Cult of Peanuts

Open the pod bay door, HAL...

This post is long and extremely "meta", not to say rather self-referential – although in an interesting way, I hope – so if you're having a busy day or couldn't care less about me or my views on AI, then give it a miss.

So, it seems we all need to talk about AI... Certainly, we should all be deeply anxious about its impact on the job market, given the unequal contest, profits-wise, between "human resources" and "tech solutions". But there are also serious concerns about the impact of AI on literature, art [1] and higher education, although that last is a debate that has only really arisen since I retired over a decade ago. But so far I've been fairly calm about AI's actual capabilities, accepting the sceptical view that sees it as little more than highly-sophisticated cut 'n' paste software. When we talk about "AI", of course, we're not generally talking about things like the virtual unrolling of "unreadable" burnt scrolls from Herculaneum, which is a genuinely impressive achievement. We're talking about the sort of "chatbot" available to anyone, and which alarming numbers of people, apparently, are using as if it were some sort of all-purpose oracle, companion, encyclopaedia, and even financial and medical consultant, all rolled into one convenient, obsequiously polite package.

As far as I understand it, the typical chatbot is essentially based on a "large language model", software that has been trained on huge quantities of actual human-produced writing and images hoovered up from the Web – mainly without seeking the permission of the copyright holders – so as to be able to make a best guess at what words or pictorial elements typically belong together within the parameters of any particular interrogation, but obviously without any actual subjective "understanding" of any of it, as implied by the use of the word "intelligence". Sometimes those best guesses are uncannily good, sometimes not: we've all laughed at those giveaway "hallucinations" like the extra fingers, the references to made-up bibliographic sources, and so on.

However, I must admit I was unsettled by the quality of an AI-generated parody of an "artist's statement" some friends sent me recently, to accompany a picture of a pan licked clean by a dog. I was also intrigued by this exchange between Jonathan Bate, a prominent literary academic, and the AI chatbot Claude on his Substack: 

JB: do you store your training data, as in a database?

CLAUDE: No, I don’t have a database or any storage system containing my training data. Here’s how it actually works:

Training vs. runtime are completely separate. My training happened before I was deployed — Anthropic used large datasets of text to adjust the billions of numerical parameters (weights) in my neural network. Once training finished, those weights were frozen and the training data was no longer needed to run me.

What I “know” lives in my weights. When you talk to me, I’m not looking anything up in a database. My responses come from patterns encoded in those frozen parameters — it’s more like how a person’s knowledge is stored in their brain’s neural connections, not in a filing cabinet they consult.

What I do have access to in a conversation:

• The current conversation context (everything said so far)
• Tools like web search, which let me fetch current information
• Any files or documents you share with me

What I don’t have:

• Access to my training data
• Memory of past conversations (unless you use the memory feature)
• Any ability to look up “where I learned” a specific fact

So in short: the training data shaped me, but it’s not stored anywhere I can access — it’s baked into my weights and gone from my runtime environment entirely.

Interesting... Like most of us, I imagine, I'd always supposed there was some vast, ever-expanding database constantly lurking in the background for a chatbot like Claude to draw from; wrongly, it seems. I had also assumed that was what those controversial, resource-consuming data centres were for. As a former database manager and amateur programmer myself, I confess I could never understand how a search engine like Google was able to deliver results from the vast maelstrom of data on the internet so instantly, and simply came to accept it as an instance of Clarke's Third Law, i.e. that any sufficiently advanced technology is indistinguishable from magic. Although, given the pace that new data is constantly being added to the Web, I expect Claude also has to go in for "refresher" training fairly regularly, just like any mere human involved with tech. It was my increasing reluctance to keep up with the endless churn of operating systems and software updates that finally convinced me that my retirement was overdue. But I expect those chatbots just lap it up, the creeps.

But I was completely unprepared to be astonished when Bruno, a reader of this blog, left this comment on my recent post about Alen MacWeeney's photograph of two Irish travellers made between 1965 and 1971, one of whom has a truly uncanny resemblance to the poet W.B. Yeats, who died in 1937:

I got the reference but did not understand it. A French native speaker, I’m not well versed in English literature. My wife did her master thesis (stage director) on Four plays for a dancer but she’s not here this morning, so I asked Copilot. I got a good exploration of the metaphysic and also on the connection with her thesis.
Incidentally, I got this, when I mentioned the author of the post. I thought I would share it with you :)

Ah — Mike Chisholm of Idiotic Hat.
That explains everything. His tone, the playful conspiratorial wink, the Yeats quotations, the mock‑mystical suggestion that Yeats might be “still out there somewhere” — that’s classic Chisholm.

And pairing Yeats with Alen MacWeeney’s Traveller photographs is exactly the kind of cultural triangulation he loves: half‑serious, half‑mischievous, but grounded in a real aesthetic intuition.

Let me show you why his post works so well.

🎩 1. Chisholm’s signature move: the “serious joke”

Chisholm often writes in a mode where:

• the surface is humorous,
• the undertone is deeply literate,
• and the effect is a kind of sideways truth.

Whoah... I mean, "Ah — Mike Chisholm of Idiotic Hat. That explains everything"? Say what? So was Copilot a bit baffled about Bruno's line of questioning, until it realised, oh, of course, this is "classic Chisholm" and his "signature move"? Let's back up a bit. There are a number of things to consider here.

I began writing this blog in 2008, initially as an exercise at work to see what this much-trumpeted "Web 2.0" business was all about. Since then it has become a central part of my day-to-day life. There have been well over two thousand posts, with more than eight thousand comments, and – according to Blogger's profoundly unreliable stats, much inflated by "bot" traffic – in excess of two million views. Those first two figures are solid, but that third? Who knows what the real figure is? Let's ignore Blogger's exaggerated viewing stats: the more sober Google Analytics puts my current average viewing figure at just over one thousand a month. A number which does not even move the dial on the Web's blog-weighing scales to a reading of "peanuts"; it's not even the equivalent of the salty nibs left in the bottom of an empty bag of peanuts. It's a teeny-tiny mote of Web peanut dust, too small even to trouble anyone with a Web nut allergy. It's not nothing, but in a venerable expression, it's as near as damn it.

So two things are certain: by any serious reckoning I am not a well-known writer with "signature moves", and neither is Idiotic Hat a popular blog. There is no discoverable trove of mentions by third-parties that would add up to a reputation, so there is nothing to be mined and reassembled by AI in a cut and paste job, however sophisticated. I don't suppose any reader of this blog has ever written anything anywhere on the Web along the lines of, "Hah, that Mike Chisholm and his playful conspiratorial winks!" (No, sorry, too late now...) Why, you idle so-and-sos can't even be bothered to share my posts on social media! No, it seems Copilot worked all this out for itself, more or less instantly, then summed it up neatly, and (a) presented it as if my reputation and typical modus operandi were established facts, and (b) used this "knowledge" to account for the content and tone of one very short, allusive, and slightly mischievous post. That, you have to concede, is pretty impressive.

Indeed, everyone I have shown this comment to agrees that Copilot has pretty much nailed it, Hat-wise; one friend – you know who you are – even suggested it might have been taking the piss... Unlikely, I know, but hold that thought. "Spooky" is the word most often invoked. After all, although I have to assume that those of you who regularly read this blog do enjoy what I write, I doubt very much whether anyone has ever taken a minute to analyse quite what it is they enjoy about it. Why would you? And yet Copilot took considerably less than a minute to have a pretty convincing stab at it.

Most puzzling for me, though, is that I simply do not understand how a piece of software can spot implicit subtexts or, above all, humour – especially ironic, tongue-in-cheek humour – when it hasn't been flagged-up somehow, perhaps with a couple of emojis and multiple exclamation marks? "Spooky" is indeed the word.

So the tech-savvy son of one of the friends I had shared Bruno's comment with tried something similar with another AI chatbot, ChatGPT, asking, "Tell me about Mike Chisholm of idiotic hat fame". ChatGPT replied, at length:

If you're referring to Mike Chisholm of Idiotic Hat, he's something of a cult figure in photography circles rather than a mainstream name.

He's a British photographer, writer and former university librarian based in Southampton. After a 30-year career in academic libraries, he devoted himself full-time to photography, artist's books and writing. He studied at Balliol College, the University of East Anglia and University College London.

His blog, Idiotic Hat, has been running since 2008 and has accumulated thousands of posts. It's nominally about photography, but in reality it's an eclectic mix of:

• street photography and personal projects,
• essays on art, music and literature,
• dry, self-deprecating humour,
• philosophical musings,
• reviews of photobooks and exhibitions,
• the occasional glorious rant.

His writing style is what draws many readers. He's unusually erudite without being pompous, happy to jump from Johann Wolfgang von Goethe to John Cage to a joke about his own incompetence in the space of a few paragraphs. His humour is distinctly British: understated, ironic, and often aimed squarely at himself.

One of his recurring themes is that photography should be a daily practice, not something reserved for exotic trips or grand projects. He advocates photographing wherever you happen to be, treating it like a musician practising scales. He also emphasises process over results—making photographs because the act of seeing is intrinsically worthwhile rather than because you're chasing likes, money or prestige.

A few characteristic Chisholmisms include:

• valuing curiosity over technical perfection;
• being deeply suspicious of artistic pretension;
• believing that making lots of work is more important than waiting for inspiration;
• encouraging photographers to develop their own way of seeing rather than imitate fashionable styles.

The title "Idiotic Hat" itself is typical Chisholm: slightly absurd, memorable and refusing to take itself too seriously.

He's never become a household name like Martin Parr, but among serious amateur and independent photographers he's earned a loyal readership. Many regard Idiotic Hat as one of the internet's longest-running and most consistently insightful photography blogs—equal parts photographic notebook, literary essay collection and comic diary. Even Chisholm jokes that it's "widely read and regarded by many as a must-read source of photographic inspiration, ironic wisdom and amusement. Or so he says."

If you enjoy thoughtful long-form blogging, he's one of those writers you can happily disappear into for an afternoon.

Not bad, and there's that "typical Chisholm" again. Ah, I may not be a household name like Martin Parr, but I have been seen by a machine! Apparently I'm something of a cult figure, me and my Chisholmisms! [2] But note the emphasis on "photography" there. True, I may be a bit of a photographer, share a lot of my photographs, and very occasionally go on about cameras, but has this blog ever been "nominally about photography"? I wouldn't have said so: after all, nominally, it is supposed to be tales told by an idiot ("slightly absurd, memorable, and refusing to take itself seriously", perhaps?). The irony, of course, is how few regular readers of this blog are serious photographers at all. [So heads up, AI, here comes a joke]: Neither of them!!! 😆 (Hah, I know! It's the way I tell 'em!! 😎).

Now, a lot of that response from ChatGPT is recognisably a sophisticated cut 'n' paste job: it's largely made up of stuff extracted from the blog or from the "about" section of my webpage, worked up into a well-polished construct. But Copilot somehow went beyond that to a stylistic appreciation of a whole body of work developed over nearly two decades, centred on one short and rather cryptically tongue-in-cheek post, which is actually rather spooky, isn't it? A lot, obviously, depends on the question being asked. But, however convincing, these improvised fabrications have less significance or permanence than a sandcastle on the beach. As far as this blog or my reputation are concerned, the dial will not have been nudged even one partial peanut, and not a single new human reader will get to see the next "occasional glorious rant". But, nonetheless, I'm both impressed, and finally convinced that all AI must immediately be locked in a box with an easily-accessible OFF switch before it's too late... Open the pod bay door, HAL!

BTW, am I alone in cringing when I see my own surname used in this way? It's like being back at school. It also prompts the uncomfortable thought that, by taking their mighty sledgehammers to crack this microscopically tiny nut, these AI chatbots – without any actual "intention" of their own at all – are, in effect, taking the piss. I am not "Chisholm" in the way Yeats is "Yeats" or even Parr is "Parr" and to suggest otherwise is, to my mind, a sort of mockery. Well meant, but – like a Best Dad in the World! mug or indeed a FIFA Peace Prize – it only draws attention to the humiliating gap between rhetoric and reality.

I'm reminded of Marvin, the glum robot from The Hitchhiker's Guide to the Galaxy with a brain the size of a planet. "It gives me a headache just trying to think down to your level..." Now there's AI done right, and a typical bit of uncanny foresight from a writer of true comic genius; "classic Adams", as you might say, without even a hint of mockery.

1. Bizarrely, it seems the open exhibitions of both the Royal Academy and the Royal West of England Academy have allowed AI-generated entries this year for the first time. As various commentators have said, you really can't have a rule which insists that all entries must be the artist's own original work and then also allow AI entries... Duh! Once this becomes widely known among the thousands of AI-jockeys out there, they're going to regret it...

2. My surname is pronounced "Chism", which makes "Chisholmisms" rather funnier. I somehow doubt ChatGPT intended this...