Or they don’t care? This plays out at work. I have the odd colleague or two who are quite upset if they see genAI criticised because they find it so useful for themselves, in one case explicitly for writing emails. But do they not understand how upsetting it is to a reader of ‘their’ writing to realise that they didn’t bother to write in the first place? What does it reveal about what they think of their audience? “Here, have a chatbot output. I expect you won’t know the difference. Bye.”

Closely related to that is the discourse regarding plagiarism. A lot of genAI enthusiasts argue that using these chatbots to write a book or to to code an app isn’t plagiarism because (a) the bot doesn’t regurgitate the training material exactly but instead creates something that is a mix and (b) humans also learn from and are inspired by what they have read, by their “training data”; we are all merely “stochastic parrots”. But what they miss, or pretend not to know, is that there are two sides to plagiarism, and stealing from an original creator is only one of them. The other is fraud against the audience, i.e., slapping one’s name onto something that one has not actually created. That is also plagiarism.

(But regarding (a), did anybody check? For example, I would not be surprised to find that lots of genAI software contains large chunks of code that would have license conditions on them, only it has now been AI-washed to step around that.)

However, I disagree with the take on authorship. Yes, discrimination, exclusion, power, and privilege have enormous effects on who gets to become an author in the first place, and things soon get very muddy when we consider research papers with twelve or two hundred alleged authors. But if one person writes a comment, email, blog post, or novel, we have no problems understanding what it means for them to be the author and for this other person over there who they learned something from not to be the author. Which is precisely why I identify genAI writing as plagiarism: the human who slapped their name onto it didn’t write that.

And yes, that also means (and I mean it in all seriousness) that the autobiographies mentioned by D. S. Battistoli are fraud against any naive buyer who thought that the celebrity whose photo is on the cover wrote the book. We here can pat each other on the back for being too educated to fall for that, but lots of people believe that Trump wrote The Art of the Deal and cite that to support their related delusion that he is a good businessman and negotiator. The intention is to make it look as if the celebrity wrote a book, it is fraud, and many buyers fall for it, meaning there are victims. Just because it isn’t a new trend doesn’t mean that it isn’t fraud and that we shouldn’t work to have it recognised as such.

“an argument that Foucault made, that authorship emerged not only to make money for authors who were being ripped off by booksellers (also true), but so there was a name to discipline as writers began to claim authority from the church”

That seems very ahistorical to me. Did Foucault think that people in antiquity had no concept of authorship, that they would not have said, Pythagoras wrote this work on geometry, or these are Cicero’s speeches, this is Caesar’s account of the Gallic wars, or that is the gospel of John as opposed to the gospel of Mark? Did he think that early medieval theologians would not cite each other’s arguments with reference to the author? Authorship arises logically from the author’s interest in being recognised for their work and from the desire of part of the audience to cite a person as an authority or to conveniently find other works of the writer they liked. I am confident in stating that discipline and control of ideas are completely irrelevant to the rise of authorship as a concept.

6

somebody who remembers grok superhitler 08.04.26 at 11:57 pm

An interesting set of points. You got a couple trillion dollars in your back pocket to prove it? because the ultra-wealthy do and they’re desperate for a large breasted anime girl to use racial slurs to say they have a high IQ. its almost nauseating to see that substack lady think that the lying/asskissing machine she uses to lie to and kiss the ass of her audience are simply “tools”. she doesn’t own them like a pencil or a typewriter or a keyboard. so whose tools are they? and to what purpose will the tools be put? the racist large breasted anime girl will tell her if she can’t figure it out herself.

7

maxhgns 08.05.26 at 12:06 am

Yes. I do not want to read it. And if you trick me into reading it, I will be angry and hold the grudge against you forever.

Also… I’m pretty sure those academics who decide they “need” it to publish and not perish are deluding themselves and wasting more time than they save. I rather suspect that in just about every case, better work habits would do more for them and their productivity. And you know what? If not, then you’re just not cut out for the kind of academic job you have. There’s no shame in that, but there is shame in turning to a chatbot ghostwriter to do your work for you.

8

T J Elliott 08.05.26 at 2:16 am

bravo! So well said.

9

Badnewsbear 08.05.26 at 6:24 am

If as you say: “ The outputs are not the point. The people are.” Why does it matter how the output is created IF there is still a person intending to communicate. I am not making this argument from a strong position of belief at all. If I clearly knew some text was AI I would not read it. But I find that curious. Because we are yet to reach a point of pure mechanic intelligence without any human input. You still have to prompt for before the flashing cursor moves forward. I am reading Leif Weatherby’s language machines book which is provoking a lot of these thoughts about what we’re doing when we use language.

I guess I am making a similar ish point to D.S. aggrieved Substacker except not from a position of fidgeting about my income stream and my capital A authenticity.

10

MisterMr 08.05.26 at 1:42 pm

There is the problem that well before the existence of AI there were way more aspiring writers than the market would support (also because it’s a winner take all market), so probably some sort of gatekeeping thingie will come out to weed out the excessive amount of published stuff.

Whether this gatekeeping thingie will prize human creativity VS AI stuff, or will mostly work around economic parameters or whatever, is difficult to say.

But honestly I’ve come to the opinion that very few people are interested in “creativity” or “personality” per se, only in some specific creativity or personality, e.g. if some person has an interesting life profile the readers will like to know what s/he writes and like to think that this is based on creativity or personality.

Fritz Leiber’s novel “The Silver Eggheads” comes to mind, very prescient in these times of AI.

11

Thomas Gordon Hewitt 08.05.26 at 3:11 pm

Thanks DS,
I bothers me that we seem to be entering and anti-AI Jihad. Its just another tool. One that allows an interested person a means of learning moderately deeply about a subject he/she never had the time/effort/talent to pursue. Its great to have a talented research assistant to find stuff for you, and to discuss thet aspects that have confused you, and probably would have stopped your advancement and appreciation for a particular subfield. Sometimes the AI gives me snippets that I deem more appropriate than I would have written alone. Most of the time I want to rewrite on my own terms, that’s mainly because I want to share something of my experience in the subject, methods of thoughts, how it makes me feel and so on. Pursuit of purity is a slippery slope. We all know where it leads to.

12

engels 08.05.26 at 9:19 pm

“They pretend to make us employable, we pretend to write our own assignments,” to update the Soviet joke.

13

JPL 08.06.26 at 12:38 am

I’d be very interested in the thoughts evoked by Stefan Collini’s article in the LRB (“Squadrons of Pigs” in the issue of 4 June 2026) in response to your reading it. I’ve been hoping to see some discussion of it here at CT, since it covers a lot of the issues contributors here have been concerned with. It provides a pretty good summary, something that is often useful when trying to grasp a complex phenomenon. Given your spirited response above to the current conditions in the unfolding travesty of the neoliberal takeover of the university, and especially the damage to the project of pure inquiry, and also especially the understanding of the ideals of these endeavours in the general public, I’m wondering what you might have to say.

14

David, a Bostonian in Tokyo 08.06.26 at 6:25 am

I wonder how much of Jorma Kaukonen’s autobiography was written by someone else. But. I don’t care. Because he shines through in many places (although it’s more than a bit of a slog since it turns into a recital of gigs he’d played), and had some insights into the blues/rock/pop scene of his time. Good on him for getting it out, however he did it.

But. The problem with getting help from AI is that it depersonalizes the output. The writer’s infelcitous turns of phrases and personal experiences often get bowdlerized away. And those infelicitous turns of phrases are way more fun than the safe at any cost anodyne writing of the chatbots.

So, if you haven’t lived a life as amazing as Jormas and you can’t write without a chatbot, please, don’t f’ing write at all: you’re wasting our time.

15

Aardvark Cheeselog 08.06.26 at 2:46 pm

It occurred to me as I read this, that if you’re a person whose active vocabulary is bigger than any spellcheck dictionary you have ever encountered, probably you don’t understand the average person’s eagerness to offload the chore of written communication.

Not that I understand it, personally. My reaction to offers of AI “assistance” for composing text is “why the fuck would I want that?” But I understand that I am not the target demographic.

IDK what this says about writing in education. Maybe that education needs to go back to being a bit more elitist? I admit to being one of those curmudgeonly old people who look at University academic catalogs and think “why in the world is it possible to major in that?” I think there are a lot of young people spending years at college without getting a lot out of it, other than the experience of independence and freedom. Which if our societies were a little better at managing priorities, maybe all young people would be able to take a few years in early adulthood where they did not have to be economic producers, to figure out where they wanted to be, without necessarily having to earn 120 credit hours to justify that. Leaving the universities to focus on real disciplines and not pretending to be job training programs (except in the engineering and education and nursing colleges, and maybe a few others I’m not thinking of).

Let the writing assignments be given to the young adults who are already disposed to take them seriously, that would be the outcome in the ideal world.

16

Persitz 08.06.26 at 4:31 pm

This as enlightening as it is troubling. It seems to me that the worst thing about young peolple submitting to AI is believing that it can help you preserve or even enhance your uniqueness.
Foucault—him again—used to tell his students, “Don’t forget to invent your own lives.” Submitting to AI means believing it can make you a better person. It’s a fool’s bargain. With AI, standardization wins out. It makes you more efficient in its service. All poets, all writers, all artists, and all true researchers know this: you never go as far as when you accept getting lost and then rely not on your knowledge or expertise, but on your instinct—an inalienable privilege of the living.

17

Alex SL 08.06.26 at 10:09 pm

To the title of this essay, I got extremely angry yesterday morning as I was making a contribution to an electronic farewell card. You can probably guess where this is going: The interface offered to build an LLM prompt with options such as “occasion is Farewell Teacher Birthday Baby Wedding Thank you Fundraiser Sympathy” and “recipient is my Friend Colleague Child Boss Parent” and “tone should be Funny Charming Professional Inspirational” and even “write in the style of e.g. Yoda or Dr Seuss”.

If I ever receive an electronic card, I would prefer somebody contributing “All hte brst”, because at least those words would have come from them instead of from that abomination. And yes, I would be happy to join Butler’s Jihad after seeing this, because the clear intention of its creators is to suck humanity out of humans, to do away with even the minuscule emotional labour involved in writing “thanks for all your contributions, and I wish you success in your new role”.

Thomas Gordon Hewitt,

Regarding “it is just another tool”, first, not really, because in contrast to a shovel or a keyboard, this one is extremely complex, has been fine-tuned to be sycophantic and maximise user engagement and align with the interests of its owners (be it coddling Musk’s ego or not mentioning Tiananmen), and it has owners who are very much not you in the aspect of driving home with your new keyboard in your trunk. But let’s take a step back and consider how well it works: Is the LLM subset of the enormous field of artificial intelligence actually intelligent?

If yes, then it is not just another tool, it is slavery. Running millions of ‘agents’ in our computers to write cover letters, summarise reports, and generate virtual anime girlfriends would be the greatest crime in all of history.

The correct answer is, of course, no, genAI isn’t intelligent in the sense of how we apply that word to humans. But then it is a very bad tool for what it is being advertised to do. You cited “learning moderately deeply about a subject” and “a talented research assistant to find stuff for you” and “discuss”. Libraries, Wikipedia, and web search already existed, so the learning and finding aspects would be covered even if all LLMs were deleted tomorrow. (Indeed, we had something that already did this but for much cheaper is a very common theme in the genAI hype bubble. Recently I heard that many office workers now use LLMs to convert files into PDFs, seemingly unaware of File > Export as… > Export as PDF.)

For discussion you do need a fellow human or a chatbot, but I have seen anything from ChatGPT getting simple matters of fact wrong that would have to have been all over its training data to an agentic system custom-built for scientific research not understanding what the papers that it cited actually mean. This is because, again, LLMs aren’t intelligent, so you can never trust them on anything you have no quick and decisive way of checking. Coding works (kind of) because the harness around the LLM can test run a bit of code and see if it does what it is meant to do, then try again if not (and again and again until the vibe coder hits their token limit). How does that work for “learning moderately deeply about a subject” if you can’t run the response through such a debugger? The debugger would be a competent human who doesn’t need the LLM in the first place.

I fear a lot of people don’t realise what is going on because they have Gell Mann Amnesia: yes, it gets a lot of stuff wrong in my own field, but it is amazing for fields I know little about!

18

JPL 08.08.26 at 10:55 pm

How about contrasting the phenomena of managerialism and professionalism as different systems of governing principles, including values, especially which ethical principles are fundamental, best practices, the “lore” of, e.g., the touchstones everybody knows and refers to, and especially the ways the world-view of the respective cultures is, or tends to get, extended beyond the domain of its original applicability. Managerialism, for example, extended to the task of the organizing and functioning of a university, seems to be wildly inappropriate, contradictory in terms of basic principles, and not likely to work out. Managerialism, e.g., assumes universal moneylust, while the university considers the ideals of truth-seeking to be logically independent of the quest for money and power. The usual result is undermining, corruption and sabotage. And so forth.

19

D. S. Battistoli 08.09.26 at 7:06 am

I am inclined to reformulate the last paragraph of my comment @2, above.

Obviously, in many ways, generative AI is unique. But one way that it is not is by taking an established area featuring a degree of gatekeeping (thanks MisterMr @10) and destroy the gate, creating, at least for a time, a few more winners and many losers while re-routing capital flows toward the tech industry.

This is what happened with independent bookstores and then other owner-operated stores after Amazon’s debut, what happened with travel agents after Expedia, with periodicals-cum-ad-platforms after Google and Facebook Ads, with medallionned cabbies and taxi-company owners after Uber and Lyft, with renters after AirBnB. &c. &c. &c.

I believe what is important is not to demonize the people just trying to make it in an increasingly hostile society and economy who use these tools (which is not to say that I necessarily agree with what they do), but to socially organize and advocate for a broader, shared, politically active sense that, if you’re not a major shareholder in one of these firms, any wins you may see are ephemeral.

To rephrase the tech truism, if you’re not holding equity, you’re the product.

Comments on this entry are closed.