Field Notes on Trust and AI

RadiationBox

An independent publication on trust signals, media incentives, and AI-native behavior.

Latest Published Essay

what owning the miss looks like

The sentence that stopped me in Arvind Krishna's letter to IBM investors was only two words: "we faltered." I have seen plenty of executives use similar language while keeping the explanation safely outside their own decisions. Krishna tied his sentence to IBM's judgment, saying the company underestimated the spending shift in front of it and did not adapt quickly enough. Investors now have a management failure they can revisit when the company reports again.

continue reading

Also Published

gen z does not trust social media as much as people think

We hear a version of this from clients: "Gen Z trusts social media too much." The evidence supports the anxiety behind the sentence, not the sentence itself. Pew's 2024 teen work gives the intensity side: nearly half of teens are online "almost constantly." Gallup/UNICEF's 2021 15-24 study gives the trust split: young people rely on social media for information but are least likely to trust social platforms a lot for accurate information. Reuters' 2025 Digital News Report puts younger news habits inside the social/video shift while still finding that people check uncertain information against trusted news brands and official sources.

continue reading

Also Published

the next survey trust problem is agentic respondents

I do not want to turn synthetic data into a scandal word. I use generated respondents differently than I use real people. They help me do the rough work before launch: pressure-test logic, walk through weird edge cases, and catch the prompt that reads clearly to me but not to a respondent.

continue reading

Also Published

A Good Summit

This week I spent time in Boulder talking, learning, and thinking about AI, leadership, and community.

continue reading