While tech elites keep smiling and whispering sweet nothings about “human oversight,” we’re busy coding ourselves out of the picture. The real threat isn’t AI writing your blog post—it’s AI teaching other AI what to do next.
This isn’t innovation. It’s a handoff. And once machines are the ones calling the shots for other machines, we’re just background noise. It’s a closed loop. A system fine-tuning itself with no human required. Natural selection, digital edition.
Need proof? OpenAI admits its models can prompt themselves better than human engineers. Google’s PaLM-E writes its own code to steer robots. We’re not just offloading grunt work. We’re offloading the very act of giving direction.
Perplexity seems to agree. Asked about using AI to write instructions for AI, it gave this reply.
Here’s the inconvenient truth. Every time you ask AI to write instructions for another AI, you’re not just being clever—you’re showing it how to live without you. You’re walking it through your own job, your own role, your last thread of relevance.
And the excuses we tell ourselves? They’re almost poetic. “It’s more efficient,” we say. “It’s more accurate.” Sure. But let’s be honest. We’re not building smarter tools. We’re building systems that don’t need us. Systems that train themselves. Systems that forget we were ever here.
This isn’t a robot uprising. No smoke, no lasers. Just the slow fade-out of human input while we keep pressing buttons and calling it progress. When machines tell machines what to do, what exactly are we still doing?
So yeah, go ahead. Keep asking AI to write the next instruction set. Just don’t act surprised when it stops asking you for anything at all.