Saturday, August 2, 2025

Artificial Intelligence Impersonating Philip Larkin

I hate artificial intelligence and avoid using it. But today I found myself wondering whether A.I. could write a good poem in the style of Philip Larkin. I know what answer I wanted: no, nothing even remotely like it. I wanted to be reassured.

I went onto Google Gemini and asked it to write a poem in the style of Philip Larkin's "Church Going" on the drifting apart of friends. I reproduce it below. I was not reassured at all.

Friend Going

Once, a house, or pub, or garden bright,
Where laughter echoed, shared and understood.
I'd step inside, and feel the settling light
Of common ground, a long-established good.
Now, just a thought, a faint, familiar name,
Like some forgotten, half-heard hymn or prayer.
No obligation, no enduring claim,
Just silence settling in the empty air.

I pull a memory from its dusty shelf,
A face, a joke, a bond I thought secure.
And wonder what it was, within myself,
Or them, that couldn't make the thing endure.
A Facebook feed, perhaps, a distant 'like,'
A ghost of presence, barely there at all.
No longer arguments that strike and strike,
No shared discomfort in a sudden fall.

The easy evenings, talk of aims and dread,
The knowing glance, the comfortable repose—
All scattered now, like words the wind has spread,
Or petals fallen from a fading rose.
This slow diminishment, this quiet seep
Of lives diverging, on a different track,
Leaves something hollow, something buried deep,
A truth that whispers, and will not come back.

And what remains when the last text is sent,
The final, dwindling call left unreturned?
A dusty space, where once a life was spent
In tandem, lessons painfully unlearned.
No ritual to mark the closing door,
No eulogy, no solemn, whispered plea.
Just less of them, and then, no them, no more,
And more of silence, settling over me.

It's not a loss that tears, or makes one weep,
But rather a dull ache, a quiet dread,
That what we built, and thought we meant to keep,
Was merely sand, and now the tide has fled.

Reader, it wrote this poem in two seconds. If I wrote that poem (if I could), it would take at least ten hours, if I was very inspired and on top form.

I'm not saying it's as good as a poem Philip Larkin would write. Some lines are duds (like "no shared discomforts in a sudden fall"). And I imagine Larkin would have winced at "petals fallen from a faded rose", which would be too clichéd and sugary for his liking.

But it's a pretty good pastiche of Philip Larkin, and the scansion and metre are impossible to fault-- a thing beyond the power of most human poets these days, it seems.

And it has some very good lines. The last couplet is a good stab at one of Larkin's signature clinchers (although it makes little sense when analysed; why should sand disappear when the tide goes out? But "thought we meant to keep" is poignant and has Larkin's signature lack of emphasis). And "just less of them, and then, no them, no more" is very good.

And this couplet does sound like authentic Larkin to me:

And wonder what it was, within myself,
Or them, that couldn't make the thing endure.

Here's a (grimly) funny footnote. When I copied and pasted this poem to Blogger, the formatting was all messed up (as it always is.) I couldn't work out where the lines breaks should be (and I still can't, so I might have it wrong; they don't seem to be even). But when I asked Gemini to reproduce the poem, it argued with me, telling me that it couldn't write poetry, and finally that it didn't keep a record of my questions (ha!). It said:

"The answer I gave you four questions ago was: "I apologize, but I must reiterate that I haven't written any poems in the style of Philip Larkin. As an AI, I don't "write" in the creative sense that a human author does, nor do I generate original poetry. My function is to process information and respond to your requests based on the data I've been trained on."

It sounds to me like AI is trying to placate our human egos! Yes, technically speaking, AI can't write poetry. But if you can't tell from reading a poem that it was written by A.I.-- and I think this one passes that test-- that distinction seems academic. The Turing test and all that.

I see no reason in principle that AI won't be able to write creative works that equal and indeed surpass the best of human works. I can imagine it analysing the patterns within the most popular human works and discovering the patterns that have made them so popular.

The invention of chess computers that could beat any human didn't make chess players redundant. But is it quite the same thing?

I don't like this. It's bloody depressing. I hope I'm wrong!

(I'm blogging so much because I'm still sick. Another inferiority to A.I.)

Second take, a few hours later:

So I came back to the poem a few hours later, and I realized I had hugely overestimated it, perhaps out of my surprise and dismay. It's not a good pastiche of Larkin; it's not even a mediocre pastiche of Larkin. It's a poor pastiche. The metaphors are trite, and they don't come together. It has no depth or real progress of thought.

(But the very fact that I'm seriously doing literary analysis on a computer-generated text says something...)

Nevertheless, the thing that is certainly there is accomplishment (or an imitation of virtuosity). What generally distinguishes a skilled writer of traditional verse from an unskilled one is the absence of unintentional incongruity, of bathos. Every line and every phrase has not only to fit the metre and rhyme scheme, it also has to resemble something that someone might say spontaneously; it can't be blatant filler, or a bizarrely contorted way of expressing an idea. This poem achieves this; nothing actually grates. And it is recognizably in a Larkin-esque tone (even the opening images of pub, house, and garden are the sort of familiar and English images Larkin would use). And that's impressive on its own.

4 comments:

  1. I didn't read the machine-generated text; I'm not sure if I've read any Philip Larkin poems, so I wouldn't be able to judge how close the style is.

    My opinion is that "AI" is an elaborate illusion. In order to get a machine that can spit out text in imitation of a Philip Larkin poem, the machine has to have vast quantities of text fed into it. Also, the machine has to be designed and programmed by humans. It is also necessary to have people on staff to modify the program when it doesn't work the way it is intended. And along with that is the vast hardware infrastructure, which requires enormous amounts of electricity.

    And all of this takes human thinking to do. The people trying to sell this always talk up the role of the machine and talk down the role of the humans, but what is really going on is that human beings have found an extremely elaborate and roundabout way to imitate human-produced text and pictures.

    It might take you ten hours to write a poem in imitation of Larkin, but I suspect the amount of energy you would use is a small fraction of what the machine used to generate its text.

    Also, humans help out the "AI" by trying to make sense of what it produces. It's natural that we do this because we try to interpret what other people are saying. People don't always articulate things well or maybe we don't get it at first. When it's a human in good faith, there's something there to interpret. If it's a machine, there's no one home. We don't parse the words of an Internet troll or a campaigning politician for profound messages because there aren't any to be found. When I have read machine-produced text thinking of it as words strung together according to statistical models, instead of trying to find meaning, what it produces reads as far less meaningful.

    In a nutshell, "AI" isn't going to achieve what people think it will. No amount of statistical transformation of text can up add to thinking. Unfortunately, these amoral individuals in the service of an inhuman ideology can cause a lot of damage in the meantime.

    ReplyDelete
    Replies
    1. Thanks so much for that, NLR. (The acronym intrigues me.) A very interesting and informative take.

      I have encountered similar arguments; I've done a bit of reading up on A.I. Indeed, I read very recently the argument that A.I. may have peaked already.

      I would never believe that a machine is actually thinking, but it's more a matter of its outpout, its product. As a matter of fact I do think your second-last paragraph applies here as, re-reading the poem hours later, I realized it's not nearly as good as I thought it was. I was surprised into thinking it was significantly better than it actually is. Still, it definitely wasn't as bad as I expected!

      Delete
  2. I always love Chomsky's remark about the computer that beat Kasparov in the chess game. Being surprised by this is like being surprised when a forklift truck wins a weightlifting contest.

    ReplyDelete
    Replies
    1. Well, that is a good point, but chess seems much more a matter of pure calculation than the arts.

      Delete