How to introduce an AI government?
The North Star for AI-Based Governance: finding alignment between humans and artificial intelligence.
TL;DR - the essence of the message:
In a world driven by the AI revolution, finding alignment between humans and artificial intelligence is a pressing challenge. But here's a thought-provoking twist: Perhaps the key to unlocking this alignment lies in finding harmony among humanity first. Without a united front, how can we expect to bridge the gap with AI?
Enter the concept of a Well-Being Index (WBI), a potential solution that could guide us towards alignment as a global society.
Imagine a world where our collective well-being takes center stage, where metrics like home prices, unemployment rates, air quality, mental health, and green spaces shape our decision-making process. This holistic approach, calculable in real-time using digital tools, could revolutionize the way we evaluate and steer our progress. Say goodbye to the sole focus on GDP growth and embrace a transparent, people-centric approach through the power of the WBI.
But how do we get there? We need a roadmap that navigates us towards AI-based governance, ensuring reliability and alignment every step of the way.
The following text is exclusively written by Max Kusmierek without AI assistance—subscribe for free to receive more captivating Greetings from the Future.
Introduction
The AI revolution is in full swing, and every week hundreds of new tools are introduced. Unfortunately, many of us are still unaware of what is going on and how many jobs will be affected sooner than we think. I would like to stimulate the discussion of how to mindfully deploy those technologies in a beneficial way for the public sector.
One of the main needs I see is to establish a North Star to drive us in an aligned direction.
The Case for AI-Based Governance
AI systems maximize the efficiency of the friction cost. The number of factors and data used for the process is beyond human capacity.
Can’t corrupt open-source transparent systems.
Even in a direct democracy, society can’t cope with the number of decisions and might not be informed and educated enough to analyze all the nuances and make decisions
Of course, we need to develop a gradual roadmap for the rollout of AI-based governance. As we still need to improve the alignment and reliability of machine learning-based systems, it’s time to establish frameworks for their implementation.
The Roadmap for AI-Based Governance
The first stage "analysis, advice, and recommendation" (AAR), will be akin to a technological brainstorming session. During this stage, AI will suggest solutions and legislation, which will then be subject to a vote in parliament or a referendum. Of course, for this to work effectively, we need to nurture the system with data - a process that's already well underway.
The North Star for AI-Based Governance
The North Star: We need a clear direction for our AI systems to effect meaningful change and achieve alignment. To set this direction for everyone, it's crucial that we engage in constructive debate. Currently, our main KPI is gross domestic product (GDP), and we are constantly trying to achieve its growth while being limited by the finite resources of the planet.
I would suggest evaluating something in the form of a Well-Being Index as a North Star.
The WBI can include elements such as:
home prices per square meter/average wage,
unemployment rate,
air quality,
percentage of the population with depression (mental health issues),
and the amount of green space per inhabitant.
With modern digital tools, this can be calculated live on a municipality, district, country, or region scale and used to evaluate each decision. The transparent approach could also make more people actively participate in WBI contributions; as for the GDP, not so many people actually care about it.
Once we set up the global direction to thrive, we can move forward.
AI-Based Governance in Practice
Large Language Models (LLMs) can propose legislative changes to simplify the law - this is the low-hanging fruit. For instance, in the European Union, every country has a Value-Added Tax (VAT) implementation that differs not only in rates but also in approach and exemptions. I truly believe we are capable of unifying VAT tax regulations in the EU "overnight," which lowers the operation cost for many companies.
Adopting AI co-governance can make politicians more efficient, leading to better development for countries, as the current approach is sub-optimal.
The feasible holy grail could be an AI-based tax system. Hopefully, due to multiple sources of live data, the new methods of automatic taxation can work out. This can lead to completely new methods of financing the budget and maybe a new economy. There is a chance that the optimal tax for maintaining a good Well-Being index is based on actual natural resource use and environmental costs.
Almost sci-fi from yesterday's perspective can be an idea for avatar-like personal representatives, like an individual bot congressman/senator/member of parliament in the democratic process. This app will learn your preferences, patterns, and life goals and vote on your behalf in the direct democratic process, aligned with the North Star consensus.
Vox populi can be heard as never before.
(I was inspired by a similar idea in a TED talk by César Hidalgo)
Safety Considerations for AI-Based Governance
This can develop faster than we think, so now is the time to set the framework for AI. As a part of the safety implementation approach, discussion is necessary. Maybe we should set hard lines for what decisions automatic systems can't make. For example, preventing decisions about death or forbidding people to meet.
Also, regulations might need to cover the datasets that are nurturing the models. For public use, large language models should have transparent sourcing, maybe sealed with blockchain.
We need to keep in mind that those systems should always serve us, not the other way around.
By correctly implementing them, we can achieve the next level of development for our civilization. True evolution.
Conclusion and Next Steps
In my view, public LLMs should work on legal and tax system research while we discuss what our North Star is and start measuring it ASAP.
Subscribe and stay tuned for more Greetings from the Future by Max Kuśmierek (@MaxKusmierek).