It’s my fourth visit to Davos and it’s fascinating how many more of the academics here are scientists rather than economists or management theorists. What is it that scientists can tell us about business? I had an opportunity to understand this when I led a debate with four professors from the University of California, Berkeley, who are experts in artificial intelligence (AI), neuroscience, and psychology.
The question we debated was “Will machines make better decisions than humans?” Of course this is a crucial question for anyone like me who is interested in the future of work. What impact will machines – be they computers or robots - have on the way we work? There are those that argue that up to 60% of the work we currently do will, within the next couple of decades, be replaced by machines. Of course this is already happening. The ‘hollowing out’ of work has already seen much medium skilled work disappear as executives’ thumbs and their smartphones replace secretaries, whilst assistants are being replaced by a combination of Wikipedia and automatic filing.
I started the debate by asking the audience what they believed: “Will machines make better decisions than humans?” About 50% agreed and 50% disagreed – it seems that even this auspicious group at Davos has no consensus. So what are the issues?
The first is well known to AI scientists. It is best described in Moravec’s paradox. This is the paradox that right now computers are really good at some things we humans find tricky - solving mathematical equations, playing chess, winning jeopardy, and even driving cars. These are tasks that require speed and precision. But here is the paradox. Machines are bad at other tasks we find easy - like clearing the coffee cups from a table or being creative. The psychologist Alison Gopnik studies babies and for her it’s clear: machines are not as smart as a two year old. Why? Well perhaps a two year old cannot play chess or drive a car, but they can do lots that a computer cannot do. They can create theories, make hypotheses, figure out how things work, imitate – and most importantly they learn and can be creative. Moravec’s argument by the way is that computers are great at reasoning, but the often unconscious, sensorimotor knowledge that has evolved in the human brain over billions of years is impossible to imitate.
The second issue, one we debated at length, was values and morality. We humans make decisions on the basis of our values. But do computers have values? The AI expert Stuart Russell argued that it is already possible to programme computers on the basis of utility in terms of gaining the highest value outcome. But whose values and how can the constant changing of consensus on values be programmed? And what of morality? As Bob Knight observed, human decisions are about emotions and we make decisions cognitively and viscerally. This comes from the evolution of the frontal cortex that took place over the last two million years.
Perhaps it’s just a matter of time? Computers cannot make decisions better than humans right now, but what about in 30 years time or a 100 years time? Or will it take another two million years for machines to become moral? Certainly with regard to sheer computing power, the technology is developing fast – recent developments in cloud computing are enabling the connection of millions of devices and that has enormous potential for amplification.
Or perhaps it’s not about time, but more about combinations? Ken Goldberg, an AI expert, believes it is in the combination of computers and people that real progress will be made. And, as I am learning in my own study of open innovation, it is the combination of diverse minds that bring forth the ideas that can be the basis of innovation.
What I have learnt from spending time with scientists is that they have much to tell us about how the world of work and organisations will develop. And even if there is no consensus on timeframes, the direction is clear: there is a great deal more to come in the human/machine interface.