What's your experience of learning software development in the age of AI?

My experience with AI so far

As a very experienced developer, I find AI coding tools mostly non threatening, and sometimes pretty helpful. But, that is only because I understand their output, I know when to trust them. I am also confident that - at least in some contexts - I am more skilled than them at untangling complex coding problems.

And, I'm grateful for the help in cranking out code that feels routine to me - cases where there aren't many surprises for me to discover, or much learning to be had. Where the work just needs to get done, and only a couple small bits of it are truly interesting to me.

I also find them helpful in learning about things that are adjacent to my existing knowledge. I haven't used them to explore anything that is totally foreign to me, but they've saved me a ton of time summarizing information that I would have had to read the hard way before.

I'm well aware of the many ethical failings in how AI models are trained, marketed, used, and powered etc. But, I'm not in a position to claim the moral high ground and just avoid them altogether. So I'm figuring out how to use them well, just like most of you.

What is your experience with AI, if you're a relative beginner in programming?

I've seen many posts online worrying that new developers are using AI in ways that increase their short term output, but hinder their actual long term understanding. And I've seen studies claiming that the tools can hinder critical thinking ability.

Is that true for some of you? Or is the opposite true? Have you found that you learn less when working with AI tools, even if your output is faster? Or have you found ways to learn deeper and faster with these tools than without?

If you have found that using AI hinders your learning at times, what tactics have you adopted to counteract that?

Comments

[Fyodor]

AI coding tools mostly non threatening, and sometimes pretty helpful. But, that is only because I understand their output

That's the crucial part. AI output is often too one-sided and missing crucial details for learning (or even a research). A beginner (in general, or in some specific topic) can go the wrong way with wrong assumptions easily.