When does AI stop reflecting us – and start shaping us?

A lot of attention is given to what our prompts reveal about us. But I’m increasingly interested in the other direction.

After hundreds or thousands of interactions with an AI system, what does the interaction itself teach us to ask, expect, avoid or repeat?

At what point are we no longer simply expressing our way of thinking through AI – but also adapting our way of thinking to the system we repeatedly interact with?

I’m curious where others notice that boundary in their own use.