Start with these videos
Stuart Russell: 3 principles for creating safer AI
Explore why uncertainty about human preferences matters. Discussion: when should an assistant ask instead of act?
Robert Miles: 9 Examples of Specification Gaming
See why optimizing a score can miss the real objective. Discussion: which metric in your organization could be gamed?
Kate Raworth: A healthy economy should be designed to thrive, not grow
Explore a different economic objective. Discussion: how should AI progress be measured beyond output?
These are foundational selections, not a newest-videos list. Links open the original publishers; no third-party player loads until you leave this site.
Technical alignment and accountable deployment
- Center for Human-Compatible AI — research on beneficial AI; start with Russell’s framing.
- NIST AI Risk Management Framework — voluntary guidance for managing AI risk across its lifecycle.
- Social Choice Should Guide AI Alignment in Dealing with Diverse Human Feedback — research on aggregating conflicting preferences.
- Collective Constitutional AI — a documented experiment connecting public input to model principles.
Shared voice and ownership
- Polis — deliberation software used in the public-input experiment above.
- Decidim — open-source infrastructure for participatory processes.
- Platform Cooperativism Consortium — democratic ownership and governance of digital platforms.
- Doughnut Economics Action Lab — meeting human needs within ecological limits.
ISITA, ISITAS, and the ISIT Game
ISITA.org and ISITAS Agora were supplied for this project. Their contents could not be verified during preparation; do not assume the domains are interchangeable or that they endorse this campaign.
The accessible ISIT Game proposal describes shared sensemaking, AI alignment, and participation-based income goals. Treat those goals as hypotheses to examine. Its philosophical framework is not an established scientific proof of alignment, and contingent participation payments are different from unconditional basic income.
Browse the ISITAS video channel. Ask how opinions are sampled, how dissent is preserved, and what revenue funds rewards.
Alternative systems worth testing
Consider worker and platform cooperatives, citizen assemblies, participatory budgeting, community data stewardship, public digital infrastructure, and shared productivity dividends. Each addresses different questions of ownership, representation, and distribution. None eliminates the need to test AI behavior.
Resource review date: October 8, 2026. Links are educational references, not partnerships or endorsements. Report broken links through Edaptus contact.