Turning your full work output over to AI is a massive gamble, for multiple reasons.
One, there's no guarantee that it can do everything you need it to be able to do, especially in its current state.
Two, there's no guarantee that you won't get stuck in a situation that it can't fulfill, leaving you powerless to step in because it's done too much of the project already and you have no idea how things work under the hood.
For skilled devs, the ONLY guaranteed outcome right now is that you'll get far dumber, and you'll lose all the dev skills, abilities, and experience you've worked so hard to acquire up to this point.
That is a 100% guaranteed fact. Removing yourself from the practice of dev makes you progressively more irrelevant each day that goes by.
In fact, this actually might be the first time in human history where an entire industry of highly technical people willingly raced from the top of the skill ladder to the bottom of the skill ladder.
They seem to be doing this because they believe in a world where AI progresses to the point of being autonomous and nearly flawless. But in that world, they're no longer really needed anyway, so what's the point?
They also seem to believe that the costs won't skyrocket, which is a third gamble.
What happens if costs do increase frantically, though (knowing that all the AI companies are currently taking massive losses)?
If AI doesn't pan out exactly the way people hope and we find ourselves in a world where the best (and most affordable) approach is actually a hybrid, AI-assist approach to development, what will all the people who turned their entire workload over to AI do at that point?
And how will all the products built by cheap AI get maintained in a world where AI is now far more expensive?
Conclusion 1: Going full-AI undoubtedly retards your actual skills. That's a huge PERSONAL risk that relies on some big gambles.
Conclusion 2: Building and selling products where iteration and maintenance require agents at the currently untenable cost structures is a huge risk to you and everyone you sell to.
Conclusion 3: The most responsible thing is to remain in command of the code. Take a hybrid approach. Stay engaged with the practice of development. Maintain your skills. Engineer your products properly.
For the people who didn't have many skills to begin with, we see why you're so hyped. You couldn't do something before, but you can seemingly do it now. You feel a newfound sense of empowerment and like you cheated the process of having to learn.
That's great. Have fun. But stop trying to convince people who have acquired tremendous skill to abandon those skills when the actual consequences and outcomes of that decision are completely unknown and terribly risky.
And remember, a lack of skill in the thing AI is doing means the person also lacks the skills to assess whether the AI is truly capable of it.
Right now we have a lot of people who aren't very skilled trying to assess the skills and abilities of AI. Dunning-Kruger is still a very real thing.
Be careful out there.