History says AI will boost productivity, wages and employment

Jon Boys, Senior Labour Market Economist at the CIPD, examines how AI might impact automation, wages and employment 

During the Industrial Revolution in the late 1700s, economic growth exploded but curiously wages stagnated, a situation termed as Engels’ pause (after philosopher Friedrich Engels). Citing the pause, Carl Frey argues that the Luddites - who protested against industrialisation and the subsequent loss of employment and fall in wages by destroying textile machinery - had a point. Technology might make us all richer in the long run, but the short run is for some a lifetime.  

With AI poised to boost productivity, are we facing another Engel’s pause or an immediate uplift to pay? 

Pay and jobs 

In the UK, real pay today has barely recovered to levels seen after the financial crash of the 1980s, with lacklustre productivity to blame. But if AI can do the heavy lifting, we could become more productive, in turn boosting wages. Meanwhile, jobs show a rosier picture. Unemployment in the UK remains incredibly low. Businesses and policymakers are worried that a lack of workers could constrain growth. As ageing populations move into retirement, workers are becoming scarcer and the imperative to adopt technologies like AI is high. 

Historical precedent suggests employment and wages will rise. The Engel’s pause was just that, a pause. We eventually became richer because of our ability to dramatically boost output. Wages grew because efficiency meant employers could afford to pay more. Employment grew because we came up with new jobs to create the products and services all this new wealth could buy. Historical precedent would suggest that AI-induced productivity gains will do the same as previous rounds of automation, boosting wages and employment.  

Cheaper means more 

By making a product or service cheaper, AI could boost demand. Machines made translation cheaper and boosted the number of people working in translation (though their average wage did fall). And there’s Uber, which by reducing the cost of a taxi service increased the number of journeys taken, and Spotify, which increased the amount of music people listen too. So what new services might AI bring within reach of the everyday consumer? Most people have never commissioned a piece of artwork for their house, a headshot for their Linkedin profile picture or had a one-to-one language or maths tutor. 

Early evidence – unorganised vs organised labour 

A fascinating study looked at the wages of freelancers on the platform Upwork. This is one of least regulated parts of the labour market as disparate freelancers around the world compete for work with no ongoing employment obligation. It may, therefore, be a bellwether for the effects of AI in the rest of the labour market. The most exposed freelancers are ‘experiencing reductions in both employment and earnings’ and highly rated freelancers are no better protected. Essentially generative AI has made people more substitutable. 

Although ultimately ineffective, it must have been cathartic for the Luddites to smash up a loom. But how can one rage against a large language model like ChatGPT? The Hollywood writers’ strike is probably the modern-day equivalent. Unlike the dispersed and unorganised freelancers on Upwork, the script writers are geographically and linguistically similar people who were able to organise through their union secured assurances over the use of AI in Hollywood. Time will tell if the threat is removed or just postponed - given the speed with which AI animation is moving, an upstart outside of Hollywood might soon be making AI movies. 

White vs blue collar work 

In a study with management consulting firm BCG, 758 consultants were assigned 18 tasks. Some had access to ChatGPT4 and some did not. The AI-supported consultants completed 12.2% more tasks, 25.1% quicker, with 40% higher quality. Interestingly, middle performing consultants were able to close the gap with their top performing colleagues when using AI, which suggests that AI could be a great leveller and may therefore reduce earnings inequality within occupations. It is notable too that this new wave of automation is most amenable to white collar knowledge work.  

What will happen to employment and jobs? 

Given what I have read and researched, I’m going to be bold and offer some possible predictions (but if you have other thoughts feel free to let me know if the comments below):  

  • We will not see mass unemployment. Employment will increase. Services will get cheaper, and people will want more of them. Whole new industries that we can barely imagine today will be created. 
  • AI will augment rather than replace jobs. People will be more productive and get paid more (in aggregate). For individual occupations it will depend on various factors including regulation, organisation/unionisation, amenability to automation, and demographic change. 
  • The pace of adoption of this technology will be quick, and therefore the benefits will also arrive soon.  
  • Most occupations and industries will be affected by AI - but to varying degrees - due to its wide applicability to different problems. 

Artificial General Intelligence 

The stated mission of open.ai, the team behind ChatGPT, is to create artificial general intelligence (AGI). Many in the company believe that this is not only possible, but, when it comes, all bets are off. The problem-solving abilities of a superintelligence mean that things that worry us today, like money, will suddenly be engineered away. I hope they’re right.  

Thank you for your comments. There may be a short delay in this going live on the blog page as we moderate the comments added to our blogs.