Deep Blues: Computerization and the Future of the Workplace
Issue #25, March 1996
"I faced an opponent that has no emotions whatsoever. It's an unbeatable weapon because it attacks. It finds the short cut to any weakness in your position and doesn't hesitate, doesn't have any doubts, it's not scared by your illusory threats."
— Garry Kasparov
"The computer design that IBM is demonstrating here can grow in ability by orders of magnitude. Sooner or later, leaps like that will leave even a Kasparov in the rear-view mirror. When that day comes, there will be a flurry of articles about "computers taking over," but they will miss the point...We humans invented machines to do things we can't do ourselves. They're tools that make us stronger, move us faster, and even let us fly. As our bodies are the servant of our will, the tools we create are the servants of both. They enlarge us. No one who will tune in the Olympics this July cares that a car could easily outrun any competitor."
Last month, people all over the world watched with interest as Garry Kasparov, the world's highest-ranked chess player, took on IBM's Deep Blue, the most powerful computer ever designed to play chess. Although Kasparov ended up winning the match handily, many were surprised how hard Deep Blue made him work. Judging from media reports, the computer's power aroused a great deal of anxiety. If a computer can give a world champion trouble, what could it do to the rest of us?
We're going to find out. Today's top-of-the-line computer technology will be the year 2020's household appliance. After his victory, Kasparov explained that he had learned a great deal from his match with Deep Blue. We can also learn from it, for Deep Blue's strengths and weaknesses tell us a great deal about the role computers will play in our future. IBM does not engage in research for research's sake. IBM wants to make money. And IBM makes its money by selling computers to businesses. In other words, the work that went into Deep Blue will eventually bear fruit in the corporate world. It will change the way people work. But it will also help determine who finds work and who does not.
Today's workplace is undergoing dramatic changes. As computers become capable of managing more and more complex tasks, vast numbers of workers are finding themselves declared expendable. Fifteen years ago, it was primarily blue-collar workers that worried about becoming obsolete. Now the computerization of the workplace is making clerical workers every bit as vulnerable to corporate downsizing. White-collar managers are also not immune. Workers in all sectors of the economy are wondering about their futures.
What is the secret to staying employed? Almost all predictions about the near future of the workplace suggest that staying employed will no longer mean staying employed at the same job. If technological innovation proceeds at its current pace, entire occupations will disappear every decade. But which ones? In order to figure out the future of the workplace, we need to know which occupations will be rendered obsolete by the use of computers. A closer look at the present-day workplace is in order.
In organizations of any size, jobs are arranged in a well-defined hierarchy. The amount of decision-making a position is imagined to require determines its place in that hierarchy. Low-level positions are assigned a minimum of responsibility. Workers in these positions perform highly repetitive tasks like working on an assembly line, filing, answering the phone, and sorting mail. While workers in these positions will often need to make minor decisions during their work, they are not expected to think about anything that is not directly related to their assigned tasks. If something goes wrong during the execution of a task, they are expected to contact their supervisor. In other words, they are not called upon to make any decisions that demand an overall knowledge of their organization. Complex decisions that take many variables into account are the responsibility of employees higher up on the organizational ladder. Sometimes these are decisions about exceptions to the rules governing lower-level employees. Sometimes they are decisions about what those rules should be. In any case, much of higher-level workers' responsibilities consist of managing the workers below them in the organizational hierarchy.
The impact of computerization is first felt in the lowest level positions. The repetitive tasks performed in these positions rapidly become routines. And even inexpensive computers are adept at executing routines efficiently. In the last ten years, many middle-level clerical workers have found themselves asked to supervise computers instead of people. Of course, supervising a computer and supervising a human being are dissimilar tasks. Middle-level workers have had to learn how to work with computer technology. Those who have not been able to do this have found themselves replaced by more 'computer-literate' workers.
In the present-day workplace, computers still tend to be reserved for routine tasks. Decision-making that involves more variables, such as troubleshooting or rule-creation, is still entrusted to humans. This is why lower-level positions were the first victims of corporate downsizing. But there are more and more signs that this is changing. As affordable computers become more and more powerful, it becomes easier to delegate multi-variable decisions to non-human 'workers'. In fact, when all the variables that go into a decision are easily converted into numbers, when they are 'quantifiable', a well-programmed computer will make decisions more rapidly than a person would. Also, even allowing for errors, the computer's decisions will be remarkably consistent, because it follows the rules programmed into it.
Chess is a perfect example. Although it is an extremely complex game, it is still played on a grid easily converted into numerical coordinates. In other words, the basic variables a chess player must take into account are quantifiable. During his match against Deep Blue, Kasparov commented that the computer's ability to 'crunch numbers' gives it a powerful advantage against human beings. It doesn't get tired. And it can work out the possible permutations of a particular move far more rapidly than even a world champion.
It is only when variables are not easily converted into numbers, when they aren't 'quantifiable', that computers are at a disadvantage. But when we turn to the contemporary business world, it is startling to see how many decisions are based solely on the variables computers handle well. There are two reasons for this. The first is simple. Capitalism encourages people to put profit first. Money is our world's least common denominator. All the rhetoric about traditional values aside, monetary value is valued most. The second reason is related to the first, but more complicated. Today's business world is dominated by short-term thinking. Few business decisions look beyond the immediate future. The complexities long-term thinking introduces into decision-making are avoided.
For a computer, the differences between short-term thinking and long-term thinking are great. If a computer is fed enough numerical data about the present and recent past, and if it is well-programmed, it can make reasonably accurate projections about the immediate future. However, when projections about the more-distant future are required, basing decisions on numbers alone becomes more and more difficult. Twentieth-century physics and mathematics have convincingly demonstrated the difficulty of predicting the distant future. No matter how many variables from the present and past that we take into account, it seems we will never fully master the randomness of our universe.
Long-term thinking demands that we consider variables that cannot be reduced to numbers, that mark the points where ambiguity and uncertainty intrude on our best-laid plans. In recent years, many of the most interesting experiments with computers have revolved around the problem of these non-quantifiable variables. Fuzzy logic, fractals and other innovations in mathematical thinking hold out the promise of a new kind of number crunching that will make it easier for computers to mimic human beings' facility for long-term thinking. Planning for uncertainty means making flexible plans, something computers are still not able to do particularly well.
Kasparov's encounter with Deep Blue did a great job of bringing computers' strengths and weaknesses into sharp relief. His own commentary on the match was particularly prescient. While he was struggling early in the match, he noted that Deep Blue's computational power is so great that "quantity becomes quality." In other words, this "monsterous machine", as he called it, can take so many quantifiable variables into account that it appears to compensate for its problems with variables that can't be reduced to numbers.
Deep Blue's seemingly magical power to turn quantity into quality inspired Kasparov to state at one point that "for the first time we are playing not only with a computer but with something that has its own intelligence." He later added that Deep Blue "creates a new, new challenge" because for the first time "we have to deal with something that understands chess. Not because it understands but because it sees too deep." This somewhat paradoxical formulation-Deep Blue 'understands' without understanding-perfectly captures the computer's power. It cannot think or understand like human beings do, but it is so much better at processing numbers that it seems to possess 'intelligence' and 'understanding', an illusion that is purely a function of the "depth of its calculation."
The computer's ability to see the consequences of possible moves several turns into the future made the match very different from a match between two human beings. Kasparov noted that "if you look at the human games, they're lost due to short-term tactical combinations" in which "one side is getting an advantage" and the other side is "crushed under the pressure." But he soon realized that against Deep Blue "there's no sure tactic" because "your opponent will never miss any short-term tactical combination." The computer's tactical mastery forced Kasparov to radically revise his mode of play. Unable to rely on his opponent's human fallibility, he needed to mimic Deep Blue's precision. As he put it, the uselessness of tactics against Deep Blue means that "part of my own style or anybody's style is just-you know, it's cut off."
For all of Deep Blue's tactical superiority, however, it proved a much weaker strategist that Kasparov. This cuts to the core of the distinction between short-term and long-term thinking. Tactics designed to force an opponent into a mistake and exploit it fail against the computer, because it calculates all short-term outcomes of a particular move. The further into the future it must look, however, the harder it is to predict its human opponent's next moves. And that human opponent's future actions involve variables that cannot be quantified.
In trying to describe his success against the computer, Kasparov repeatedly contrasted his intuition against its calculation. After winning the match, he stated "I must say that I am very reassured, when I think that the computer can analyze 50 billion positions every 3 minutes and the human being can still win, well, that's great for the human intuition and the greatest thing in the world, the human mind." As Kasparov's comments suggest, "intuition" is one of the words we use to describe the amorphous qualities that make us into something more than machines. It is true that some still dream of a science that will be able to translate these qualities into exact quantities. But it is also true that we are not even remotely close to achieving this. Many of us doubt that we ever will.
To return to our discussion of organizational hierarchies, we can see that computers are now making it possible to replace not only those low-level workers who perform simple filing and sorting tasks, but also those who make more complex decisions based on numbers. The only kind of decision-making computers are not yet suited to perform is decision-making that involves variables that aren't quantifiable. In the business world, these are typically thought of as the 'intangibles' that can make the numbers lie. These non-quantifiable variables include all of the so-called 'intangibles' that human beings are able to take into account when they make decisions. Intuition is based on intangibles. And long-term thinking requires the flexibility of intuition.
When an organization discourages all but the highest-level employees from using intangibles to make decisions, it treats the rest of them like computers. The path to computerization has already been paved by corporate policy. In this sort of organization, replacing employees with computers appears perfectly rational. In fact, so long as an organization bases most of its decision on numbers alone, the majority of its employees become expendable.
There is a memorable television commercial for Saturn, an American car company. In the ad an African-American who works the assembly line explains that the innovative management techniques used at the unionized Saturn plant give him the responsibility to stop the entire assembly line when he spots a problem. In theory, the organizational model touted in this commercial really does give middle and lower-level workers more freedom to make decisions that affect their workplace as a whole.
From a business standpoint, the problem with this organizational model is simple: workers with a little freedom are likely to begin wondering why they don't have more. In places like Japan, this model has usually worked because the social pressure to conform curbs free-thinking. In less homogenous places, it could be downright dangerous for business. Our capitalist system maintains itself by preventing most workers from thinking about anything beyond the narrow confines of their particular occupation. When workers are suddenly given the responsibility of thinking about their organization as a whole, the logical next step is for them to start thinking about their organization's place in the global economy. They may begin asking why they are allowed to make decisions that affect production, but not about distribution of the profits that production yields.
The rush to computerize the workplace is leading to a paradoxical situation. In the future, human beings will be called upon to complement the calculating power of computers with qualities computers lack, such as intuition and a facility for coping with uncertainty. But at present, the right to take full advantage of these qualities is denied to the majority of employees. Many of the people who do take advantage of these qualities do so in the name of profit. If the workplace of the future is to provide opportunities for the ranks of the unemployed, this will have to change. If Deep Blue teaches us anything, it's that a society in which human beings will no longer be able to work like machines will have a hard time creating jobs. That is, unless it has very different priorities from the society we live in today.
When he wrote this, Charlie Bertsch was a Ph.D. student at UC-Berkeley. He now teaches at the University of Arizona. He welcomes your feedback: email@example.com