When a friend posted up the image of this wonderful, daft costume from this year’s Karl-Lagerfeld-themed Met Gala, I thought three things very quickly:
1) Oh. It’s that horrible twat Jared Leto inside. Shame.
2) But that outfit is absolutely brilliant and hilarious. What a delightful joke.
3) Hang on. AI. I bet that never actually happened. I’ll have to Google it to make sure.
And by the time I’d Googled it and found out that, yes, it had happened, and a group of brilliant costume designers at a company called SCPS had taken a break from doing Marvel and Star Wars movies to make a disturbingly accurate wearable version of Lagerfeld’s cat Choupette for the creepy Dolmio-accented House of Gucci pillock to swan about in, I’d completely stopped enjoying it.
And whose fault was that? You guessed it. Who else? The Pope.
‘The Pope’ had briefly gone viral in an unexpected get-up which then turned out to be an AI generated image. Looking at it now, I can’t help feeling that his Holiness’ plump white reinvention was one that a couple of years ago AI image recognition software would probably have tagged as the same image as the kitten costume. They share a cuddly Stay-Puftness that makes both pictures incredibly appealing. They both feel like a real-life cartoon; some crazy Kawai shit dropped into our real world, that makes you double-take, and then try and work out what the hell’s happening.
With the kitten, it works perfectly. Something crazy did happen, in the real world, and it was a huge effort, and a huge surprise, and a delight. Aren’t humans strange and baffling and brilliant? I love us. I think I can work out what happened, and why, and it’s a joy.
But with the Puffa Pope, there’s a glitch. Because it didn’t happen. The Pope didn’t make a nuts wardrobe decision you have to try and read. Nobody made him a coat. Nobody even spent ages cutting up pictures and rescaling and matching the tones and shadows, in order to waste their time on a daft gag.
Unlike the cat image, the Pope image contains no implications, and no juice. Someone thought of it, and then a machine quickly did it, and somehow because the making of the joke lacks the implied man-hours that might make you think ‘well, it has to mean something or they wouldn’t have bothered’, there’s no weight behind the gag, no behind-the-scenes story. It’s gossamer light. It’s not even got the bulk of that coat.
The advent of AI is already starting to damage the speed at which we can play the basic human game of trying out various sense templates to parse and process apparent nonsense. Adding in the possibility that something was seamlessly made at the stroke of a key, with no human intervention, or graft, purely to fool the eye, affects how we read the intent of a joke.
Because I needed to know the dumb cat suit had been a massive waste of someone’s time for it to be properly funny.
It might seem strange, but the wasting of the creator’s time is crucial to comedy like this. If nobody wasted very much time making that joke, then they just wasted yours. In joke terms, at best, it’s a Nelson Muntz gag, and the ‘ha ha’ has a victim: us.
With AI art, we are not invited to wonder at the effort that a joker put in to making something elaborate and ridiculous, the time they wasted on their own folly. The joke isn’t on the fool who made the costume, or the fool who wore it,. The joke with any AI image that escapes its context and passes as real, or human-made, is on us.
Even if someone explains it, and it’s contextualised as ‘Twenty great images of Wes Anderson’s Star Wars’, or ‘The Coronation but done as a working class wedding’, the wonder and richness of decoding the joke is fleeting because there’s nothing more to discover beyond the title strap. We know what to expect, otherwise we wouldn’t be able to ‘read’ the mash-up, and there’s no fun in imagining someone painstakingly painting the parts of the image, making complicated decisions, deploying their skills. The joy lasts as long as the prompt took to type.
It’s the difference between this marvellous piece of art by illustrator Chris Barker, imagining Jack Vetriano’s version of a classic Victoria Wood sketch:
And this bit of AI responding to a prompt requesting Wes Anderson’s Star Wars:
And that’s the difference between the Met Gala cat costume (real) and the Puffa Pope (virtual).
Even though the juxtapositions in these examples might seem identical – a scene by one creator in the style of another, or somebody unexpected in a mad outfit – there’s no wonder, or joy, or comic depth in AI generated gags like the Pope coat or the Wes Star Wars, beyond appreciating the initial idea, and then witnessing a data handling technique that very few laypeople remotely comprehend. There’s no space to find joy in being fooled, or laughing at imagining the long, long hours which the joker ‘wasted’ on a stupid idea. It’s like CGI stuntwork compared to a 70s car chase. It’s hard to care.
Both images – cat costume and puffa Pope – could be created by AI in seconds. Just type in what you want. But in contrast to a computer generated image of a hip hop pope, the giant swanky-do cat contains shitloads of tasty human comic data that we can analyse and play with, and delight in. Who did this? Why? What were they thinking? Was it hot in there? Where’s the zip when he needs a piss?
With a hand-made production number like the cat, we can read hidden character, and story. We can imagine the sort of person who would commission the costume, how much they laughed thinking of it, the hard work the craftsmen put in, the times they must have tried it on and cracked up laughing. It’s a rich game that we’re joining in with them. If we delight in the image, the hard work behind it deepens that pleasure. We are in a little gang, together, giggling.
The effort that was put in to do something like this, something so pointless feels thrillingly extravagant. Silly. A total waste of time. It’s a celebration of safety, of there being time to make art, to craft, that we are not under threat, or short of time or resources, just like a ballet or the Sistine Chapel ceiling. It’s what art is for. For us to take a chance to look and wonder: what in god’s name were they thinking? That must have taken ages!
It’s really, really funny.
I know lots of people who make a living or a reputation – or just get enjoyment – from skilfully making fake images of impossible funny things, and all of them have been depressed by how easy AI art generation makes creating these sort of mash-ups and cultural juxtapositions. And they’re right to feel threatened, because it is ruining not only their chances of impressing us with their hard-earned Photoshop skills, but also our fun as an audience in consuming them. Someone has taken the grind out of something where it was the pointlessnes of the grind that made the joke funny.
Here’s one done by my brother, by hand, done just to amuse himself and us:
That probably took a bit of time. Not forever. It’s just a cut up. You can see the joins. But it’s a lovely idea, well executed. That’s what’s funny about it. That it sort of wasn’t worth the effort. But because thinking of combining the worlds of Emu and Black Sabbath is a funny idea, and then not tossing the idea aside, but laboriously doing it is a funny thing for a person to do with their time, that makes the wasted effort not a wasted effort. Coded into that image is not just the warm recognition of shared cultural signifiers, but also the profligate joy of the time wasted.
Automating that process with AI so that generating Grotbags’ debut doom metal album takes seconds doesn’t take the drudgery out of comedy. It takes the comedy out of comedy. In the name of making comedy more efficient, AI sprays defoliant gas into all the creative crevices where extra jokes might grow.
It’s not just pop culture image mash ups. If you think of the things that have made people laugh the most, ever, they’re very often enormous, wasteful, profligate production numbers. They might involve huge numbers of people who’ve wasted ages learning a song about spunk.
Or a production number that was obviously a terrible idea all along, but, hey, they went the whole ‘nein’ yards.
Or a dignified public figure who’s put the hours in rehearsing a special skill for no explicable reason but to surprise us.
And how brilliant is it that a professional crew wasted days building and rigging a 1950s cafe for someone dressed as a Teddy Boy to sing Mr Boombastic on a Kirby Wire, and for that to be on screen for a matter of a minute?
Or that a huge all-star choir (including Indiana Fucking Jones) gathered to sing a song based on a half-serviceable callback shock gag about how a late night US talk show host might be fucking another A-list Hollywood star. I mean, I know there was a writers’ strike on at the time, but every cutaway shot in this makes me imagine another really complicated day of pointless phone calls to agents, and that makes it funnier and funnier.
AI art is always doing this Jimmy Kimmel style gag: generating unlikely casts, dropping stars in the wrong places. And that destroys a really basic gag that we humans really enjoy: guest stars. We really applaud when they come on. We love them being game for a laugh. That’s our favourite joke. A status gag. Kings as clowns. That’s why we’re asking the machines to make it easier, to produce more, to give us more of that sweet juxtapositional star sugar.
But I can’t stress this enough: we automate that process at our peril, because we kill its potential to make us wonder and laugh.
Star mash-ups are why most Comic Relief or charity telethon sketches really work – that you can’t believe somebody went to the effort of booking the actual (insert star name here) to do (stupid thing). Create the footage by typing in the stars’ names to an AI that generates an identical viral video, seamlessly achieved without any backstage graft or negotiation, and I don’t care, it’s shit. It’s not funny. There’s nothing to unpack, and so I’m not laughing.
It’s not just an argument that artisanal work is better than factory product. It’s that we read so much more into jokes where someone has wasted loads and loads of time. Who are these maniacs? What were they thinking? How did this happen? It turns a surprise tea-splutter reaction at the basic idea, into a rolling laugh as the implications sink in. It’s rich and human and funny. It’s the comedy version of gasping at the effort and craft put into a big stage show or an elaborate magic trick.
That’s the difference between seeing a viral image of the Met Gala kitten costume and seeing a viral image of the Puffa jacket Pope. In one, there is a whole world of human implications zipped and compressed into the image. So much stupid graft is in that fur. In the other, it’s just machine data compression and unpacking.
I think a very special sort of human joke is all about wasted effort. These gags are often my favourite jokes of all.
Because wasted effort for a joke is not wasted. It was spent well, with kind intent, to make an audience happy.
And I worry that if we streamline the graft of making silly things, that if a dumb idea can be realised as quickly as it was invented, rather than through the painstaking, stubborn, bloodyminded wasting of our precious time on this earth, in the name of making a big fake Goodies cat and then dressing up in it, we potentially kill so much comedy, and a lot of human joy.
Wasting our time on properly dumb shit makes us happy.
Why would we automate that?
If you want to read more of my writing on how humans comprehend and use comedy, why not buy a copy of my forthcoming book? Because that’s where I’ve put loads of it.