D751. The Impact of AI on Programming and Life

When I learned to program for the Sperry-Rand Univac I in 1957, upon arrival at Harvard--and learned to operate that huge machine, which was much more fun--this was a revelatory experience. Using machine language, every instruction was a tiny step. (The memory held only 1000 instructions, or numbers.)

When push came to shove--I didn't like it. Slow, tedious, repetitive, unrewarding. I correctly envisioned that what computers could do for the next 10 or 20 years was very limited, and I turned in a different direction--far away from the Harvard Computation Lab, which had been my "first love." I try to be practical.

Since then--a series of new eras. In the 21st century, there has been an increasingly widespread belief that every smart, "quantitative" kid should learn coding in elementary school. Maybe that's was no more useless than the Math. (Never mind the Social Studies!)

Still, the latest "news" is that unemployment is now higher for new coders and for Computer Science graduates than it is in many other fields. (I don't recommend Journalism, either. Physics graduates also have high unemployment.)

This drop in demand for coding skills is attributed to the increasing role of AI in coding. The CEO of Microsoft says that 30% of the coding there is already carried out by AI. Programmers at the giant tech companies can tell a somewhat different tale. They say that the use of AI has led to 50% cuts in staff, coupled with speed-ups. What used to be done in weeks now must be done in days.

Worse, using AI, their focus has largely turned from writing code to reading and correcting code created by the AI. They don't like that either! (Odd, I prefer editing text to writing it.)

Man-machine integration will continue to improve as we learn to define AI's tasks with greater precision. How about this, from Anthropic: "Small shifts in how you work with Claude" (their AI) "can help you preserve more energy for what truly matters." How promising, "what truly matters." The next thing you know, computer jocks might even be offered a possible route to a "meaningful life," whatever that is.

Until that happens, their underlying complaint is that AI has deprived them of the dignity and centrality that they held for so long in these monster organizations. They're no longer the "source" of the companies' "products." It's the difference between making a suit of clothes from a bolt of cloth, versus merely doing "alterations" on what comes out of the factory.

Until the AI is also trained to do most of the debugging--which is the next, predictable advance. So, how do we advise young students now? To study AI? About a dozen major US institutions already offer a major in AI. So, then you can join the armies of "engineers" that have spent literally decades trying (and failing) to make AI drive cars? That actually sounds very chancy--and unrewarding even in the short run.

The original Industrial Revolution wasn't so great for "human dignity" (living a "meaningful life?") either. Ah, to be young again!