I am quite happy with the predictions I made back in December. Back then, Pope Leo’s encyclical on artificial intelligence had not yet appeared. I suggested that when it did, its main concern would not be whether machines can think, but whether human beings would gradually give up thinking for themselves. In particular, that we might start treating it as a substitute for conscience.

Now that Magnifica Humanitas has appeared, I think that I was pretty spot on. Pope Leo is asking what artificial intelligence does to the human person. That is the right thing to be thinking about: not whether a chatbot can be conscious, or whether the “singularity” will arrive before the next general election, or whether a sufficiently large pile of graphics cards will eventually replace the majority of white-collar workers. What happens to man if he delegates judgement, memory, attention, language and decision-making to systems that cannot love him, answer for him or suffer with him?

That is a real danger. It is also quite different from saying that artificial intelligence is about to change everything from the productivity point of view. In this respect, I think AI’s potential is greatly exaggerated, at least for the near future.

I do not say this as someone who never uses it – I use it quite a lot. In some cases it can be very impressive. Before my wedding, I built an RSVP website with a database of all the guests, so they could log in, etc. Doing it from scratch would probably have taken me the better part of a week and would have involved a lot of unpleasant debugging, but with the help of AI, I had a working version in an afternoon. That was genuinely useful. It saved me, a hobbyist programmer, a bit of time. But that was about it.

I had not abolished labour, nor had I entered a post-scarcity economy. This is roughly most people’s experience of AI in general. It is useful, but also often wrong and overly confident. The gap between the usefulness of the thing and the mythology surrounding it has reached absurd levels.

It is worth remembering how many of these technological apocalypses we have lived through already. Nanotechnology was going to transform everything. Autonomous cars were going to remove the need for drivers. Blockchain was going to rebuild finance, contracts, identity, art and probably the parish newsletter if you gave it long enough. The Metaverse was going to become the next internet. For a while every large institution was expected to have a “Web3 strategy”, which in practice usually meant paying consultants to explain why a database should be worse.

This is not to say that none of these technologies achieved anything. That is almost never how these things work. The point is that the gap between the promise and the result was enormous. Autonomous vehicles exist, but the world is still full of drivers. Cryptocurrency exists, but most people still pay for things using money rather than a speculative token named after a dog. Virtual reality exists, but most people still prefer reality.

We are constantly told that everything is about to change: jobs will vanish, universities will collapse, artists will be replaced, lawyers will disappear, programmers will be automated, children will be educated by bots, soldiers will be commanded by algorithms, the economy will become unrecognisable; human thought itself will be superseded.

Where is the great productivity explosion? Real technological progress has a habit of showing up in the real world. It does not stay trapped in forecasts.

For most of human history, cloth was expensive because making it required an enormous amount of human labour. Then came mechanical looms and industrial textile production. The result was not merely that people wrote excited essays about the future of cloth. The result was that cloth became cheap. More cloth could be produced by fewer people at lower cost. Ordinary people could own more clothes. The improvement was visible, measurable and useful.

The same is true of serious advances in transport, power, agriculture or manufacturing. If a new machine produces more output with less labour, less fuel, less time or less capital, one does not need a prophet to explain that a revolution has occurred. The revolution appears in prices, wages, abundance and ordinary life.

So if AI is the new loom, where is the cheap cloth? If artificial intelligence is really about to make software production close to free, where is the cheap, excellent software?

Perhaps the answer is that the revolution has not arrived yet. But that is exactly the point. The case for AI is always being pushed slightly into the future. The present systems hallucinate, but the next ones will reason. The current products are expensive, but the next wave will be cheap. This is not engineering – it’s eschatology!

The strangest thing about the AI debate is that the boosters and the doomers often agree on the main point. Both assume that AI is the central event of the age. One side says it will save us; the other says it may destroy us. But both speak as though history has been handed over to the data centre.

Pope Leo’s encyclical is valuable because it refuses that premise. AI may not be coming for our jobs quite as quickly as advertised. It may be coming for our judgement instead.