Almost twenty years ago, Garry Kasparov was beaten at chess by a computer. People started to worry that computers would soon be taking over the world. But that doesn’t seem to have happened. Instead, a more interesting issue has emerged: the way in which humans can use computers to make themselves stronger and more capable.
We’re not talking in a James Bond villain-type way, but about ordinary people improving their own capability. For example, in 2005, a chess tournament was held in which people were allowed to use computers. You’d probably predict that a grandmaster with a laptop would beat a supercomputer by itself. But what was interesting was that the final winner wasn’t a grand master with a supercomputer, but two amateur chess players working with three relatively weak laptops. So what was the issue here? Shyam Sankar, in a TED talk on the rise of human-computer co-operation, suggests that it was the process. To put this in equation terms:
Weak human + weak machine + good process > strong human + strong machine + poor process
Sankar argues that instead of thinking about artificial intelligence, we should be thinking about intelligence augmentation, that is, the idea that machines can be used to augment the intelligence of humans. Instead of trying to turn a toaster into a rocket, we need to think about taking a human and making him or her more capable. As an example, computer scientists created a computer game about protein folding, where individuals mess about with possible structures of a protein, and the computer works out the forces acting on the proposed model to see if it is possible. Using this game, three individuals worked out the structure of one protein that had eluded computer analysis for over ten years. You can, and many people have, not least in the comments on Sankar’s TED talk, argue that this so-called ‘co-operation’ is in fact simply humans using tools, just as they have always done. But putting aside the term itself, there seems to be very little question that using computers makes humans more capable, and able to do things that were previously impossible, such as crunch large amounts of data and find insights.
So if we start from the point of view that human-computer cooperation or symbiosis is making humans more capable, then the next question is how to improve the process of co-operation. Sankar suggests that the issue is the interface between human and machine, and most importantly, minimising the friction. It is this that can make a weak human/computer combination stronger than a strong one.
The impact of computer-human cooperation
Sankar further suggests that this type of co-operation is crucial for solving human problems. He is thinking particularly of what he describes as problems with ‘adaptive adversaries’, such as other humans. Computers are incapable of detecting new patterns of behaviour, which means that they can’t detect new ways of defrauding the system. But by using computers, and particularly big data, humans can move more quickly, ‘think’ faster, and get ahead of adversaries such as terrorists or organised crime syndicates. Human-computer interaction is also useful in disaster relief efforts, to pull together disparate data from multiple sources quickly, decide what’s really important and enable swift coordinated action. An earthquake in Haiti, which left 10% homeless, also destroyed all information about flooding in the country, and meant that the location of disaster relief camps was something of a lottery. But by using computers, aid organisations were able to analyse data from the US Army, geospatial data from satellites, online data from an environmental risk conference and similar sources. The data was pulled together inside two days to see which temporary camps were most at risk from flooding during the forthcoming rainy season, and enable them to be moved.
Another example of human-computer cooperation is deciphering old texts. Have you ever wondered where at least some of those funny letters come from that you are asked to read and enter on some websites to ensure that you’re human? They’re from really old texts. And asking individuals to read and enter what they read helps computers to hone their algorithms for reading such texts.
Human-human or human-computer?
We’ve often discussed the importance of human-human interaction, and that remains crucial. But it seems that the use of machines or tools to augment our human abilities, something which we also see apes doing, is also a fundamental part of life. And by using computers, in particular, we are able to improve our capabilities to solve very human problems.
Image credit: Wooden Town Mecavnik, Serbia by Uros Petrovic