Artificial intelligence: the Internet of Things really is on its way
Artificial intelligence could bring us utopia or wipe us out, says Calum Chace. It depends how we harness it...
I believe that historians in 2100 will look back on our century as the age of the two singularities. The word “singularity” is used in maths and physics to mean a point where change has become so rapid that the normal rules no longer apply.
The two singularities coming our way this century are the economic singularity and the technological singularity. Both present enormous opportunities and challenges. If we manage them successfully our future as a species is beyond wonderful. If we fail, it could be miserable – and probably short.
The reason for this is artificial intelligence (AI) – humanity’s most powerful technology. Software that solves problems and turns data into insight has already made big impacts: your smartphone employs AI to deliver maps and apps; Google to answer your questions. But wonderful as these things are, the AI revolution has barely begun, and it is accelerating fast.
In the next few decades you’ll see self-driving cars on the streets and have conversations with Siri, which will transform it into an invaluable friend. With trillions of tiny sensors and computer chips embedded in vehicles, clothing, buildings, street lamps and roads, your environment will become intelligible: the Internet of Things really is on its way.
In 2015, AI was rarely out of the headlines, and with good reason. The technology is approaching a tipping point at which machines perform at superhuman level many tasks that were previously deemed uniquely human. They are on the cusp of recognising faces and other images better than we do, and understanding and processing natural speech as well as we do.
The range of possible consequences is wide, from terrible to wonderful, and they are not pre-determined. They will be selected partly by happenstance, partly by their own internal logic but partly also by the policies embraced at all levels of society. The argument of my book Surviving AI... is that we should monitor the changes that are happening, and adopt policies which will encourage best possible outcomes.
AI researcher Demis Hassabis likes to say that humanity’s plan for the future should involve two steps. Step one is to solve intelligence (i.e. create powerful AIs). Step two is to use that intelligence to solve everything else. How to cure disease, and even stop and reverse the ageing process. How to harness more solar energy and generate clean energy.
These huge problems and many more can be solved if we tackle them together with machines that can assimilate and process information better than we can. But like any powerful technology, AI has its risks. The two biggest are technological unemployment and superintelligence, and it is these which will generate the two singularities that kicked off this article.
Technological unemployment is what will happen if, two or three decades hence, the automation of jobs by machines renders large numbers of people unable to find paid work because there is no work they can do that cannot be done cheaper, faster and more reliably by machines. If we are smart we could create an economy of “radical abundance”, where AIs and robots do all the work and humans enjoy lives of leisure and play, spending our days in conversation with friends, learning, playing sport, creating art and travelling.
But to make this world a reality we will probably need to evolve an entirely new economy, which is why I call it an economic singularity. If we get it wrong, an elite may own the means of production and suppress the rest of us in a dystopian technological authoritarian regime. Or the process of getting from where we are now to the new economy we want could prove too challenging, with devastating consequences for our economies, our societies and perhaps our entire civilisation.
The arrival of superintelligence, which could happen from two (unlikely) to seven (very likely) or more decades hence, will represent a technological singularity, and the most significant event in human history bar none. Being the second-smartest species on the planet is an uncomfortable position, as chimpanzees could tell you if they understood how precarious their position is. Working out how to survive this transition is the most important challenge facing humanity in this and the next generation.
If the superintelligence values us, it could improve our lives in ways quite literally beyond our imagination. A superintelligence that recursively improved its own architecture and expanded its capabilities could very plausibly solve almost any human problem you can think of. Death could become optional and we could enjoy lives of constant bliss and excitement. If it is indifferent to us, or even hostile, the result for us could be extinction.
Surviving AI, and two singularities, is the great challenge of this century.
Calum Chace’s Surviving AI: The Promise and Peril of Artificial Intelligence is out now (Three Cs, £9.99)