The Swift Centre is launching an open competition to help bridge the gap between AI forecasting and government policy. We are looking for participants who want to take rigorous forecasts produced by the Swift Centre's professional team and translate them into policy advice that set out options and recommendations for specific decision-makers.

Why take part?

As someone who has facilitated numerous AI policy workshops, I know that the hardest part of governance isn't just predicting a threat, but deciding what a political leader can actually do about it. This competition aims to:

• Provide chance for those new to AI governance to experience writing professional policy advice.
• Demonstrate how probabilistic thinking should be the foundation for designing tangible, and well targeted, policy options.
• Improve the explicitness of reasoning in policy writing, moving away from vibes towards transparent advice that confidently outlines what is known and what isn't, and how a decision maker should act.

My hope is that this provides a fun and unique way for people to think about important AI threats and events to develop policy advice.

Register if you'd like to take part.

How it'll work:

We aren't asking for essays. We are asking you to act as a Policy Advisor to a Secretary of State.

If you have signed up to take part (using this form), you will be sent a report with 5 forecasts our team have made, along with detailed rationales. You will then:

1. Analyse the Forecast: Review and choose one of our 5 forecasts. What are the causes? What are the uncertainities? What are the risks? Where do you disagree, and why?
2. Develop Solutions: Outline up to 3 policy options for the relevant Secretary of State. Do nothing, invest in diplomatic action, introduce or remove regulation? You should develop your options based on the forecasts and rationales given (explaining where you disagree and why if so).
3. The Finished Advice: We will share a 3 page template for your advice to be provided in. This will align to a typical format policy advice is provided to Secretary of States in the UK (based on my own experience). It will cover:
   • Context/Background of the forecast you have chosen and what the Secretary of State needs to be aware to understand the issue.
   • Your three policy options including the risks, costs, and how they'd be implemented (e.g. new law, meeting at a international event etc.)
   • Recommendation of which option the Secretary of State should take forward and why.

Timeline:

• Now: Register if you'd like to take part.
• Start of March: Full rationales and templates will be sent to registered participants.
• April 6th: Submission deadline.
• April 20th: Dashboard with all submission goes live, with "competition winners" announced.

The Questions

The forecasted questions will be fairly high level. This will enable participants to have a fair amount of freedom to interpret the impact of the event how they desire and thus develop unique policy options.

To maximise the decision relevance of the questions, they are all focused on potential events in the next 18-24 months. The topics are:

• Agentic Capabilities: Autonomy thresholds for frontier AI.
• Workforce Impact: Structural shifts in UK Finance and IT sectors.
• Market Stability: "AI Bubble" risks.
• Autonomous Weapons: G20 decisions on 'human-out-of-the-loop' systems.
• Biosecurity: Addressing LLM-facilitated biological threats.

The "Competition" Element

We believe this offers a unique outlet to the many people currently seeking to try, improve, or build experience in writing policy. 

However, to add a little extra incentive and impact, we will:

1. Publish all submissions sent to us on a dashboard that will be linked on our website (swiftcentre.org)
2. Circulate the top submissions for each of the questions to relevant connections in the AI policy and governance space, including those working in think tanks, funding bodies, and Government departments. 

Register using this form