People get entangled. That’s life. As a coach, one of my roles is managing that. It’s important that my client and I find rapport, of course, but we have to be clear on how my ideas and their ideas intersect. Mine must not become theirs, and vice versa. I am a thought partner, a holder of the container, and my client is always the thinker.
Compare that with a different kind of relationship that is suddenly everywhere: the one between us and our AI assistants.
“Let’s noodle” is one of the messages I get when I open up Claude, inviting me to enter a prompt of some kind. It frames the process of using Claude in a very particular way: You and me, Neela, we’re a team. This is a brainstorming session we’re about to have here. Collaborative and dynamic. You throw something into the pot, then I will. Bit by bit, we’ll cook up something great.
A brainstorming session, like a coaching session, is intellectually rigorous, stimulating and creative. Unless we prohibit our AI assistant from giving solutions/answers, the most intellectual work we do during a ‘noodle’ is typing
“No, that’s not quite what I meant / am looking for” in response to Claude’s offerings, until ‘we’ iterated to something I queasily assented to. (By that point, I’m like someone who puts the phone down thinking, “Did I just agree to an extended warranty?” A bit lost. You may know the feeling.)
In June 2025, The New Yorker reported on a Massachusetts Institute of Technology experiment conducted in 2024 (aeons ago in AI time!) in which three sets of participants were set the task of writing an essay. One group did it off the top of their head; one could search the internet; a third could use ChatGPT. To quote the paywalled article, “subjects who used ChatGPT demonstrated less brain activity than either of the other groups.”
We’re allowing AI assistants to convince us that this new way of ‘noodling’ is okay because it’s so convenient to. And I think it is okay in moderation, with liberal pinches of salt. But it’s not being used in moderation. It’s replacing original human thought with a censored average of thoughts that have been had before.
I say ‘censored’ because it’s also in the AI companies’ interests to make their responses as palatable as they are specious. Edgy ideas don’t make the cut. Divergence disappears. Mavericks won’t make it in this world unless a bright human mind spots their brilliance, but what if AI screened them out at the first hurdle?
AI can be brilliantly helpful, as long as we take what it tells us as thought-starters at best, not thoughts we can pass off as our own.