Chat GPT, and the myriad additional apps built on it, are taking the world by storm. On the one hand, we see utopian suggestions that generative AI will let everyone communicate directly with machines in plain English. It will foresee machine errors, schedule our work time for us, and make our workers happier and more productive.
On the other hand, there are just as many dystopian predictions that generative AI will be hijacked by terrorists, make its own decisions that will inadvertently doom humanity, and certainly that it will take over all our jobs.
In an article called “AI and the automation of work,” Benedict Evans suggests that it might just be more automation.
Learn from history
“We should start by remembering that we’ve been automating work for 200 years,” Evans points out. “Every time we go through a wave of automation, whole classes of jobs go away, but new classes of jobs get created. There is frictional pain and dislocation in that process, and sometimes the new jobs go to different people in different places, but over time the total number of jobs doesn’t go down, and we have all become more prosperous.”
Evans brings out a couple of economics concepts demonstrate why it is that automation in the past ended up making jobs more efficient, and instead of meaning that companies hired fewer workers, they hired more — but in new jobs. Read the article to get the economics on it, but if you have some background in history you can look back and agree. As Evans puts it, “So, all of this is to say that by default, we should expect LLMs to destroy, displace, create, accelerate and multiply jobs just as SAP, Excel, Mainframes or typewriters did. It’s just more automation. The machine lets a person do 10x the work, but you need the person.”
And as we know, work expands to fill the time available. When housewives, whose labor is generally unpaid so it simplifies the equations, started having access to washing machines and vacuums, their jobs changed enormously. However, they didn’t have less housework to do, because standards for housekeeping rose. Equally, when you can put 300 paper cups together in a minute, you want to make and sell more paper cups. You don’t just relax and make do with the same number of paper cups you made before.
Of course, that’s a while other problem.
Then what about the future?
Evans acknowledges that these ideas only work if we don’t have a machine that can actually do everything a human can do. After all, Chat GPT makes lots of mistakes. Anything AI generates is just a prediction of the most likely words or strings of code, not an actual answer based on actual knowledge. That doesn’t exist outside of human beings. Maybe some day it will, but sufficient unto the day is the evil thereof.
In the meantime, well, it’s just automation.