As you may recall, I have written from time to time about the effects of artificial intelligence, particularly large language models like ChatGPT and others, on the interpretation and heritage tourism sector. It has been a minute since I wrote on this subject, so… let’s jump in.
As I was out walking today I started to think about art museums and how they interpret their artworks. A classic method has long been the digital audio tour. In days of old you rented a piece of hardware with headphones; nowadays you pay a modest fee and get access to a curator’s well-researched thoughts through your own device.
As an interpretive writer and consultant, I write those things. Well, I haven’t done an audio guide in a while, but I do make a substantial chunk of my income from writing interpretive text destined for public eyes and ears.
So I wanted to see how close I am to being functionally obsolete in that field. I crafted a prompt for ChatGPT.
If I upload a pic of a museum artwork label, can you create some interpretation in the style of a stop on an official museum audio tour? As if I was standing there looking at it. About the work, why it’s important, about the artist, any movement or school the work belongs to? Accuracy and tone are important.
Chat GPT immediately replied, SURE I can do that for you. It also asked for more details: what tone of authority I’d like to hear; it asked me to upload an image of the artwork along with the label; and a few other things. But, reader, I offered it none of those things. Just the museum’s name and the label. Witness:
MOMA.
It took Chat GPT exactly five seconds to produce the following. I will reproduce the artwork for you here so you can have the museum-like experience; I did not show Chat the image.
Audio stop: Georges Braque, Soda, 1912
You’re standing in front of a work created at one of the most radical moments in the history of modern art.
In 1912, French artist Georges Braque was working closely with Pablo Picasso to develop what became known as Cubism — a movement that challenged centuries of European painting tradition. Instead of showing a subject from a single viewpoint, Cubist artists broke objects apart and reassembled them from multiple perspectives at once.
At first glance, Soda may feel fragmented or difficult to read. But Braque is not trying to confuse the viewer. He is asking a new question: how do we actually experience objects in space and time?
Look carefully and you may begin to recognize hints of café culture — perhaps parts of a soda bottle, lettering, glassware, or a tabletop. Braque often drew inspiration from everyday Parisian life: cafés, newspapers, musical instruments, bottles, pipes. Rather than painting these objects realistically, he compresses and overlaps them into shifting planes of muted colour and geometric form.
This painting belongs to a phase often called Analytic Cubism. During this period, Braque and Picasso reduced colour dramatically — using browns, greys, blacks, and ochres — so that viewers would focus on structure rather than surface beauty.
There’s also something deeply modern happening here. Europe in 1912 was experiencing rapid change: photography, cinema, industrialization, urban speed. Traditional painting no longer seemed capable of expressing this fractured, fast-moving world. Cubism became one of the first major artistic languages of modernity.
Braque himself was quieter and perhaps less mythologized than Picasso, but many art historians see him as equally essential to Cubism’s development. He brought a strong sense of balance, rhythm, and material sensitivity to the movement. In fact, Braque would soon begin experimenting with collage — gluing printed paper directly onto canvases — helping to open the door to much of twentieth-century modern art.
So while Soda may appear abstract, it is also very grounded in ordinary life. A drink. A café table. A fleeting moment in Paris. Braque transforms those familiar fragments into an entirely new way of seeing.
Reader, once again, I am shook.
The facts and the interpretation overall seem to pass the smell test, though I’m not an art historian. The tone is curatorial yet accessible. Interesting. I’d listen to it in the gallery.
Now, I’m enough of an art museum maven to tell you that this isn’t cutting edge interpretation; part of what makes this uncanny is that museum audio guides themselves are often written in a broadly accessible, safe institutional voice. AI happens to be very good at reproducing that voice.
And really, this is a very famous work. Generating a plausible overview of a foundational Cubist painting is probably low-hanging fruit for a large language model.
Let’s try something less canonical, and considerably more difficult. Attila Richard Lukacs is a Calgary-born artist that I have followed for some time.
Gibson Gallery: MCLEAN / COPELAND COLLECTION
Attila Richard Lukacs – S.H.A.R.P. Lacquer on Wooden Nesting Dolls Original Prototype, Made in Moscow & Berlin Various Sizes ranging from 8 1/2 x 5 x 5 in.
It took exactly five seconds again.
(Reader, I’m not going to reproduce the image here but you can find it online at this link. Content warning: Lukacs’ work can be graphic, politically charged, and… troubling, for reasons that the AI audio stop highlights below.)
Audio stop: Attila Richard Lukacs, S.H.A.R.P.
At first glance, these objects may feel almost playful.
They’re Russian nesting dolls — familiar souvenirs associated with folk craft, childhood, and cultural nostalgia. But Attila Richard Lukacs uses them to explore something far darker and more confrontational.
The title, S.H.A.R.P., refers to “Skinheads Against Racial Prejudice,” an anti-racist movement that emerged within skinhead culture in the late twentieth century. Lukacs became known in the 1980s and 1990s for paintings and installations exploring masculinity, extremism, violence, homoeroticism, youth subcultures, and political symbolism. His work often unsettles viewers because it refuses easy moral distance. Rather than presenting ideology as something abstract or historical, he places us uncomfortably close to its imagery, seduction, and contradictions.
Here, the nesting dolls become loaded symbols. Traditionally, matryoshka dolls contain smaller and smaller versions of themselves — identities nested inside identities. Lukacs transforms that familiar structure into something psychologically and politically charged. The figures may suggest performance, tribal identity, conformity, or the way ideologies reproduce themselves socially and culturally.
The fact that these prototypes were made in Moscow and Berlin is also significant. Both cities carry enormous historical weight in relation to nationalism, authoritarianism, Cold War memory, and contested identities in Europe. Lukacs frequently worked internationally and drew from transnational youth culture, punk aesthetics, and political iconography.
Notice the tension between craft and content. The lacquered surfaces are meticulous, seductive, even beautiful. Yet the imagery resists comfort. That collision — attraction mixed with unease — is central to Lukacs’s practice.
Importantly, Lukacs is not simply illustrating political positions. His work tends to investigate ambiguity itself: how symbols acquire power, how identity becomes theatrical, and how aesthetics can both attract and disturb.
Standing here, you’re not just looking at decorative objects. You’re looking at a meditation on ideology, performance, masculinity, and the unstable layers inside cultural identity — fittingly embedded within objects that literally contain versions of themselves within themselves.
Well, then.
Not to centre MYSELF in all of this, but as someone who earns part of his living writing interpretation, this is the first time I’ve felt genuinely uncertain about where some of that work is headed.
But there’s more. If I used the ChatGPT app, I can click on the ‘Listen’ button and a pleasant and very human sounding voice will read the above text in my ear while I stand in front of the piece. So it’s not just interpretive writers who can start to sweat here, but museum media producers in general.
And that’s not all.
While the curatorial interpretation is interesting and relevant to me, I have no idea whose thought or scholarship the AI is using here. Is it synthesizing a consensus scholarship, or is it appropriating a particular authority’s work? I have no idea.
And while I am all about diversifying the museum voice and taking hegemonic power from single-script, single-authority museum approaches… I’m just not sure algorithmically generated interpretation is the way to go. Thinking about it, a lot of institutional interpretation already follows familiar conventions: simple contextual framing, accessible tone, concise thematic storytelling, broad consensus scholarship. This AI approach doesn’t help. Large language models are frighteningly good at speaking in exactly that safe institutional register. They are weak-to-useless at original curatorial thinking, unique institutional positioning, emotional risk, and controversial/alternative interpretation. You know, the stuff we really need.
Lastly, museums make money from this stuff.
If I want to access a curator’s detailed thoughts in an audio tour, I usually have to pony up a few dollars to do so, or perhaps invest in a membership to the museum. This do-it-yourself audio tour thing pretty much does an end run on the museum’s revenue stream.
All that to say… what a time to be alive. What a time indeed.
What are your thoughts?
Hey does this kind of writing tickle your brain? Sign up for monthly goodness from donenright.com.
Wanna see what past newsletters look like? Clickity click.
Leave a Reply