by As you’ve no doubt noticed, we’re in the midst of a feeding frenzy about something called generative AI. Legions of ordinary people – and economists – are riding a wave of irrational wisdom about its transformative potential. It’s the newest, newest thing.
For anyone suffering from fever, two antidotes are recommended. The first is the hype cycle monitor produced by the consultants Gartner, which shows the technology currently at the “peak of inflated expectations”, before a sharp decline in the “effort tank”. The other is Hofstadter’s law, about the difficulty of estimating how long difficult tasks will take, which says “It always takes longer than you expect, even when you put Hofstadter’s law included”. Just because a powerful industry and its media boosters are losing their marbles about something, doesn’t mean it will sweep like a tsunami through society at large. Reality moves at a more leisurely pace.
In its Christmas edition, the economist there was an educational article entitled “A short history of tractors in English” (a rare homage to the 2005 comic novel by Marina Lewycka, Short history of tractors in Ukrainian). The article aimed to explain “what the tractor and the horse tell you about generational AI”. The lesson was that although tractors go back a long way, it took aeons before they changed agriculture. Three reasons for that: early versions were not as useful as their supporters believed; changes in labor markets need to be adopted; and farms themselves needed to be renovated to use them.
History suggests, therefore, that whatever transformations AI hype merchants are predicting, they will arrive more slowly than they expected.
There may be one exception to this rule, however: computer programming, or the business of writing software. Since the invention of digital computers, people have had to be able to tell the machines what they want them to do. Since the machines did not speak English, generations of programming languages emerged – machine code, Fortran, Algol, Pascal, C, C++, Haskell, Python etc. So if you wanted to communicate with the machine, you had to learn to speak Fortran, C++ or whatever, a tedious process for many people. And programming turned out to be a wise craft, as implied by the title that the great Donald Knuth gave to the first book in his five-volume basic guide, The art of programming. As the world became digitized, this craft became industrialized, and was rebranded as “software engineering” to reduce its occupational origins. But her mastery remained a prudent and valuable skill.
Such evidence as we have suggests that programmers are taking to AI assistance like ducks to water
And then came ChatGPT and the amazing discovery that he could write software as well as compose clear sentences. Even more impressive: you could set it a task in plain English prompts, and the machine would write the Python code needed to complete it. Often the code was not perfect, but could be debugged by further interaction with the machine. And suddenly there was a whole new prospect – that non-programmers would be able to instruct computers to do things for them without having to learn a computer speaker.
In New York recently, programmer James Somers wrote an elegant essay about the implications of this development. “Bodies of knowledge and skills that traditionally took lifetimes to master are being swallowed up,” he said. “Coding has always felt to me like an infinitely deep and rich domain. Now I want to write a recommendation for him. I keep thinking about Lee Sedol. Sedol was one of the best Go players in the world, and a national hero in South Korea, but is now best known for losing, in 2016, to a computer program called AlphaGo.” To Somers, Sedol “seemed to be weighed down by a question that began to feel familiar, and urgent: What will become of this thing that I’ve given so much of my life to?”
That sounds a bit OTT to me. Such evidence as we have suggests that programmers are taking to AI assistance like ducks to water. A recent survey of software developers, for example, found that 70% are using, or plan to use, AI tools in their work this year and 77% have “favorable or very favorable” opinions on these tools. They see them as ways to increase their productivity as programmers, speed up learning and even “improve accuracy” in writing computer code.
This is not my view, but the view of the professionals who see this technology as “directive power for the mind”, as they say. Anyway, they don’t sound like horses economist‘ a story. But just as the tractor ultimately changed agriculture, this technology will transform the way software is developed. In that case software engineers will have to be more like engineers and less like craftsmen. About time too, (says this engineer-cum-columnist).
What I was reading
Smart move?
Great discussion by Gary Marcus on his Substack blog on the lobbying of AI companies to be immune from responsibility for copyright infringement.
Control mechanism
A very thoughtful piece by Diana Enríquez on the Tech Policy Press website about how to be “managed” by an algorithm.
Out with their heads
A beautiful post on Margaret Atwood’s Substack on films about the French Revolution, starting with Ridley Scott’s Napoleon.
