Is AI coming for my job? Only if I’m already too futile to matter

Is my job about to be taken by a robot? Not according to the robot itself. ChatGPT – the AI language generator that has gone viral since its launch in November – assures me that it is not capable of replacing a human journalist. Sweetly self-deprecating, it claims to lack the “critical analysis and human perspective that is essential to good journalism”. But it would say that, wouldn’t it?

If you haven’t tried it yet, ChatGPT is an online software that can write like a human. Type in just about any question, and its bots will scour the internet for answers before replying in clear and unnervingly fluent English.

It can explain the origins of the Israel-Palestine conflict, suggest birthday party ideas for a teenage boy, or describe the landscape of Devon in the style of T S Eliot (“the sound of the waves crashing against the shore was like a dirge, mourning the passing of something once beautiful”).

I made it do all this today, before being turfed off the site for asking too many questions. Like a real writer, ChatGPT can be quite temperamental.

Similar online AI can create any image, in any style, on demand. Type in: “Photograph of a spooky woman sitting on a giant chess board”, as I just did, and that is exactly what appears. It looks like art: not great art, but good enough to put on a postcard, and a lot cheaper than hiring an actual graphic designer.

You already know, don’t you, that this technology is going to end in tears? If the human race had an ounce of collective will, we would strangle it at birth. Why unleash upon ourselves another tool for digital mischief and misery? Why make it any easier to create revenge porn and fake news, or to cheat, plagiarise and undercut each other?

A class action suit filed in California this week by a group of artists points out that image-generating AI “scrapes” the internet for pictures by real artists, before using this “data” to create new images. In other words, it mimics human content and then makes us redundant.

The singer Nick Cave warned this week of the “emerging horror” of this kind of technology. His argument is chiefly aesthetic: robotic writing, he argues, is only ever a “grotesque mockery” of human creativity. This is certainly true when it attempts to mimic any kind of genius. (Asked to write a song in Cave’s lushly gothic style, ChatGPT came up with: “In the depths of the night, I hear a call/ A voice that echoes through the hall …”)

But the bigger worry, I think, is how good it is at mediocre writing. ChatGPT excels at the kind of bland, textureless copy we have come to expect all around us. It is fluent in the language of council websites, of banks and corporations, funeral homes, rental agencies and NHS leaflets.

Such institutions may well be tempted to cut costs by using a robot instead of a copywriter. But how will they hold a robot to account if (when) it gets things wrong? ChatGPT gets all its facts, ideas and language from the internet, after all. In its own words, “it can generate text that is biased due to the data it’s trained on”.

Are we really stupid enough to outsource the human voice to a digital ventriloquist trained in lazy pieties, conspiracy theories, bigotry and sloppy thinking? If so, perhaps the robots deserve to win.