The AI productivity myth (and two more)

I’m trying something new this week, although it will likely be a one-off. Three subjects caught my attention over the last seven days. Rather than chose one, I thought I’d share all three.

Will AI really make us more productive?

It has become a highly useful article of faith that AI will ‘improve productivity’. Every political leader is practically obsessed with the notion, and I understand why. It promises better results without spending more money. We could do with some of that.

This week Kier Starmer was banging the AI productivity drum again. Turbocharged. Transformative.

Will it? Are we really sure about that?

Anyone who’s actually used a generativeAI tool like ChatGPT in their work will know that it can allow you to produce much more stuff in less time. There are even a handful of academic papers that show it really does increase productivity (and even make staff happier and more engaged). Let’s take teaching, which Sir K/S mentioned explicitly. I have no doubt it can help teachers mark papers more quickly, generate lesson plans, create materials, complete paperwork vastly quicker. All the things that most teachers say takes up too much time.

However!

Technology has always promised remarkable and stunning productivity improvements, and yet so often fails to actually deliver them. In the 1980s the economist Robert Solow famously said, “You can see the computer age everywhere except in the productivity statistics.” He was referring to the strange paradox where companies were all introducing machines into the workplace which sped everything up – and yet for 20 years productivity actually decreased.

The AI productivity assumption is partly based on a very reasonable proposition, which underpins a lot of the current hype: that if we simply have more information and more computing power, problems will be solved more quickly. But this is not obvious. In the 1990s the great Neil Postman (in respect of computing more broadly - AI wasn’t even a thing then) complained that: ‘The computer argues, to put it baldly, that the most serious problems confronting us at both personal and public levels require technical solutions through fast access to information otherwise unavailable. I would argue that this is, on the face of it, nonsense’.