The idea in a nutshell : The consensus seems to be that AI will widen income inequality, negatively affecting low skilled men more than any other social group. The government should focus on the means of retraining these people (a classic approach to technological unemployment), ideally using the very technology which displaced them to help, as a key initiative in the fight to find them work. They should begin now.

Jason Furman’s view on the imminent impacts of AI is different to most

I’ve covered many aspects of the potential impact of AI on the economy and jobs in other articles on this blog. Then I came across some work by Jason Furman, Chariman, Council of Economic Advirors at NYU.

Good things and bad things about Furman’s report

Furman’s report, which prompted this article, is well worth reading. I’ve linked to it above. I had come to similar conclusions as him about how the benefits of the new technology would be split within the economy (unfairly).

  • One of the best things about his paper is that he reinforces his point about the distribution of the benefits by correlating likelihood a worker’s job will be automated according with how much each worker gets paid. There is a strong correlation between lower wages and a higher lihlihood workers will be displaced.
  • He also articulates better the reasons for a general decrease in wages better than I did in my version of that article.
  • I disagree with his notions of increasing the collective power of workers (I literally cannot remember a time an economist proposed that sort of things as a way to improve the circumstances)

To avoid the natural concerns associated with unequal distribution of the benefits of AI, Furman proposes reframing tax policies to encourage work, expanding training, raising the minimum wage and increasing the collective bargaining power of employees. Some of which I agree with.

The Economist suggests educational reform is required

It’s actually very hard to say anything about this subject that The Economist has not said a good deal better. In this old but still interesting and important article, the magazine suggests (comprising some work from Brynjolfsson and Andrew McAfee) a focus on educational reform and increased educational investment will be beneficial to the country in the circumstances that will arise.

A relevant example ?

I grew up in the North of England when I was younger, at the time of the Miner’s Strike.

The miners knew a cataclysmic change was coming to them, their jobs and their whole way of life for years before Thatcher’s policies hit them.

Unfortunately, many never worked again when their positions in the pits disappeared. They spent the remainder of their lives ‘hidden away on disability benefit’.

At a more macro level, entire regions suffered. Areas involved in the miner’s strike and the unemployment that followed came up in the 1994 European enquiry in to poverty. Every single ward in the city of Wakefield was classified as ‘in need of assistance.

Government spending either in the form of

In some ways, the arrival of Artificial Intelligence represents is a new threat. In other ways, it’s not a new threat at all. There are established ‘textbook’ methods of dealing with technological unemployment. Education is one of those methods. (Along with redistributive tax schemes, mandatory hiring of humans, regulation including slowing down technological progress).

The government needs to match training gaps to educational services and newly trained workers to job gaps as they appear

Private sector job matching facilities (like the website ‘Seek’) exist and are getting better each year, ironically, sometimes, through the use of AI algorithms. Increasingly, these sites match not just keywords to resume but the essence of a job to the essence of someone’s skills.

Each company has an incentive to look for the skills it wants in the job market. They have an incentive to build an algorithm which connects existing job seeker / worker requirement dots.

No one company, however, has an incentive to identify displaced workers that need new skills, determine the jobs the economy needs and then train them to address the gap.

Unfortunately, unless that happens, displaced employees may detract from rather than contribute to the economy for the rest of their lives – just like the miners from the North Of England.

Save the real basics of the job centre, job matching is something the government has failed to be involved in. Paradoxically, the private sector argues ‘often’ that they are receiving graduates who don’t have the skills business needs when they leave university.

In the not too distant future, swathes of people with out of date skills are going to be displaced. Warning and education on the threat and the need to plan for multiple careers seems like a first useful step for the government to undertake.

Providing better information on the jobs which are arising in the economy and partnerships with organisations that need to hire people seems a useful second.

It may not be the only thing but it’s a useful step

The shame from my point of view is that this facility for retraining was not part of the government’s narrative in the recent 2016 election. Assisting with retraining is a standard government response to large numbers of displaced workers as the Economist and the text book responses point out.

The STEM training the government are proposing seems fledgling and focussed on the very young (those of school age). That doesn’t seem inappropriate as much as it does inadequate. When ‘typical Australians’ (and I say that with all the respect due to them) don’t see the structural technological change heading towards them it can be all the harder to accept when it lands on them. As we’ve seen, those affected tend to ask the government for financial assistance at those times.

In many ways, it comes down to personal responsibility, informed by the government. My experience, having grown up near the coal mines in the midlands of England when the Miners strike was on was that they felt strongly. They were smart men, motivated to improve things for themselves and their family. They had a strong sense of personal responsibility.

Linking that motivation to re-training in the right areas and a facility which will help them execute seems like an obvious step. It’s something we should be planning for, given the information we have.

An example of the sort of retraining that’s possible

It’s early days for AI in education but there are examples. ‘Learn English with Aco’ is a free to download bot you can pick up in the Google Play store. It interacts with you like a person and talks to you about the things you’re interested in – in English – targeted at those who speak languages other than English. In the process, users learn more about the language when they are offered lessons and prompts by the bot.

Obviously, once created, this sort of algorithm gets better with each interaction it has. It is scalable to thousands of people at one time. Why couldn’t such a thing be developed for STEM courses ?

It could be smarter algorithms which retrain people

Ironically, it could be the technology which displaces people which ends up educating them. In the same way Seek and other job matching companies are using more intelligent software to match candidates to jobs, sophisticated algorithms are becoming available to train people in new skills. Personalised education could soon be widely available and cheap to deliver.

Linking those retrained workers to a platform which tracks growing gaps in the workforce will more usefully manoeuver people from the scrap heap to maintaining their position as contributing members of society.