Summary

Charity Majors answers a question about how companies can develop shared AI norms, drawing on Honeycomb's process of starting with open listening sessions before formalizing anything. She frames this as the opening salvo of a planned public dialogue with research psychologist Dr. Cat Hicks about norms, ethics, and learning in the AI era. The post touches on the broader industry anxiety around AI productivity expectations — and the uncomfortable truth that no one yet knows what 'enough' output looks like.

Key Insight

The most effective path to company-wide AI norms is bottom-up listening first — structured open sessions that surface real frustrations and build consensus, so that when formal norms arrive, the cultural groundwork is already laid.

Spicy Quotes (click to share)

• 2

  Start by listening.

• 6

  You might be delivering twice as much twice as fast, yet falling far short of expectations, if your managers, execs, or investors are expecting 10x, 100x or more. Are their expectations unreasonable, or has your output not sufficiently caught up or adjusted for what is now table stakes? The only honest answer is that no one really knows yet.

• 6

  There's a vast sense of unsettledness and insecurity in our industry, with ecstatic possibilities coexisting alongside existential dread, and we're all along for the ride with our meat sacks and Paleolithic psychologies.

• 4

  It also felt like an excellent steam valve to release some of the pent-up anger and frustration people were feeling, to the point that when we did eventually release the AI norms and values documents internally, it felt like the heat had simmered down quite a bit.

• 3

  What feeds greatness, in people and in teams? What are our responsibilities to each other? How do we engage with complex sociotechnical systems, as builders and caretakers, and leave them better off than we found them?

Tone

reflective, conversational, collaborative