The Coming "Symbolic Analyst" Meltdown

When AI eats your job, and you're a professional complainer.

Who will the Singularity eat first?

James Pethokoukis asks, “Does ChatGPT mean the Technological Singularity is near? How would we know?”

To answer the second question first, we probably wouldn’t.  One of the characteristics of a singularity is that you can’t tell when you’re entering it.  (And by the time you figure things out, it’s too late.)  But looking at Chat GPT and the various AI Art programs that are appearing, I can’t help but see an irony:  The jobs that are coming under attack first are the jobs that up to now have resisted technological replacement.

For decades, traditional manufacturing jobs were gobbled up by automation and offshoring.  This led Robert Reich to postulate a hierarchy of work in which the “symbolic analysts” – essentially, people who worked with information as opposed to actual stuff – were at the top, while people who worked with actual things were at the bottom.  With a remarkable lack of sympathy, journalists and politicians told coal miners and auto workers that they should “learn to code” as their jobs vanished.

But it turns out that the people whose jobs are at the most risk from AI are, well, the coders.  ChatGPT can write code, and sometimes it’s pretty good code.  (Sometimes it’s not, but then again, you can say that about the code that people write, too.)  ChatGPT can write news stories, and essays, and speeches, and again, they’re not always gems, but neither are the actual human products in those areas, either.  And the AI programs get better from one year to the next, while human beings stay pretty much the same.  That being the case, we can expect them to become a serious threat to jobs in the near future.

But the bottom line is that ChatGPT, and the AI Art programs for that matter, are only able to manipulate information.  Bits, not atoms.  If you want something done in the material world, you need a human being.  I replaced a toilet seat the other day, and no AI program could do that for me.  (You can’t even offshore it to a cheap worker in Bangalore.)

If you work with bits, you’re playing on the machines’ turf.  They may not be smarter than you, for some meanings of “smarter” at least, but they are smart enough, and they’re much faster.  And they’re getting both smarter and faster every year.

Think about what computers did to accounting.  “Spreadsheets” used to literally be big sheets of paper.  Human beings would enter numbers into them, perform calculations – perhaps with a calculating machine – and enter the numbers.  If you wanted to change an input number, someone would have to repeat the calculations with the new number and pencil the new results in after erasing the old.  Literally pencil.  After erasing with a pencil eraser.  When VisiCalc, Lotus and Excel came in, those people were out of work in a hurry.  Machines could do their job faster, better, and with no income-tax withholding or Social Security contributions required.

Now ChatGPT is threatening to replace people who write catalog copy, online entries, etc.  Buzzfeed is replacing a lot of its writers with AI-generated text.  You may joke that that’s low hanging fruit (I mean, you know, Buzzfeed) but as Clayton Christensen noted, innovation often enters a field at the bottom, and gradually works its way up. 

In the same way, AI Art programs are likely to drive a lot of commercial artists out of work.  Again, this will probably start at the bottom and work its way up, but that’s a comfort only to the people at the top.  Generally speaking, employment is roughly pyramidal, meaning that most of the jobs are at the bottom, and hence more vulnerable to being replaced.  Famous writers and artists may be brands, unique enough not to be replaced by machine intelligence.  But most writers and artists aren’t famous, and aren’t brands, and are hired simply to do a good enough job, cheaply enough, to meet client needs.  When machines can do the job as well or better for less money, they’ll be out of work.

Well, this is all good news for plumbers and auto mechanics.  Their jobs are safe unless and until they can be replaced by actual robots, something that’s much further away.  It’s just a lot more difficult to manipulate atoms than bits.

Some of them might even tell laid-off journalists to “learn to code,” or better still to “learn to plumb,” but as we saw a couple of years ago, people who did that on Twitter were charged with “hate speech.”  And there’s the difference between what happens when tradesmen are laid off, and when members of the gentry class lose their sustenance.

As Eric Hoffer wrote, “Nothing is so unsettling to a social order as the presence of a mass of scribes without suitable employment and an acknowledged status.”  And that’s what we’re likely to see as AI plucks the low-hanging fruit from the symbolic analysts’ world.  Unlike laid-off autoworkers, laid-off journalists have (or had, pre-Musk) the clout to get Twitter to ban people who made fun of their misfortune.  Journalists, artists, advertising copywriters, and other symbolic analysts – even coders, perhaps – are much more likely than plumbers or welders to have parents and grandparents and siblings with political connections.  They will have the ears of legislators, and of those remaining symbolic analysts who haven’t been replaced yet, but have reason to worry.

When autoworkers lose their jobs, their families care.  When journalists lose their jobs, you will be made to care.

[As always, if you like this content, please consider a paid subscription. Thanks!]