When we seek to create collective intelligence, we cannot ignore the future promise of artificial intelligence, or AI, and the great strides the field has taken over recent years. Today AI can master several subsets of AI functions such as speech recognition, machine learning, neural networks, strategic gaming, data mining, pattern recognition, robotics, autonomous driving, coding etc.

The most recent breakthrough has come in the form of large language models that have astounded the public with their ability to create cogent texts and photorealistic images. As these models are capable of processing a great share of all the information found online, they exhibit a specific kind of collective intelligence that consolidates the cumulative knowledge of humanity.

To work, however, the system we’re building doesn’t need any advanced AI functions. Like a social media service, our system primarily relies on the information and intelligence provided by its users. But, AI can make the service much more effective. It could also help us perfect the code and algorithms the platform will be based on. Parts of this book can in fact be used as prompts to generate that code.

While AI is still far off from performing higher cognitive functions, it’s clear that AI is already beating humans in many tasks. Anything that requires monotonous and repetitive tasks or involves massive data sets is something that the AI technology of today can perform much faster, better and cheaper than humans. The sheer data processing power of a single server farm can decimate all human competition in a number of narrow data-intensive tasks.

The ability to process large amounts of data in the blink of an eye helps to create the impression of a real intelligence.

The next frontier is the ability to create artificial general intelligence, or AGI, which would hopefully enable AI to master more complex functions of human cognition like problem solving, reasoning, planning and the ability to interact with objects in their environment. The predicted end state of this development is the technological singularity, where AI will rapidly evolve into a form of superintelligence that surpasses humanity’s ability to control it. In this dystopian scenario, this would also bring an end to humanity, as the AI could see humanity as a threat to its existence.

Human consciousness is one of the enduring mysteries of science. New theories are presented regularly, yet none of them have fully satisfied the scientific community. What does it mean to be conscious? Why are we conscious? Are all living organisms conscious? Or is matter the true source of consciousness as panpsychism claims? Can consciousness be generated electronically? These questions have become ever more relevant as humans attempt to build AI that would rival and exceed the capabilities of the human brain.

Artificial intelligence is often made up of an array of algorithms that process the information they receive. An algorithm is like a recipe we follow when we cook food. It tells us all the ingredients we need, what to do with them and in what order. Data are entered into an algorithm and it cooks up the result. Since AI is usually built from many overlapping algorithms, it can perhaps, if we extend the metaphor, be compared to a whole cook book, which can be used to transform whatever ingredients are found in cupboards and fridges into a healthy and tasty meal.

While algorithms have inputs and outputs, machine learning, a subset of AI, works backwards just like mechanism design. If you show the software the starting point and the end point, it will try to create an algorithm, or the specific recipe by which one is turned into the other. The AlphaGo program, which was the first machine to win over a champion player in a game of go, was able to perfect its game playing algorithms by training with other players and finally by playing against itself.

Now this becomes interesting. This implies that we could, in theory, plug in the outcome we want–to maximize the well-being of everybody within planetary boundaries–and the relevant data sets of our current existence, and receive suggestions on how the system itself should be structured to produce the outcome we want.

While this task might still be overwhelming for today’s machine learning, it could be a genuine help in designing some specific aspects of the system. Machine learning could at least help the designers of the platform relinquish their preconceived notions of how to structure the system they are designing. As we create new governance models, AI could be employed to study their viability using computer simulations.

A citizen-centric system lends itself perfectly to agent-based modeling, since every citizen can be modeled as an individual agent in this computer simulation. Agent-based modeling can be especially helpful in simulating the cumulative impact of the actions and interactions that individual citizens can have within society under different sets of rules. Agent-based simulations are especially adept at modeling emergent phenomena, where the individual actions of individuals add up to something much bigger than the sum of their parts.

In our simulations the agents would represent average citizens living under the new financial and political system outlined in this book. Their needs, on average, would be knowable, as would the overall resources at their disposal. We can then propose various kinds of mechanisms and systems and run numerous simulations based on them, making tweaks and changes along the way. By adding data and refining the model itself, we can produce better and better simulations. Positive results can eventually lead to small-scale trials in real life.

Later, when the platform is operational, AI’s ability to suggest matches from a massive batch of data in a marketplace where supply and demand meet should greatly enhance the effectiveness of that marketplace. It could steer people to make better choices and reach better outcomes for both the community and the individual.

At this point, however, no general intelligence is even desired by the system we are creating. Since the computer code of the platform becomes their user’s de facto constitution, the transparency of every line of code becomes paramount. Complex algorithms can obscure the inner workings of the platform, which would quickly render the system incomprehensible to the average user. What we don’t want are black boxes that are vulnerable to manipulations or capture. AI, when employed, should therefore be utilized in narrow silos and their interactions should remain clear, transparent and predictable.

Needless to say, it is important that AI remains our servant and never becomes our master, as many technologists warn could happen. Since the platform we create will combine our society’s most vital functions, control of the system should never fall into any single set of hands, let alone be under the control of AI. Having citizens’ control taken away at the root, at the level of the code itself, would be tantamount to a coup and render the platform useless.

We must therefore make sure the platform is built correctly, its functions remain transparent and that it will always be controlled by its citizens. If and when AI is employed, it should structurally be guaranteed that no one party gains undue advantage from it or manages to monopolize its power. Artificial intelligence should serve the population like a public utility everybody can access whenever they need it.