Ask Paul: February 16 (Premium)

It's been a crazy, busy, and stressful week, and I suspect no one is looking forward to the weekend more than me. But first, another great set of reader questions.
Slice in time career advice
spacecamel asks:

With all of the industry changes and changes in technology, is programming still an excellent profession to recommend to university students? I am thinking of your story of a group of on-prem admins right before Microsoft moved everything to the cloud. Are we close to a similar moment with things like ChatGPT?

That's a great question. I feel like the answer could change weekly or monthly going forward.

A couple of quick thoughts first.

On one hand, I am sort of impressed (and depressed) by the AI denier crowd. Someone commented the other day to whatever article that AI is a scam being led by Microsoft. And I just want to scream. You're not seeing the asteroid that's careening towards this planet, and that it's not a question of it changing things, it's how fast and how much everything changes. Because everything is changing. There's just too much evidence---meaning shipping products/services that work---to claim otherwise.

On the other, even people like me who see this AI revolution in motion can be surprised by the speed at which this stuff improves almost in real time. If you haven't, look at OpenAI Sora and then silently ponder how that level of capability applied to much simpler tasks will change things. So I'm sitting here, begging anyone to listen about this AI stuff, and what's really happening is far more dramatic that what I imagined. I look at those Sora examples, and I'm at a loss. What does this mean for ... almost anyone who creates content of any kind? I figured I could ride out this career over many more years. Now I have some doubts. Not good.

Anyhoo.

To your question specifically, it's impossible to see too far out with this stuff, but for now, I feel comfortable saying that, given the nature of this work, we will "always" need human beings in the chain. And that, at worst, it's likely that the job changes over time to overseeing, verifying, and continually testing AI-generated code bases to ensure that they are correct in every sense of the word (semantically correct, of course, but also address the need of whatever solution and in the most efficient way possible).

More specifically, imagine you're a developer, and you're tasked with creating a library that does some thing, it doesn't matter what. Up to today, you might have had some level of understanding of how to write this from scratch, from completely to not at all, and you use whatever resources---books, back in the day, Stack Overflow, etc. more recently) to get you to something that at least works. To make this as efficient as possible, you might use built-in IDE tools or third-party tools, debug as needed, test, etc. In the end, you create this thing.

Today, things are a bit different. You might use AI instead of Stack Overflow t...