AI and Fear of Job Losses

As AI gets better at more and more things, it invokes fear. Of job losses. Ben Evans wrote this excellent article on the topic. On the one hand, he says:

“Every time we go through a wave of automation, whole classes of jobs go away, but new classes of jobs get created… over time the total number of jobs doesn’t go down, and we have all become more prosperous.”

But that is just historical data. In practice we worry:

“When this is happening to your own generation, it seems natural and intuitive to worry that this time, there aren’t going to be those new jobs. We can see the jobs that are going away, but we can’t predict what the new jobs will be, and often they don’t exist yet.”

 

Then there’s what economists called the “Lump of Labour fallacy”:

“The Lump of Labour fallacy is the misconception that there is a fixed amount of work to be done, and that if some work is taken by a machine then there will be less work for people.”

Say, machines reduce the price of production of shoes drastically. Yes, lots of shoemakers will lose their jobs. But since the new shoes cost less, people will have more money left with them now. This extra cash will create a demand for some new product or service to fulfil, which will create new jobs.

 

Did calculators wipe out accountants? No, once something becomes cheaper and more efficient, we often find new uses for that capability. New uses translates into new jobs.

“It also tends to mean that you change what you do. To begin with, we make the new tool fit the old way of working, but over time, we change how we work to fit the tool.”

 

But is this time different? There are two reasons many feel that way. Evans looks at each reason in detail.

 

Reason #1: It is happening so much faster than ever before.

True, but… Rolling in any new tech into the workplace takes time. Years, not weeks. At the workplace, a lot of existing knowledge involves not just technical but also institutional knowledge. Next, he points out that companies don’t buy technologies; they buy products.

“I don’t think a text prompt, a ‘go’ button and a black-box, general purpose text generation engine (like ChatGPT) make up a product, and product takes time.”

Or, as he puts it in a snarky way:

“The future takes a while, and the world outside Silicon Valley is complicated.”

Third, he rightly says current AI is error prone:

“People call this hallucinations, making things up, lying or bullshitting - it’s the ‘overconfident undergraduate’ problem.”

 

Reason #2: It looks like a general purpose technology, i.e., it impacts multiple industries (and thus jobs)

Think of Excel, he says. It is a general purpose tool – how many industries did it disrupt? The Internet and smartphone were general purpose, and they have disrupted many industries. But are we seeing unemployment rates much higher today than say, when the Internet got started? Secondly, yes, you could build things on top of each other and cumulatively their effect could be huge, but that takes time. Lastly, and most critically, he says we don’t have general purpose AI… at least, not yet. Each AI today does one thing - art, answering questions, music, and so on. The real world is messy:

“You might also suggest that the idea this one magic piece of software will change everything, and override all the complexity of real people, real companies and the real economy, and can now be deployed in weeks instead of years, sounds like classic tech solutionism.”

 

All that is why Evans ends by saying:

“As an analyst, though, I tend to prefer Hume’s empiricism over Descartes - I can only analyse what we can know. We don’t have AGI (Artificial General Intelligence), and without that, we have only another wave of automation, and we don’t seem to have any a priori reason why this must be more or less painful than all the others.”