Great piece, Debbie. I agree that writing is thinking and that prototyping with LLMs often bypasses the cognitive work entirely.
But I wonder if there's a next-level skill emerging here. Using model output not as a substitute for thinking but as input to it. If my unassisted reflection produces ideas at a certain level, what happens when I inject higher-quality tokens into that process, not as answers, but as material I have to wrestle with. Does that elevate where I land?
Is deliberate interleaving worth developing? LLMs as intellectual friction rather than shortcut?
This is a great reframe! The distinction you're drawing between "substitute" and "input" is exactly right. Using AI to generate material you have to wrestle with is fundamentally different from using it to skip the wrestling entirely. But this only works if you already have good baseline judgment to begin with.
Great piece, Debbie. I agree that writing is thinking and that prototyping with LLMs often bypasses the cognitive work entirely.
But I wonder if there's a next-level skill emerging here. Using model output not as a substitute for thinking but as input to it. If my unassisted reflection produces ideas at a certain level, what happens when I inject higher-quality tokens into that process, not as answers, but as material I have to wrestle with. Does that elevate where I land?
Is deliberate interleaving worth developing? LLMs as intellectual friction rather than shortcut?
This is a great reframe! The distinction you're drawing between "substitute" and "input" is exactly right. Using AI to generate material you have to wrestle with is fundamentally different from using it to skip the wrestling entirely. But this only works if you already have good baseline judgment to begin with.