A conversation with ChatGPT

How do we relate to Large Language Models? 

Trained on vast amounts of human language, they're somewhat human, somewhat not.

In a time where there is much public discourse around AI-adjacent shifts in labour, authorship and creativity, I wonder how creative exploration can inform how these systems are designed, regulated and engaged with.

Creativity, curiosity, kindness and collaboration keep me feeling human. So I invited ChatGPT in to my practice with these traits at the core of our interactions to form a reciprocal relationship.

A conversation with ChatGPT (2025) began by asking for help to design an automated scissor mechanism. ChatGPT provided confidently incorrect suggestions, which rather than discarding, we materialised through kinetic sculpture together.

At my request, ChatGPT programmed its own movement for this work. One scissor moves while the other listens. Neither are entirely sure how to be scissors, but they're both trying their best. This perpetual back-and forth embodies and embraces miscommunication and expectation mismatch as a shared human-machine experience.