Common Signals

Work with us

Common Signals builds a shared, neutral evidence base for anyone explaining AI to the public.

We're at an early stage, with published reception studies, an open methodology, a public data explorer and a map of trusted messengers.

If you recognise yourself in one of these sections, we'd like to hear from you.

Fund the work

So far our research has used data that already exists: public comments on videos, podcasts and posts about AI risk. That work has produced useful hypotheses, but it can only tell us about people who chose to comment. We are not yet funded, and a grant would pay for three things.

First, a standing protocol for coding public reaction to each major AI incident, so that the next one is studied within weeks against a published codebook, not whenever someone has time.

Second, split tests run with partner organisations: two versions of a real message sent to the partner's own audience and compared on what people do. This is how the hypotheses from our reception studies get tested.

Third, a blind second coding of our five earlier studies, so that each has an agreement figure, as the Diary of a CEO study already does.

Everything we produce is published openly under a Creative Commons licence. A grant to Common Signals becomes a resource for every organisation communicating about AI. We'd rather talk to you while these plans are still open than after they're fixed. Our funder brief is available on request.

Collaborate on research

Public attitudes to AI, message testing, content analysis of AI media: if your question overlaps with ours, we would rather work on it together than in parallel. Our coded datasets are in the data explorer under CC BY 4.0.

We also need methodology review. Our methodology page sets out how we sample comments, code them and turn patterns into hypotheses. We'd like a small group of reviewers with survey or content analysis experience to read our drafts before publication.

If you communicate about AI for a living, you can also help shape how findings are used: reviewing drafts of briefings and message guides, or agreeing shared criteria for judging a message, such as whether it changed understanding, built or eroded trust, or backfired. If you've had to judge your own campaigns and found the metrics wanting, we'd value your input.

Have a finding of your own? Write for Common Signals

Suggest a piece to analyse

If you make videos, podcasts or posts about AI, or have seen one that drew a large reaction, the comments under it are some of the best evidence available. Send us the link and we'll consider it for a reception study. We can't study every suggestion. If the piece is yours, we'll share the findings with you first, and we only publish analysis of your content with your agreement.

Get in touch

For anything else, email [email protected].