I came across this article today. Following on from that, I came across this extremely similar article, but on a much more readable website. Both articles date from early last year, and both concern the same event: a machine has written poetry.
I don’t approve – it is important to me, I find, that there are some arenas in which mankind is not surpassed by machines. I’m happy to concede efficiency, I’m happy to admit that a computer can simulate and calculate far more rapidly and accurately than I could ever hope to achieve. All the privilege I claim for my own species (it is not a very enviable one: you need not covet it), is that of creating: of making something new and original, of infusing ink and dead trees with beauty and meaning, of finding the figure trapped inside a block of stone.
Computers can do many things, but they cannot truly think – their complexity is the result of extrapolated rules and repeated algorithms. There is no spark, no vivifying fire. A computer can reproduce the Mona Lisa by flinging paint at a digital canvas, but it can’t make something equally beautiful and deliberate.
Or so I would like to believe. But that viewpoint is a beleaguered one, and there are so many different arguments that can be made, about the true nature of consciousness and whether our minds are simply emergent phenomena from an equally constrained but less well-known system of petty rules that mirrors our current computing architecture.
There is neither space nor time here to discuss the various different philosophies and approaches to this idea, the myriad possible challenges that my hoped-for consciousness faces. It must suffice to know that my belief is not secure – at any point, it may be discovered that we are nothing more sufficiently advanced technologies, that the universe is fully deterministic, that choice and art and beauty are simply shadows on the cave wall.
I don’t know that our minds are markedly different from any that we might create. I believe that they are, and I hope that they are, and I would be less than thrilled to be proved wrong.
Thus, I become alarmed when I hear the news that a machine has usurped humanity’s creative prerogative. I was particularly alarmed in this instance because the example poem given in the first article wasn’t actually that bad.
When I in dreams behold thy fairest shade
Whose shade in dreams doth wake the sleeping morn
The daytime shadow of my love betray’d
Lends hideous night to dreaming’s faded form
Were painted frowns to gild mere false rebuff
Then shoulds’t my heart be patient as the sands
For nature’s smile is ornament enough
When thy gold lips unloose their drooping bands
As clouds occlude the globe’s enshrouded fears
Which can by no astron’my be assail’d
Thus, thyne appearance tears in atmospheres
No fond perceptions nor no gaze unveils
Disperse the clouds which banish light from thee
For no tears be true, until we truly see
It doesn’t read like something hacked together by a clunky algorithm. Any idiot can do that, and it suggests no intelligence on the computer’s part. I myself, whilst dabbling in Python, managed to quickly throw together a program that made centos by chopping and changing different lines of poetry. After I ran it a few times, I ended up with things that almost made sense, but it was clear that this was something generated, not written; there was no element of conscious craft.
This poem is different. It doesn’t clearly hang together, there isn’t a coherent message, but it is an awful lot more sophisticated than my paltry programming. It makes sense, even if it doesn’t sear the soul. You can read it, and work out what is being said without too much effort. A person could have written it, and should a person claim to have done so, I wouldn’t be particularly surprised.
This left me somewhat disquieted. If machines could write poetry of this calibre, then it stands to reason that they could eventually write poetry that matches even some of our greatest writers. I did not care for the possibility. Still, a strong hope remained that, like a lot of online journalism, this story was bunkum.
So I looked into it, read a few more articles, and it turns out be a load of nonsense. Massively exaggerated headlines putting the most outlandish possible spin on things. The sonnet wasn’t written by a computer, it was written by a man sitting in front of a screen, and choosing from a list of suggested options. The computer just provided the possible predictions, based on Shakespeare’s lexical fields, while it was the human brain – the living, unique, creative brain – that selected words and hammered raw nonsense into something a little more coherent.
This time, it was a false alarm. Nobody created a machine that can do everything we can do. And so my beliefs remain unchallenged, and I am free to continue believing that I am more than just a panicky animal that follows instinct in an extremely roundabout way. I find that very soothing.
The whole thing did remind me of a short story though – Roald Dahl’s The Great Automatic Grammatizator. Like many of his stories written for adults, this one is a lot darker than his more famous work – the Dahl in his short stories has always struck me as a rather unpleasant man. Talented, but spiteful and sordid in his imaginations.
The Great Automatic Grammatizator is one of his better ones. It concerns the invention of a machine that can write. Through careful management of its controls, the characters discover that they can churn out coherent novels with only a minimum of effort.
The story ends with a change in perspective, swapping from the viewpoint of the machine’s inventors to a struggling artist – by this point in the story, all books are automatically created, and authors are steadily being driven to starvation by a market flooded by books that are just as good as their own.
The title of this post is taken from that story’s final line, as the artist begs for the strength to resist the lure of easy money – to allow himself and his family to die in order not to compromise his artistry.
That, on the face of it, is a poor strategy: it guarantees failure. But it is also a noble one, and one I agree with: I don’t want art to be outsourced to machines. Because art is something precious and necessary; its profitability shouldn’t change that.
I don’t want to see the day, should it ever come, when computers reach the point at which they can write with the same power and beauty that we can. Should it turn out that nothing is preventing them save time and research, we will lose something valuable.
Our ability to create is one of the best things about us. We need that spark, and I don’t want to discover that so much of what has driven us and improved us has just been an illusion, an accident of complexity and not the pinnacle of sentience.
I tend to let my prose run away with me, especially when I get overly involved in a topic. Forgive the gushing, forgive the overly sentimental outpouring of sentiment. For these final paragraphs, I will strive to be more grounded and less over the top.
I feel that it is important that something sets us apart from machines. I don’t really want to be just a complex machine. Our ability to create is so far an ability denied to computers, and I would like to keep it that way. But I worry that with increasing complexity in computers, we will eventually approach the point at which our AI is as complex as us, and displays the same abilities.
I don’t want that. It is more than likely that the day will come when we clearly demonstrate that there is nothing special about humanity, that there is nothing more to us than there is to anything else. But I hope that day never comes. I choose to believe that there is something that sets us above our tools.