AI and ChatGPT - the demise of the white-collar job?

What’s the best way to get to Carnegie Hall? Lots of practice. Or at least it was until about two months ago when OpenAI released ChatGPT. By now, most people have heard of ChatGPT, while over 100 million have used it in January alone, with some extraordinary results. While Microsoft’s Tay AI-chatbot was taken off-line back in 2016 after programming glitches meant it became neo-nazi in less than a day, ChatGPT seems to be the real thing¹.

For the last fifty years at least, a good education and particularly a good degree from a top university has been a great stepping-stone into a successful and profitable career, and the media reaction to the release of ChatGPT reflects a stark appreciation of what this level of AI means for the formal education process.

No more homework? (Alice Cooper, Schools Out, 1972, Copyright Warner Bros.)

Already we are hearing of universities considering a return to written exams as the only form of assessment to prevent ‘cheating’ in coursework via ChatGPT. In education-obsessed Britain where well-off parents send Mungo and Polenta to private schools, head teachers at fee-paying schools are already apparently moving away from formal written homework in favour of research and preparation for “discussions and assessments” in class². Clearly something very big is happening.

ChatGPT is not perfect. It apparently struggles with maths, and has a propensity to make up facts at various points. Debate in the US has inevitably moved towards the topic of whether ChatGPT is biased in terms of political outlook. But with a database consisting of everything written on the internet until the end of 2021, it has a remarkable ability to mimic high-level responses to questions, expressing answers in language eerily similar to that which one might expect from a very smart, well-informed person.

In his unambitiously-titled essay De Profundis, poet Oscar Wilde noted, “Most people are other people. Their thoughts are someone else’s opinions, their lives a mimicry, their passions a quotation.” If this is the case, we can bypass the question of AI and the nature of originality (a minority problem according to Wilde) and cut straight to the chase. Much that passes for formal education is to a greater or lesser extent an exercise in repetition, mimicry and arm’s-length plagiarism, in the nicest possible sense. The reaction to ChatGPT seems to suggest this is so - the bot is better than the student.

While education has pretty much always been a good way of getting ahead, its impact on wealth, success, household formation and morals has come to the fore in recent years as the different life experiences of blue- and white-collar workers has become more extreme. The topic was examined in detail by Charles Murray in his 2012 book ‘Coming Apart’ and elsewhere. The basic premise is that the life experiences of the two groups has never been more different, with those of the poorer members of society collapsing, especially in the context of traditional paradigms of household formation (ie the nuclear family).

While conservative economists such as Thomas Sowell might highlight the negative effect of welfarism on household formation and the general wellbeing of the poor, it is clear that globalisation has played its part as well, with manufacturing jobs moving abroad to take advantage of lower wages and production costs in the developing world. This offshoring and outsourcing has denuded the industrial heartlands of the developed world and driven a growing wealth disparity between rich and poor marked by labour’s overall share of profits declining over a multi-decade period.

America’s rustbelt - a potent symbol of blue-collar decline.

In the context of globalisation, deregulation, the reduction of capital controls and the ensuing freer movement of capital around the world since the 1970s, the differing fortunes of blue- and white-collar workers can be summarised as follows. The highly-educated prospered from globalisation while the skilled and semi-skilled worker suffered. Brains turned out to be mobile while brawn turned out to be static. The question that AI raises is this: will the technological advances of ChatGPT et al prove to be to white-collar workers what globalisation was to blue-collar workers?

The reaction to ChatGPT and the way it makes obvious the massive leaps in the power of AI in recent years perhaps indicates that the threat is real. Those who remember the 1990s will recall the angry protests during the passage of GATT (the General Agreement on Trades and Tariffs) in Europe and elsewhere as workers and business owners alike saw their future value being eroded by trade liberalisation³. People knew at the time what the downside to globalisation was, and they have largely been proved right. Just as Leonard Cohen sang - everybody knows.

Old King Ludd - machine breakers in the early nineteenth century.

Before we get a Luddite reaction from management consultants and lawyers as they angrily unplug computer servers and defiantly change the formatting of power-point presentations, it is worth noting that a quiet quasi-revolution has already happened in finance and asset management in the past decade or so.

Active investing has been gradually supplanted by passive investing (in exchange-traded funds or ETF’s), trading has shifted from open outcry in the pits of the exchange to screen-based algorithms, and python-coded computer programmes seek to find pricing anomalies where once the star stock-picker or macro guru held sway. While there has been plenty of money to be made in the short term from those with the right technology to profit from this transition, in general finance is probably on a road to being less profitable - at least for the majority of people who work in the industry.

The great thing about technological advance is that it allows us to do more with less. The problem with the same advance is that the results of this efficiency tend to be strongly deflationary over time. This deflation is not just a monetary phenomenon, but one of declining value or self-worth amongst those adversely affected by technological change. The problems and misconceptions that this raises are perfectly summarised in the quotation below from a Sunday Times article about the likely effect of AI on jobs (presumably the sort of jobs that Sunday Times readers have - ie white-collar ones):

“Is your job safe? Probably, if smartphones and the internet are a guide. Indeed, despite the life-altering power of those technologies, economic productivity has remained stagnant in recent decades. Huge swathes of the economy, from healthcare to housing and energy, have proven largely impervious to the march of progress. It is unclear that AI will break through where others have not.”⁴

Healthcare, housing and energy are all ultimately about stuff - medicines, commodities, molecules and the like, all of which are consumed. To compare the building trade to the tech sector is really a chalk-and-cheese argument which doesn’t stand up to the most simple critique. There is stuff and there is code, and they are very different.

Mr Forston of The Sunday Times might want to pay closer attention to the examples he uses. The production of medicine, especially ‘generics’ which have passed their patent-expiry date, is now mainly done at low cost in the developing world. If much of the meat of white-collar work could be similarly outsourced to AI, wouldn’t that put some jobs at risk - or at the very least pressure how much one might expect to get paid for them? This is the deflationary aspect of technological advance.

During the first industrial revolution, workers poured into the cities to toil in factories, and, if people like Marx are to be believed, they began to acquire a class-consciousness and an angry awareness that they had been alienated from the product of their labour. While Marx borrowed the labour theory of value from Ricardo, he overlooked the fact it was really the coal and the steam (the harnessing of which was the truly revolutionary part of the industrial revolution) which was doing all the extra work, not the labourers in the factory, even if it didn’t quite seem like it at the time.

We are possibly still at the stage where doctors, consultants, lawyers and other white-collar workers might think they’ll be ok and their unique skills will retain the premium for which they get handsomely paid. But as explained above, many in finance are already aware that the winds are changing.

But if it’s taken about eight weeks for ChatGPT to show that much of modern education can be undermined at the click of a button, then what price education in the future - especially in the context of the extraordinary inflation in college fees we have witnessed in the past few decades? If the post-graduate jobs which a fancy education used to guarantee fail to command their old premium, why would you pay all that money to go to university if all that awaited you was a white-collar pound shop not the VIP tent of the global elite?

Perversely, we may end up in the sort of situation envisioned by John Maynard Keynes nearly a century ago when he wrote ‘Economic Possibilities of Our Grandchildren’ in which, when the material needs of subsistence were taken care of, he imagined that people might lose their love of money and only work a few hours a day, spending the rest of their time at leisure or improving their minds. Imagine that - education for its own sake.

(This article was written without the help of ChatGPT.)

1. Amy Kraft, Microsoft shuts down AI chatbot after it turned into a Nazi, CBS News, 25/03/2016.

2. Nicola Woolcock, ChatGPT marks end of homework at Alleyns School, The Times, 29/01/2023.

3. Bernard Connolly, The Rotten Heart of Europe, Faber & Faber, 1995, p216.

4. Danny Fortson, The ChatGPT revolution is real, Sunday Times, 05/02/2023.