Making Policy Costs Personal With AI
People support bad policies—until they learn what they'll personally pay
Most people want the government to fight climate change. But when asked how much they'd personally pay, the vast majority cap it at £25 per month.
Meanwhile, the UK's net-zero policies already cost households far more than that, even before counting massive opportunity costs e.g., job losses from having some of the world's highest energy prices.
This is just one striking example of a policy voters would never support if they knew the full costs.
AI can make the costs of policies tangible and personal
I'm building an AI-powered dashboard that bridges this gap between policy support and policy costs, starting with UK energy policy. Energy and climate are a natural starting point because of my work on AlexAI, an energy and climate expert AI for Alex Epstein.
Check out the work-in-progress: policy-impact.uk
Using credible explanations grounded in facts, AI can help voters understand whether the policies they support align with their values — or rethink their values if they find that the cost-benefit doesn't add up!
Here is how it works:
At the top of the website, there are a few hard-hitting stats and tangible impacts, personalized according to their UK region and user profile.
Further down, a news feed features energy-focused stories, each with suggested actions citizens can take – like emailing their MP, posting on social media, or filing a regulatory complaint.
AI is really good at drafting high-quality emails, but here’s the key feature — the AI first generates the draft after the user selects which personalized talking points to emphasize.
The project is still in an experimental stage and many details may change in the coming weeks/months!
Getting the methodology right is key
The idea for each metric and news story is to have a straightforward, personalized explanation showing which policy is responsible for it and to what extent.
I am under no illusion that this is an easy task!
But I believe that this can be done by using a combination of official government numbers (like average energy bill), comparing UK policy outcomes to uncontroversial baselines like France (e.g. on nuclear) or even UK itself a few decades ago.
Here I am inspired by Our World in Data and UK Foundations essay in particular. I think there is potential to systematize a similar approach and guide AI to connect the dots on a much larger scale.
Next steps
I am working to validate the overall concept and methodology. I want to see who the project resonates with.
I'm looking for feedback, advisors, and collaborators. If this sounds like something you'd want to be part of, or if you can connect me with others, do get in touch.
Long-term, I plan to expand it to other policy domains (housing, for instance) and other countries, including the US.