Fooled by the duck

In 1739 the Grenoble inventor Jacques de Vaucanson unveiled his latest invention to the people of France. The Canard Digérateur, or Digesting Duck was a life-sized model of a waterbird that appeared to be able to take food in its beak from a human handler, and then miraculously poop it out of its other end.

The magical fowl sat atop an enormous box in which, de Vaucanson claimed, sat the complex scientific mechanism that allowed the food to be digested.

It was truly a miracle of the then modern engineering, and it brought fame to de Vaucanson.

It’s a story that often comes to my mind when the latest wave of AI hype hits the news. In 2022 we have seen both AI Art in the form of Dall-E and most recently the miracle of natural language processing Chat GPT. We now have a discourse that asks whether the work of writers and artists are at threat given the amazing progress that artificial intelligence continues to make at pace?

I guess that there are many professions that are at risk, as there were at most stages of automation throughout the time since de Vaucanson. It’s also worth remembering, however, that in the UK at the moment we have a huge shortage of people to do all sorts of jobs.

Ultimately it comes down to why we do the things that we do. I write to try to make sense of the world. I of course could ask a chat AI to create a blog article in the style of Matt Ballantine on the theme of AI using de Vaucanson’s duck as a metaphor, but what would be the point? I wouldn’t have spent the weekend with the ideas bouncing around at the back of my head with the idea of this piece being a way to structure my thinking. I would have learned nothing. And most importantly, neither would the AI.

I could spend my spare time instructing AI music generators to generate music for me, but I would far rather spend an hour or two noodling about with eight bars of nothing in particular because it will clear my mind.

However, I could quite happily ask an AI text generator to create the next 25,000-word procurement submission that I’m asked to produce to win the next big contract. And at that point, no doubt, the people commissioning the next big contract will have an AI read all of the submissions for them.

This is my biggest fear with what I see going on at the moment. Not the loss of jobs, but the automation of stupidity, in the same way that we see stupid paper forms replicated into stupid online forms rather than people actually addressing the issues that daft processes cause.

I can see a world in the near future where content-generating text bots create content marketing to exclusively target the virtual eyeballs of content-reading consumer bots. If you thought your local newspaper’s website was unreadable already, you ain’t seen nothing yet.

It’s important to keep the duck in mind here. Image-generating AI bots aren’t creating art, they are using algorithms to generate image file output based on the statistical analysis of millions of other pictures. Prose-generating AI bots aren’t creating literature, merely using algorithms to generate text output based on the statistical analysis of billions of other words. They aren’t really digesting and shitting, they’re just pretending.

And that’s where we need to take a step back and not get fooled into thinking that they are doing what we are doing. They are doing a very good impression of what we are doing, but with a massive box underneath them to do a considerable amount of processing to maintain that illusion.

Yes, some jobs will be rendered obsolete by all of this. There are many people in the creative industries whose jobs are anything but creative. And there are massed ranks of people in creative roles who will find their jobs much harder to sustain when the lower-value bits of their industries are removed.

There are huge implications. How do you learn the trade if there aren’t low-level jobs in which to learn? How can you afford to work in an industry which doesn’t pay at the entry point?

But there are lots of things that we do that could be done by machines that maybe would be better done by them. And there are also lots of things that we could automate that really we should just stop doing altogether. Crucial in that process needs to be the realisation that they do these things without “thinking”.

After all, if you built an amazing mechanical duck, would you really put in the effort to make it shit*?

* it actually turns out that de Vaucanson didn’t go to all the effort to make his duck shit. It was all a hoax and the mechanism, whilst clever, didn’t actually process the feed at all.