Why a handful of nations shaping the global governance of AI is a recipe for disaster

Thanks to OpenAI's launch of ChatGPT last November and UK's hosting of an AI Safety Summit today and tomorrow, the entire world is now cognisant of the immense risks posed by AI for human safety and for an unaccountable concentration of power and wealth, and the need for unprecedented global action to tackle them. 

Many do's and don't can be learned from the ways we managed the risks nuclear weapons and the promise of nuclear energy, after WW2.

The choices of the five winner of WW2, and only veto-holders of the UN Security Council, prevented a nuclear catastrophe so far, but we came very close to it several times, and nuclear risk is higher today than it ever was.

They chose to manage the risks via a informal loose coordination of their intelligence agencies and various pressures on states, while they kept mostly for themselves the science of nuclear energy. It was only after all five of them had achieved solid nuclear weapons capability that the IAEA was established in 1957, to prevent others from doing the same.

Today, the risks with AI are even larger, its proliferation harder to control, and time to act shorter. 

Attempts by a handful of nations and firms to shape and control the global governance of AI, as it happened for nuclear, are a sure recipe for disaster, mainly because not enough nations and people will trust it enough to comply with the global bans and oversights that will be needed.

Enacting a process to build such governance that reconciles participation, inclusivity and effectiveness, is ripe with huge complexities and risks. 

For these reasons, last June 28th, our Trustless Computing Association launched the Harnessing AI Risk Initiative to call for a critical mass of globally-diverse nations to gather in meetings, in Geneva and online, to design and agree upon the Rules for the Election of an Open Transnational Constituent Assembly for AI and Digital Communications to create suitable intergovernmental organizations:
https://www.trustlesscomputing.org/harnessing-ai-risk-proposal

We are seeking partners fo all kinds, and donors for a $1.5m grant proposal to carry on this work:
https://www.trustlesscomputing.org/join-or-donate

Join us!