As you have no doubt noticed, we are in the midst of a feeding frenzy regarding something called generative artificial intelligence. By now an army of normal people – and economists – are surfing a wave of irrational enthusiasm about its transformative potential. This is the newest new thing.
Two antidotes are recommended for anyone with a fever. The first is the excitement cycle monitor produced by consultants Gartner, which shows technology currently at the “peak of inflated expectations” before a steep decline into the “trough of disappointment.” The other is Hofstadter’s law, which relates to the difficulty of predicting how long difficult tasks will take; This law states: “It always takes longer than you expect, even if you take Hofstadter’s law into account.” Just because a powerful industry and its media supporters lose their courage on an issue does not mean that it will sweep through society like a tsunami. Reality moves at a slower pace.
In the Christmas issue, Economist There was an instructive article titled “A Brief History of Tractors in English” (an understated tribute to Marina Lewycka’s 2005 comic novel), Brief History of Tractors in Ukraine). The article set out to explain “what the tractor and the horse tell you about productive AI.” The lesson was this: Tractors are ancient, but they took a long time to transform agriculture. There are three reasons for this: Early versions were less useful than their supporters believed; their adoption required changes in labor markets; and farms had to renew themselves to use them.
So history shows that whatever transformations AI hype traders envision, they will happen more slowly than they expect.
But there is one possible exception to this rule: the work of computer programming or software writing. Ever since digital computers were invented, people needed to be able to tell machines what they wanted them to do. Since machines do not speak English, programming languages have evolved over the generations: 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, is a tedious process for many people. And programming has become a kind of secret trade, as the great Donald Knuth’s title to the first book in his seminal five-volume guide suggests: The Art of Computer Programming. As the world became digital, the craft became industrialized and renamed “software engineering” to belittle its craftsmanship roots. But mastering it remained a secret and valuable skill.
Our evidence shows that programmers are taking to the help of AI like ducks to water
And then came ChatGPT and the surprising discovery that it could write software as well as create seemingly comprehensible sentences. Even more remarkable: You could summarize a task in simple English commands, and the machine would write the Python code needed to accomplish it. Most of the time the code wasn’t perfect but could be debugged with more interaction with the machine. And suddenly a completely new possibility opened up; non-programmers were able to instruct computers to do things for them without having to learn how to talk.
Inside New Yorker Recently programmer James Somers wrote an elegiac article about the consequences of this development. “Knowledge and skills that traditionally took lifetimes to master are being gobbled up,” he said. “Coding has always seemed like an extremely deep and rich field to me. Now I find myself wanting to write a eulogy for it. I think about Lee Sedol all the time. Sedol was one of the world’s best Go players and a national hero in South Korea, but he is now best known for losing to a computer program called AlphaGo in 2016.” According to Somers, Sedol “seemed to be buckling under the weight of a question that was beginning to feel familiar and urgent: What will happen to this thing to which I have given so much of my life?”
This seems a little strange to me. Our evidence shows that programmers take to the help of AI like ducks to water. A recent survey of software developers, for example, finds that 70% are using or plan to use AI tools in their work this year, and 77% have a “favorable or very positive” view of these tools. They see these as ways to increase their productivity as programmers, speed up learning, and even “increase accuracy” in writing computer code.
This does not seem to me like admission of defeat, but rather the attitude of professionals who see this technology as “power steering of the mind”, so to speak. In any case, they do not sound like horses. Economist‘s story. But just as the tractor eventually transformed agriculture, this technology will eventually change the way software is developed. In this case, software engineers will need to look more like engineers and less like craftsmen. The time is coming (says this engineer-columnist).
What was I reading
Smart move?
Here’s a great explanation by Gary Marcus on his Substack blog about AI companies lobbying to be exempt from liability for copyright infringement.
control mechanism
A really thoughtful piece by Diana Enríquez on the Tech Policy Press website about what it’s like to be “managed” by an algorithm.
their heads are closed
Nice article on Margaret Atwood’s Substack for films about the French Revolution, starting with Ridley Scott’s Napoleon.