This might sound like laziness, but it's actually the opposite. The lazy choice is sticking with what I already know and staying comfortable in my current workflow. The real challenge is shipping more while maintaining or even improving quality.

As humans, we need to stop being the bottleneck. "AI slop" sounds like an excuse at this point, if the output isn't good enough, use AI to fix it. Prompt again. Fix the algorithm or data structure. Generate 800 test cases if needed. Adjust the style. Add another agent to review the code.

The real question now is how many agents I can run at the same time, how much i can iterate, and what the most efficient pipeline is to ship fast with confidence.

As a human, I'm still responsible for the output. My job is to express decisions with clarity and fearlessness, without being the bottleneck.

I think we need to detach our egos from something that is now cheap. Writing code is no longer the hard part of building something, the hard part is finding the right prompts, avoiding the "you are absolutely right" loops and knowing when to give the model more autonomy to find the truth.

That said, I think learning to code is still a good investment. anything that helps you understand how the world works, like math, code, or physics, is worth learning.

But if you have a forklift and you're still carrying bricks up the stairs by hand, you're wasting time and misusing resources.

There's a massive opportunity right now. So much code isn't being written, not because it's impossible but because LLMs don't have goals. They need direction and as programmers, our job is becoming less about writing code and more about giving intelligence a purpose.