Human-machine superintelligence: how to live and prosper with AI

Marvin Minsky, one of the fathers of AI, published a book in 1986 with the title "Society of Mind" where he described a theory of intelligence. Minky's big idea was that the human mind was the result of many mind-less agents coming together, interacting, and producing complex mental phenomena such as language, memory, free will, and consciousness. In essence, Minsky's "society of mind" was a way of understanding swarm intelligence as a fundamental principle of how nervous systems work. Similarly to how termites swarm to build architecturally complex nests, neurons interconnect and exchange signals that translate to our feeling of selfhood, and more. Therefore, Minsky argued, if you want to build intelligent machines you only need to build many simple computational agents, connect them in an appropriate way, and let intelligence emerge.

Another interpretation, or application, of swarm intelligence takes place in crowdsourcing. In this case computers and computer networks are used as facilitators that connect humans together. The "agents" in this case are not mind-less in themselves, nevertheless they are mind-less when it comes to the computational processes that takes place at the level of the crowdsourced human network as a whole. Take for example the pioneering Galaxy Zoo project that managed to harness the collective power of 150,000 human volunteers and classify 50 million galaxies in the first year of operation. No single human, or any number of human teams, could have achieved this herculean task without the help of computers. Crowdsourcing, citizen science, collective intelligence, are nowadays used to capitalize on the power of human networks to solve complex problems. In every case, computers are used to facilitate human interactions, as well as to processing and storing data. However, as data increase exponentially, both in volume and complexity, we are nearing the ultimate limits of what our fastest supercomputers can do. Wouldn't it be nice if the human agents in a crowdsourced network could augment their individual processing power - aka their intelligence? Wouldn't it be also nice if the networking computers could learn from human interactions and optimized their connectivity accordingly, much like neural axons modulate the strength of their connections inside the human brain?

So what if we added AI to the human crowdsourcing mix? What if we expanded on Minsky's big idea and included humans in the "society of mind", connecting and interacting with machine agents? What if we created a human-machine superintelligence?

We may indeed have to do all of the above. Indeed the most likely scenario could be that human-machine superintelligence will emerge soon, not by design but due to evolutionary necessity and pressure. The survival of our species is increasingly dependent on making correct collective decisions. The human institutions that currently exist to facilitate collective human decisions are demonstrably inadequate. Faced with complex, interconnected financial, political, humanitarian, environmental and health crises, we seem helpless to manage our collective affairs. One crisis leads to the next creating a world with an explosive mix of uncertainty and instability. If only we could communicate with each other better, understand each other's minds and points of view, find the win-win of collaboration quickly, agree and act in unison. But to do so it seems that we are in need of help: in dealing with the deluge of data, in accessing and processing knowledge fast, in predicting emerging patterns, in seeing multiple points of view concurrently. Embedding Artificial Intelligence in human institutions could therefore be the next logical step in dealing with humanity's biggest problems.

Our future survival may depend on the harmonious symbiosis of humans and intelligent machines, or perhaps the ultimate emergence of an evolved type of human whose mind will be continuously connected to a much larger, cybernetic mind of planetary dimensions.