Here’s a boring interesting chart showing the changing composition of the UK workforce, by industry, over the past 25 years.
The chart looks boring in email - but much better here.
There are some very significant changes. The proportion in manufacturing has more than halved over the past quarter-century, and there have also been sizeable drops in the number in construction and retail. As we become an increasingly service based economy the proportion in information and communication, public administration and professional services have all grown. The impact of the ageing society is also evident in the growth in health and care work.
The size of the labour market – 26 million people in 1997 and 32 million in 2022 – also means that even small increases in proportion can translate into hundreds of thousands of extra jobs. The proportion working in real estate has only grown by 0.4% (perhaps a surprise given how many estate agents there are on the high streets these days) but in terms of people that’s more than 200,000 extra in the sector (not all of them estate agents).
But the other thing that jumps out is how slowly these shifts happen. This is a long era in which globalisation and the steady adoption of the internet helped transform businesses and the way we work. But while the composition of the workforce today is significantly different to the 1990s it isn’t radically different. There are still more than 2.5m manufacturing jobs in the UK and more than 3.5m retail jobs. Although the share of jobs that are in construction has dropped, there are more people working in construction today (2.1m) than there were in 1997 (1.9m).
All of which brings us neatly onto ChatGPT, other Large Language Models and the emergence of generative AI programmes that can do everything from analyse code, draft essays and win photography competitions. A few months on from the release into the wild of OpenAI’s chatbot, ChatGPT, examples of its output have flooded the internet. In terms of the ability to create detailed, structured, human-like text it is hugely impressive, and quite possibly transformative for many businesses and sectors.
In amongst all the reactions is a common theme: whose jobs will AI take, and how many? Could we all be out of work in the future?
Really annoyingly, it’s no longer possible to embed tweets in Substack posts. You can access the original tweet here (for now)
ChatGPT’s ability to respond to prompts with plausible and seemingly authoritative content is remarkable. Already, it has raised questions about what and who it could replace. Some (alright, that one guy) have suggested that ‘AI has replaced CEOs’ because ChatGPT can generate a team meeting agenda and draft an email telling staff to work faster. In my experience, CEOs tend to do a little more than that. You’d also be able to find the same information by searching for it, in the old fashioned way.
When thinking about something new and exciting, like ChatGPT, we need to check ourselves. As we’ve seen, changes to the workforce happen slowly, even when the technology available leaps forward.
The excitement over this latest step forward in AI is both revealing and predictable. Throughout history, from the agricultural revolution and crop rotation to the industrial revolution and the spinning jenny, new technology has changed the way we work and the composition of the workforce. In the words of Ray Kurzweil:
“We have already eliminated all jobs several times in human history.”
That change happens slowly, not suddenly. One of the great myths of automation is that new technology replaces jobs. This isn’t quite right: it replaces tasks. Eventually, it might replace enough tasks situated close enough together to render a specific role obsolete. But that happens slowly, generally giving workers, businesses and consumers time to react and adjust.
Here’s this edition’s boring chart: the proportion of people in employment over the past 50 years.
It’s remarkable. The proportion of the population in work in February 1971: 60.8%. The proportion in work exactly 51 years later: 60.7%. There’s been a lot of technological disruption in that time, and it doesn’t seem to have prevented us from working.
Fortunately (or unfortunately, depending on your view) the same is likely to be true of the next 50 years.