Project summary

A 12-month project to establish AI risk communication for public preparedness as a defined field (more at https://theophilusfemialawonde\.com/ai\-capabilities\-communication\-versus\-ai\-risk\-communication\)\. Analysis shows that frontier AI labs strategically communicate more about increasing AI capabilities than about AI risks, or even about AI risks in ways that prepare the public for what may (or may not) emerge from AI becoming AGI. Essentially, while frontier AI labs are best positioned to prepare the public effectively for AI risks, they are not necessarily incentivized to do so. This project will focus on publishing a foundational research agenda, conceptualizing AI risk communication for public preparedness and validating construct dimensions (what constitutes AI risk communication for public preparedness?), and examining methods for scaling (to reach wider audiences) AI risk communication for public preparedness and ensuring this communication is effective (actionable, improves self-efficacy). The project will also engage in field-building efforts by hosting a convention of AI risk communication scholars focused on public preparedness, with working papers as convention outcomes.

What are this project's goals? How will you achieve them?

1) Conceptualization, achieved by publishing a foundational research agenda defining the construct and validating its dimensions, building on prior published analysis of the capabilities-risk communication asymmetry.

2) Watchdog efforts, achieved by running a longitudinal study with periodical reports on the State of AI Capabilities-AI Risk Communication Asymmetry.

3) Develop effective methods for scaling AI risk communication for public preparedness through research

4) Field-building: Position AI risk communication for public preparedness as a critical risk communication category through a convention that regroups risk communication scholars, with working papers as the convention outcome

How will this funding be used?

Lead researcher compensation (full-time, 12 months): $60,000
Phase 1 construct validation methodologist (part-time, 3 months): $10,000
Phase 2 research assistant for data collection and coding (part-time): $6,500
In-person convention (Venue, catering): $20,000
AI tooling (3 ChatGPT Business premium seats, 12 months): $3,600
Website and infrastructure (domain, Cloudflare, hosting): $500
Research software and subscriptions (survey platform, transcription, reference management): $3,000
Contingency (7%): $7,252

Total: $110,852

Who is on your team? What's your track record on similar projects?

Theophilus Femi Alawonde, MA in Communication, Clemson University; MBA, Quantic School of Business and Technology. I have several publications on AI capabilities and AI risks, including a treatise on the asymmetry between AI capabilities communication and AI risk communication.

What are the most likely causes and outcomes if this project fails?

The most likely cause of failure, as I see it currently, are:
1) The Pause the Frontier Movement doesn't succeed, and AI companies ramp up their AI development efforts even faster than we've seen before. That could mean AI models with better capabilities are released almost every month, a pace our expected lean organization won't be able to match. It also means that the consistency in increasing AI capabilities might lead to AGI or broadly available and insanely capable, agentic, self-directing AI models before we have properly defined the AI risk communication for public preparedness construct or even researched ways to effectively communicate about AI risks to the larger public.

That would effectively render our mission moot, since AI risk communication would have become AI crisis communication, and scaled government interventions would be needed.

2) AI risk communication for public preparedness fails the construct validity process. In that case, it would be difficult to standardize what AI risk communication for public preparedness is. A fallback plan would be to still examine how the term could stay conceptually coherent, especially for distinguishing between the forms of (risk) communication about AI. That way, we would still be able to research ways to effectively scale communication (for public preparedness) about AI risks.

How much money have you raised in the last 12 months, and from where?

We have not raised any money yet.