Why this matters
Most people are worried about AI, but worry alone has not produced widespread public pressure for change. In the absence of that pressure, tech industry interests are shaping outcomes largely unchallenged. This study identifies which harms can close the gap between caring and doing - giving campaigners, funders, and movement strategists an evidence base for choosing what to lead with and how to frame it.
What we found
Participants were randomly assigned to read a short news-style vignette about one of eleven AI harms - or no vignette - and then measured on concern and willingness to act. AI-enabled warfare (autonomous weapons making life-and-death decisions without human oversight) shifted attitudes across every measure. Environmental harms and AI bias and discrimination showed the strongest effects on willingness to act, suggesting real campaign potential for both. Job displacement - the issue people mention most spontaneously - ranked poorly as a mobiliser, showing that salience is not a reliable proxy for action. Extinction risk ranked last for direct concern, but there was a notable indirect route: reading about AI-enabled warfare raised concern about existential risk more than reading about existential risk itself, suggesting that concrete present-day harms may be more effective entry points to communicating about catastrophic ones. Over a quarter of participants clicked through to sign a real AI regulation petition during the study, indicating significant latent concern.
What it means for campaigners and funders
The findings point to three practical choices for those seeking to build public pressure on AI. First, lead with warfare and environmental harms - both are strong mobilisers and, unlike extinction risk, are concrete and proximate enough to generate anger rather than fatalism. Second, use anger strategically: messaging that generates justified outrage outperforms messaging that generates fear or anxiety. Third, treat existential and near-term risk advocates as potential allies rather than rivals - concrete harms appear to be a more effective route into public concern about catastrophic outcomes than abstract long-term framings.
Read the full report
The findings above are a summary. The full report, including methodology and supporting evidence, is available on socialchangelab.org.
Read the report