How AI, Web3 and Humans Can Work Together to Solve Complex, Global Problems

(Yuichiro Chino/Getty Images)

Cultural and regulatory conversation worldwide today show that there is a lot of fear around AI. AI is going to take over the world, it’s coming for everyone’s job, it’s going to destroy Hollywood: choose your own dystopian adventure. I don’t want any of those things to happen, either, but I’m an AI optimist at heart. There are a lot of things that humans are pretty bad at, and we should want AI to help us with those things, and work together with us to fix them, especially when it comes to solving complex problems.

That is the exact thing that the Web3 space has been addressing since the first Bitcoin block: how to coordinate big, decentralized, complex groups of people around a common purpose. I would propose that there are some areas where AI and Web3 combined can truly help society tackle its most complicated problems. Here’s one potential view of what this could look like, with 100% feasibility based on the state of the tech today.

Some things AI is good at:

• Sourcing and consolidating a lot of information, even globally
• Evaluating outcomes on a set of parameters
• Carrying out research and tasks according to clear prompts from experts
• Some things blockchains are good at:
• Governance frameworks
• Global pool of funds deployed across a variety of jurisdictions
• Helping scale global networks of participants around a common protocol
• Some things humans are good at:
• Deep holistic expertise built from organic experience
• Making informed and nuanced decisions
• Sharing knowledge and enthusiasm on subjects they care about with communities

What happens when we mix all of these together in a symbiotic relationship to tackle a complex problem? I mean a REALLY complex, global-scale problem that has eluded humans for generations, or a “wicked problem,” like curing cancer. This requires coordinating the actions and wishes of thousands of people; making high stakes decisions about strategy and resource allocation; and a massive amount of management across many industries and jurisdictions.

How It Could All Work

So let’s imagine a Curing Cancer DAO. Founded by a consortium of scientific research labs, academic departments, and disease communities, totaling 1,000 people at the start. They decide on their shared mission and assign a small governance group of experts who will make strategic decisions. They initiate a DAO with membership NFTs for each participant with different governance responsibility levels and allocate a pool of funding to kick it off, and they instate an AI agent to manage the project as it scales.

The governance board assigns a list of KPIs to the AI agent, asking it to complete a set of management tasks for the community. Let’s say those start out as: manage donations to the treasury, keep track of new members, and disburse payments for work done on behalf of the DAO according to a clear set of deliverables. This will save a lot of time and overhead, which tend to consume a lot of resources for a not-for-profit organization or for a DAO.

More importantly, the board also tasks the AI agent with assessing the requirements to make progress on curing cancer and crafting a proposal with a roadmap of work, sub-projects, possible participants and institutions around the world who would be good to get involved, and specific tasks that will need to be done. The agent comes up with a long-term plan and proposes a set of steps to execute on it, then submits to the board of experts for review.

The board makes adjustments and prioritizes across the proposed roadmap from the AI to cover the first six months of work on Curing Cancer DAO. They empower the AI agent to recruit people to do those tasks, assign people work (no matter how big or small), and assess how well the work was done, paying them from the treasury.

The AI agent creates updates to the roadmap and reports progress frequently to all stakeholders in the DAO, managing a global view of this complex project in a way that makes it much easier for local contributions to be made effectively and in real-time context. Over time, it can expand the project by proposing sub-communities, managing experiments, and helping coordinate across the growing membership, even interacting with multiple boards managing different areas of expertise.

The DAO board can veto any proposal from the agent or rework it so things can get more efficient as they make progress. Over time, if the governance board observes that the AI agent is not doing well at something, they can commission new training data in that specific area in order to improve the model and fine-tune it to fit their needs. This can even be crowd sourced onchain from among an expert community, who can review the work and assess the AI agent’s improvement.

AI, Web3’s Role

This vision actually wouldn’t be possible without both AI and Web3. We need the Web3 part for governance, for the financial aspect, for the coordination tooling. All of the AI’s actions are onchain, membership of the group and donations are managed via blockchain tooling, and it can work directly with the DAO treasury to carry out transactions with total transparency. AI can streamline every part of the process of running the Curing Cancer DAO, as long as it is working alongside expert humans and with oversight and transparency assisted by blockchains. If it’s all on a blockchain, we can monitor risk and manage trust much better than we can even in today’s primarily social systems.

This is a very high level example, but I hope it prompts us to think in an optimistic direction about how we can become more effective at problem solving when we use AI and Web3 creatively. We will be able to scale up a lot of things that have been too complicated to manage just socially, or just with top-down command and control, or just with blockchains. This decentralized science community-building example could also apply to any global coordination problem or research effort.

None of these technologies are as interesting in isolation as they are when combined: It’s not about AI doing whole jobs themselves, but rather, about it doing what we are bad at, and helping us coordinate and do it faster. We will start to see some powerful experiments emerging in the next few years if we focus on productive building with proactive risk management, instating checks and balances that maximize the best of human and tech participants in service of a shared mission.

Edited by James Rubin.