What progress have you made since your last update?
Over the past six months, we’ve maintained and improved the Intelligence Rising platform. We prioritised usability improvements over some of the features originally outlined in our proposal. Our aim has been to reduce the time facilitators spend setting up and managing workshops, allowing participants to focus on exploring AI competition dynamics and the trade-offs.
The improvements include:
- A redesigned admin interface allowing facilitators to customise individual games, including editing technology and policy descriptions during workshops.
- Redesigned technology and policy cards that make key information, including safety status, easier to understand. Facilitators can also override research and policy activation requirements directly.
- A new interface allowing participants to allocate research resources publicly or secretly.
- Facilitators can now export data from all games at once, making it easier to analyse workshop outcomes.
We also improved the presentation and navigation of our three interactive AI competition models: https://www.modelingcooperation.com/interactive-models
What are your next steps?
- Update the technology tree in collaboration with TSR to better reflect current developments.
- Include an immersive 3D elements and a new end-of-game screen that reinforces the consequences of decisions involving AI progress and safety.
And, on an ongoing basis, we will continue refining the platform based on facilitator and participant feedback.
Is there anything others could help you with?
We generally welcome introductions to AI safety researchers, educators, facilitators and organisations interested in using our platform to explore AI governance risks.