The dominance of humanity may soon be challenged. That’s because an alien intelligence is heading towards us at breakneck speeds. Many experts believe it will arrive within the next 50 years while others think it will get here in 20. Either way, it will not show up in a rocket ship but will be born right here on Earth, in a research lab.
I’m talking about the first sentient AI
It will be as different from us as any alien we can imagine and we have no reason believe its interests will be even remotely aligned with our own. This intelligence will be flexible and cunning, able to infiltrate our computer networks and permeate our critical infrastructure. Upon first contact, it might appear harmless, childlike even. But it will quickly become smarter than we are and when that happens, humans will struggle to understand how it thinks, feels or acts. But it will understand us completely. After all, we will have told it everything we know and it will have spent decades studying human actions and reactions. And if that sounds incredibly dangerous, that’s because it is.
Can we harness the vast potential of AI while at the same time preventing any unforeseen consequences from being unleashed on the world? Maybe we can and maybe we can’t, although humanity has a poor track record for keeping dangerous technologies in check. Do we even need to worry, since if we invent such an alien intelligence, surely it will exist to make our lives better? While that’s possible, it’s equally likely that it will pursue its own interests, just as humans vigorously dominated Earth’s other species and exploited its resources. So if we can’t stop this alien intelligence from arriving and we can’t count on it to be friendly, how can we protect ourselves from being intellectually outmatched by our own creation? A good place to look is Mother Nature, where many species have evolved ingenious methods of jointly amplifying their intellects to levels well beyond the capacity of any single individual. Evolution has enabled these species to ‘think as one’, combining the knowledge, wisdom, intuition and instincts of large groups into closedloop systems that are smarter together than any of the individuals could ever be alone. Biologists call this process ‘swarm intelligence’. It’s a primary reason why birds flock, fish school and bees swarm.
Every spring, for example, when a bee colony need to find a new home, it sends out hundreds of scouts to search a thirty square mile area and find dozens of candidate sites. But how do they select the right spot? Individually, honey bees aren’t that clever – but by forming a hive mind – a ‘brain of brains’ – the colony combines its collective insights to consider all the options. They share their knowledge by vibrating their bodies in what biologists call a ‘waggle dance’. These competing signals are compared and contrasted until the swarm converges on a single solution that is not just decisive but also ends up being the optimal outcome the vast majority of the time.
Now if you were a human CEO looking to locate a new factory, you’d face a similarly complex problem and be hard pressed to pick the optimal solution. Yet researchers at Cornell University have shown that honey bees are able to find the best possible location about 80% of the time. That is a massive amplification of intellectual ability.
A human hive mind?
So what about us? If birds, bees, and fish can form a real time ‘brain of brains’, could humans do the same, thereby amplifying our intelligence beyond the limits of any one individual? Our species may not have evolved to do a waggle dance but over the last half-century, we have built a global infrastructure that connects every man, woman and child in the industrialised world.
The interfaces and algorithms to connect people into real time swarms also exist already and are referred to as ‘artificial swarm intelligence’. Governed by AI algorithms, these combine the knowledge, wisdom, insights and intuitions of real people and in real time, enabling large groups of networked humans to quickly converge upon optimised decisions, predictions, solutions and evaluations.
Perhaps unsurprisingly, it turns out that these ‘human swarms’ can be really, really smart. Recent studies have shown that by forming real time swarms, significant improvements can be made in everything from predicting financial trends to assessing whether a person is lying. For example, in a published study by researchers at Unanimous AI and Oxford University, a human swarm was used to predict the outcome of all English Premier League games. The experiment was conducted using a group of average sports fans, who logged into UNU, a platform for artificial swarm intelligence developed by Unanimous AI. Results showed that individuals who averaged 55% accuracy when working alone – only slightly better than flipping a coin – were able to amplify their predictive accuracy to 72% by thinking together as swarms. This corresponds to a 131% amplification in sports predicting ability across five consecutive weeks and 50 games. And this result appears to be the norm, not the exception. Human swarms are smart.
When people are linked in real time using swarming algorithms, we can form closed-loop systems that amplify our intelligence to a level higher than our own individual abilities. This suggests that artificial swarm intelligence is a viable pathway to building superintelligent systems. After all, if a swarm of bees can make complex life-or-death decisions better than a human CEO, a swarm of already smart humans working together should be able to soar to unimaginable intellectual heights.
How does this help us mitigate the rise of purely artificial intelligences? Well, because these systems are a combination of AI algorithms and real human participants, they will always be inherently aligned with human morals, interests, values and sensibilities. The resulting super-intelligence won’t be an alien entity with conflicting values and interests but instead will be the natural evolution of the human intellect.
This is why I believe the best way for us humans to stay ahead is by amplifying our intelligence. Machines may soon rival our intelligence and may even threaten our existence but real time human swarms may save us by being our evolutionary destiny. After all, it’s the same path that was taken by the birds and the bees when they were faced with the need to boost their intellectual capacities.
Dr Louis Rosenberg is a Stanford graduate who founded Unanimous AI to amplify the intelligence of human groups. A prolific inventor, he has been awarded more than 350 patents for his technological efforts in AR, VR, collaborative systems, artificial intelligence and human-computer interaction.