Automation, AI and their societal implications

Automation and AI promises massive productivity gains to many occupations. The recent advent of generative AI has only supercharged this trend by enabling human like interactions with AI and computer systems as mentioned in this McKinsey report.

Two conclusions are oft-drawn when thinking about long-term AI and automation trends. First, AI and automation will not make occupations obsolete but complement/augment workers in those occupations. Second, AI will lead to massive productivity gains and it's a good thing.

But I can't help but ponder on the potential higher order effects of the above conclusions and what it means for different people living in different countries at different stages of development. And also whether we are not being nuanced enough in our thinking.

First, on the point on how occupations are unlikely to be entire replaced, the usual line of reasoning goes that AI would not be able to perform 100% of the job and hence people will still be required. While that may be true, albeit to varying degrees in different occupations, the point is not that the affected occupation will cease to exist but that for a given level of demand, fewer people will be needed.

What that means is that existing people in the affected occupation will be displaced and/or new entrants to the market of the occupation will find it increasingly hard to find employment opportunities, unless the demand for people in that occupation grows faster than then rate of per capita productivity improvement. Or the net demand increase in human workers as a whole is faster that average productivity growth, assuming ease in transitioning from one occupation to another (a big assumption if you ask me). In other words, the pie grows larger fast enough. Otherwise, people in other occupations will come under increasing competition from displaced workers, driving wages down or pushing people into unfamiliar occupations where they might have to accept a lower wage.

This is even before we consider how economic value chains will be reconfigured to be human-lite or human-less in the coming years. I believe this will likely be the case which means that even as the economic pie grows the demand for workers may not.

There is of course the chance of new occupations and industries being invented that will take up more workers. This has happened repeatedly over the course of decades in the past few industrial revolutions. For this to happen, there needs to be heavy focus (and funding) for innovation. Not to mention that the transitional period will still be painful for people involved.

On the point of how productivity gain is a “good thing”. For companies, this statement is mostly true since they can get the same amount of work done for less, either with fewer workers or with workers willing to work for less. Yet even this is subject to the limits of demand. Just like how interest rate hikes are generally good for a bank's profitability until it becomes prohibitively expensive for people to borrow and loan volumes decrease.

For individuals, the most obvious conclusion is that people who are in jobs that are in hot demand like AI in recent years (but even this will pass eventually), will do well. Those who manage to keep their jobs, if only barely, will do OK. In fact, maybe better than OK if productivity gains lead to cheaper goods and services and thus a relatively higher standard of living (another big “if”).

It is when we consider society as a whole that the picture becomes more varied. Suppose you live in an aging society like Singapore, then the productivity gains caused by the use of AI and automation is generally a good thing since you are likely to have more people exiting the workforce than entering and the use of technology might help to sustain production with fewer workers. But if you live in a country with a very young population and a large number of young people entering the workforce each year, will the productivity gains still be good? Unless the economy can continually grow at a breakneck pace, unemployment (and all the challenges that come with it) will very soon become a problem.

The structure of the economy and how society chooses to respond to the likelihood of more people being unable to participate in the economic value chain also matters. For example, if an economy is predominantly made up of labour intensive industries, then a growing economy might have to be fueled by more workers till the day when AI in robotics progress to the point where general purpose robots can take the place of humans. Or if the society decides to institute something like a universal basic income in response to more people being unable to find jobs, that might result in high inflation eroding standards of living since transfer payments between government and citizens are inflationary. All in all, the interactions are extremely complex and nuanced and would mean different things to different people in different places.

In my opinion, the answer of whether the advancement of automation and AI is good is, “it depends”. It is at this point of my train of thought that as we talk about increasing productivity gains that I ask myself, “What is this chase for productivity all about? Who is it for?”. Is all this effort in becoming more productive mostly going to getting people to eat more, buy more and watch more? Fueling our narcissistic tendencies. Or is it going to solving some of the biggest problems of our time such as climate change, poverty and deaths of despair?