…some historical precedents.
There is currently much discussion of the impending effects of artificial intelligence and robotics on employment, the economy, and our society as a whole. (here, for example)Â I think it’s useful to look at some historical precedents, always keeping in mind the caution that ‘past results do not guarantee future outcomes.’
Peter Gaskell’s book Artisans and Machinery is about the effects of the industrial revolution, as seen by a contemporary observer. I reviewed and excerpted it here, along with some much later commentary by the British writer and scientist CP Snow.
My post Attack of the Job-Killing Robots (three-part series) is a 30,000-foot view of the history of automation over the centuries and of some resulting automation panics.
Your thoughts?
There are two thrusts for the accelerated use of robots, humanoid or not. First is in the twin fields of construction and agriculture. What will the legions of immigrants do when these fields are near 100% automated? Go to Service industry? A small percent may find temporary refuge as cleaners, nannies and home owner gardeners, perhaps care givers for the aged. Most will be obsolete. Think that is n the far horizon? Already a supplier of major agricultural machines sells GPS and computer-driven combines that plow and harvest thousands of acres of grain crops at a time, with only a ride-along operator.
Same company is beta-testing large scale strawberry pickers and weeding devices.
The second big thrust will be to automate routine office work. Watch out QA departments. SOPs, training (soon to be unnecessary), lower level financial and legal briefs. On and on. Recently some judges have allowed the presentation of AI-generated legal briefs, with “appropriate human review”. Who has not noticed that telephone trees in customer service have become increasingly sophisticated. Where will these office workers go? Database management and development, lower level coding can only absorb so many.
And if forgot to mention the impact on the medical profession. Some AI programs have been shown superior in diagnosing routine conditions. But surgeons? Mostly a thing of the past. Well established surgeries, dental and eye corrections will bee performed by robot while you under anesthesia. Bet on it. No more operations on the wrong leg.
So, one analyst at Goldman Sachs predicts the immanent demise of 300,000,000 jobs. While at the same time, I have no doubt, others there are touting investments that depend on those same 300,000,000 continuing to pay mortgages and consume products above bare subsistence levels. All of them can’t be right. This contradiction never seems to bother the consumers of economic prognostication. If you look long enough, you’ll always find an “expert” that agrees perfectly with your own preconceptions and prejudices.
As you point out, automation is nothing new and isn’t going to stop. My prediction is that it will continue to advance incrementally as it has in the past. The factors that govern this remain.
The first is financial, human labor remains far more scalable and flexible than machinery. Try laying off half of a million dollar machining center. I know there are leases that are based on production, but all that does is shift the cost to a different party that still needs to make up the difference, effectively fixing a floor on cost.
Moreover, while a machine has perfectly fixed specifications, humans are adaptable. Humans deal effortlessly with materials and conditions that persistently frustrate machine designers. I remember a story from a couple of years ago about an exciting development in robotics. They had, after much, work managed to program a robot to fold towels. This isn’t a joke, soft floppy materials have bedeviled automation from the beginning. The nimble fingers of the women and children in the mills that Gaskell talks about were there because it took a very long time to invent a machine that could tie knots and feed threads into the machines.
The same is true for “knowledge” work. The march of progress has eliminated the steno pool. So now you find six figure executives spending their time doubling as clerk typists while seven figure executives have six figure assistants doing the same. I have never seen a payroll system that would function without constant human intervention.
I’m still not buying all the AI hype. The systems I see are doing nothing more than cutting and pasting snippets of their training sets that seem to fit whatever prompt they’ve been given. Very facilely and fluidly, but like the “AI” art generators, very impressive at first glance but doesn’t stand up to close scrutiny. Not today and not tomorrow, maybe the day after.
Here’s a rather long video that shows a great deal about the actual state of the art in CNC metal machining.
https://www.youtube.com/watch?v=uWDfSfSBIto
“So now you find six figure executives spending their time doubling as clerk typists”…years ago, one software commented that “The main thing we’ve accomplished with the computer revolution so far is to turn high-paid executives into incompetent clerk-typists”
FWIW, here’s some data on US labor productivity:
Overall year-over-year productivity growth rate
https://www.ceicdata.com/en/indicator/united-states/labour-productivity-growth
Index values for the nonfarm business sector:
https://fred.stlouisfed.org/series/OPHNFB#0
https://www.wsj.com/business/robot-straighten-candle-wick-difficult-e1c80bb8. It turns out that automating straightening candle wicks is hard.
Once you have the process there’s one (or only a few) way(s) to build a thing, but the much larger variety of ways it can break demands a variety of ways to repair it. AI might (might) help with diagnosis, but I doubt that robotics and AI will reliably repair things, unless you’re satisfied with BFI bulk replacements.
I’d not trust an AI robot to try to dust my desk. There’s a reason things are in the position they are in.
Back about half a century ago, there was talk of a “leisure society”. Automation would enable the working week to be cut to only 2 or 3 days, leaving people with lots of time for their preferred recreational activities. While it is true that we all today work many fewer hours than people did prior to the Industrial Revolution, the projected “leisure society” has still to appear.
AI is effectively outsourcing work to local machines — as opposed to outsourcing work to remote human beings in India or China. Our Betters have been outsourcing to other countries for almost half a century, so we can look at that real world evidence to make a prediction about the impact of AI:
it will likely involve the growth of overhead, the creation of millions of “jobs” writing & implementing government regulations and of other “jobs” of smart lawyers & accountants devising ways to exploit or avoid those regulations.
None of that busywork adds to the real productive economy. Remember — Production Precedes Consumption. Thus, the likely impact of AI will be to accelerate the coming inevitable collapse of the economy.
I’ve got a quarter century of involvement in student robotics matters.
On the plus side the increasing sophistication of automation has been critical in “re-shoring” manufacturing. If you want to make something of quality, something that is not needed in the tens of millions, something that you can get next week….don’t look to the East.
On the neutral to negative side, there is still an insatiable appetite for mass produced krep from China. That will take a change in us, not in the tools we reach for.
Most of the AI I’ve seen in the graphic realms is disturbing but improving. I don’t like it. In the written fields I can usually identify the mixture of confidence, con and anodyne that is the tell for ChatGPT and its clones. Maybe writing will go the way of beer. You can get bland, watered down swill (that subsidizes Progressives!) or something with a bit of character. So far I’m not going to be fooled by the difference between Weizenbock and Bud Lite. With head to head comparisons (of writing, not foam) I am still doing well at finding Skynet, especially if you ask a few simple follow up questions.
Anyway, the students I’m working with are all ridiculously employable.
T
When Goldman Sachs tells you something, the first question is: Does Goldman Sachs believe it and are they acting in conformance to that belief? The second, most especially when they share their usually very expensive insight for free, is: Why am I the beneficiary of this largess?
The answer to the first is very obviously: No. They do not believe that 300,000,000 workers, 2/3 portioned between North America and Europe with the rest scattered through the rest of the world, will be shown the door by summer. This would produce unemployment rates somewhere near 50%. The repercussions would be noticeable.
The second, on reflection, is also pretty obvious. Goldman Sachs is in the business of promoting, tout is such an ugly and legally actionable term, various companies to high net-worth investors. I am pretty sure some of these companies can plausibly, or maybe not so plausibly, be described as “occupying the AI space”. The goal is to convince these investors to buy the sizzle while ignoring the fact that all these companies have generated impressive accounts payable with accounts receivable, if any, all in the future. Also ignoring, that even if some portion of this prediction comes to pass eventually, there will be ample opportunity for many companies to be born and proceed into bankruptcy while others succeed.
Also, bear in mind that unless your name is Bezos or Gates, you’re hearing about this “opportunity” because others with much more money than you have already passed. By the time you or I hear of it in the open press, pretty much anybody with money has already passed.
Finally, how many of these 300,000,000 about to become redundant should have already been displaced by any or all of the preceding waves of office automation over the last 70-100 years? Yet they continued to be employed in apparent contradiction to economic utility. How many pages to turn or clicks to find the latest horror story of an informatics transformation gone so wrong as to threaten the solvency of the company involved? How many of these stories have you read? I lost count about 40 years ago.
So, no, the world is not about to end because of AI, especially, AI that may not actually exist. I make no such sanguine prediction about the continued existence of many investments in this “space”.
Here’s the Goldman report, or at least some Goldman report on the subject:
https://www.key4biz.it/wp-content/uploads/2023/03/Global-Economics-Analyst_-The-Potentially-Large-Effects-of-Artificial-Intelligence-on-Economic-Growth-Briggs_Kodnani.pdf
and here is a mediocre (surprise!) Forbes summary:
https://www.forbes.com/sites/jackkelly/2023/03/31/goldman-sachs-predicts-300-million-jobs-will-be-lost-or-degraded-by-artificial-intelligence/?sh=4b500591782b
The Forbes pieces says “The investment bank estimates 300 million jobs could be lost or diminished by this fast-growing technology”…it doesn’t pay much attention to jobs that could be created or enhanced, although it does cite Goldman as saying that automation creates innovation, which leads to new types of jobs.
Forbes also says that (businesses) “deploy their resources toward building and growing businesses, ultimately increasing annual global GDP by 7%”…this is kind of bizarre…the ultimate effect of AI will be only a global GDP improvement of 7%? Or do they mean 7% a year? And if the latter, starting when and for how long?