This past Spring, AlphaGo — an artificial intelligence (AI) machine — beat the human world champion in a five-game match of Go, an ancient Chinese board game once considered so difficult that a computer could never master it. Since then an improved version of the game, called Master, has been slaying some of the game’s top players with a streak of 60 wins.
Last year, ride-sharing company Uber debuted a pilot programme involving self-driving cars. Elsewhere in the world, factories have become automated, churning out products using robotic workers designed with AI. With the buzz generated by these headlines, you might think AI machines are poised to take over just about any job a human can do. That idea might not sit well with you, for lots of reasons. But here’s an alternative: what if machine learning programmes were designed to help us rather than replace us?
AI generally refers to efforts to replace people with machines. But AI has a counterpart, known as intelligence augmentation, or IA, that instead aims to use similar machine learning technologies to assist — rather than replace — humans. IA may now be at a tipping point to take over from AI when it comes to progress and headlines.
Scientists have calculated that, as a global population, we’re close to generating 10 billion megabytes of new information every second. Despite all this information at our fingertips, innate human intelligence may be plateauing: while the average IQ rose 20 points in the past 80 years, owing to better health and education, it’s projected to only rise by about three points in the next 40. Our shrinking attention spans and cognitive inability to keep up with so much data add to the problem. In medicine alone, some 2,000 research papers are being published daily, beyond the ability of even the best doctor to keep up. Today, a small fraction of the data being generated by researchers in all fields is analysed for useful insights, even though we all agree that doing a better job of this could be better for our health, economy and the whole of society.
If traditional AI has shortcomings, they’re that AI has never been able to best humans when it comes to certain tasks involving nuances in language, complex problem-solving, and emotional and social intelligence. But where machine learning programmes shine is in sifting through data, finding connections and noting trends — exactly where humans, in our society of information overload, are struggling. Combining machine learning with the existing power of the human brain means we get the best of both worlds. Indeed, IBM has said the company is now focused on IA, hoping to boost human capabilities; and Carnegie Mellon’s School of Computer Science has noted that 98% of AI researchers are focused on work that fits the IA mould rather than strictly AI.
So what does the future hold for IA? Supercomputers and connected devices, like smartphones, will be increasingly used to sort through data and feed us information that’s pertinent — possibly without us having to search for the information or even know that we need it.
IBM’s Watson supercomputer is beginning to help doctors make better decisions about how to treat cancer by providing clinicians with the most relevant papers, data and clinical trials for each patient. As fields like materials science, brain science and genomics continue to mature, they’re increasingly intersecting with IA. Chips in our brains, electrical stimulation to turn on brain networks involved in creativity or genetic engineering could one day shape human intelligence. Entrepreneur Elon Musk recently revealed that he’s developing neural lace, a wireless brain-computer interface that can access internet information by thought without interfering with normal brain function.
It will take time for all these brain technologies to come of age — and for society to adapt to their use, but, in a few decades, there may be no such thing as strictly human IQ. Instead, augmented IQ will measure just how good we’ve become at melding the power of machines with the human brain.