After taking a deep dive into collaborating with LLMs (ChatGPT and OpenSeek) while developing software, I came to this conclusion:
Implementation code is a commodity. It's dime-a-dozen. As machines keep getting better at it, our coding skills are rapidly becoming less and less relevant.
Now for the bad news: what is now becoming increasingly relevant is our testing skills. LLMs appear not to have a good grasp at formulating good expectations. This is where a good testing mindset becomes indispensable.
Why is that bad news? Because, from what I know and from what I've seen, a large majority of software developers/coders look down upon testing. To them, testing is a nuisance, something to be avoided or, at best, left for some later time.
That mindset doesn't bode well with the ways the software industry is currently shaping up. I have realized that if I improve my skills when it comes to crafting focused, pointed, atomic, and legible expectations, the machine will serve me well and will produce code that works as expected. I also realized that acquiring that skillset is far from being trivial. I would now claim that it is much harder to become good at formulating crisp expectations than it is to become good at implementing those expectations. And the majority of software professionals I worked with are 100% focused on being good at implementing the expectations. That is where they're eventually going to lose to the machines.
You could be up to something here.
Well done for formulating it this crisply.