Paired prompting

I was talking to Tracey Camilleri yesterday about workplace design in the the age of LLMs, and then Chris Noessel about his forthcoming book, Cognitive Debt (and the potential for Cognitive Surplus), and I have a head full of ideas and desk full of new cards.

Of those, this one felt worth trying to quickly articulate more formally – what might paired prompting as a deliberate strategy look like in an organisation?

Instead of everyone having access to a personal LLM that they use to tease out information, and then relay this to colleagues, what if this was strictly a co-operative effort?

I’m taking direct inspiration from pair programming of course:

Pair programming is a software development technique in which two programmers work together at one workstation. One, the driver, writes code while the other, the observer or navigator, reviews each line of code as it is typed in. The two programmers switch roles frequently.

And it’s a little like that thing in movies where if a machine is sufficiently dangerous, you need two keys and two people to operate it…