News Article 05/12/2007

Roger Schank: I like to study people instead of machines (Economic Outlook, November 2007, p. 14)
Exclusively to Economic Outlook

Machines are getting better and smarter and by smarter it means operating more like humans. For his new pioneer project, Professor Schank tells Outlook that what he does is based on a simple practice applied to humans by sharing your experience with your colleagues and with fellow men and learning by doing so.
By implementing this project not only in the secretive shipping field, but also in the other industries, Dr. Schank explains the important role it can be play to prevent accidents and save time and money.
Schank is one of the world's leading researchers in artificial intelligence (AI), learning theory, cognitive science, and the building of virtual learning environments. Schank is Professor Emeritus at Northwestern University. Previously he was the Chief Educational Officer of Carnegie Mellon West. In 2000, Schank won the Distinguished Contribution to Workplace Learning and Performance award of the ASTD.

Q: Let’s talk about your project. How did you come up with the idea to get involved in this process? What is the main idea behind it?
A: My field is artificial intelligence, I am iAnterested in getting machines to be smarter. But in order for machines to be smart, you have to understand what makes people smart, because the best AI strategy is to copy what people do. So I am actually more interested in people than I am in machines. Most software is not only not smart it is often quite annoying, and it rarely corresponds to how people think. The real issue is to get machines to be smarter, more helpful, and more useful. To do this it is important to see how people function and to ask what their needs are, rather than to focus on adding more features to an already complex piece of software.

Q: Now more specifically, how can we apply this process to shipping industry?
A: The shipping industry is not unlike many other regulated industries in that there are a lot of rules promulgated by legal bureaucracies who miss the point with respect to how people operate. Their biggest mistake is thinking that if they tell you the rules then you will know them and be able to follow them. People don’t learn by being told. We like to imagine that people can learn this way, since it would be so easy to just change people’s behavior if we could just tell them what to do, but it doesn’t work that way. Plato said this by the way. So this is an old idea. And the idea that you can learn new behavior from a list of regulations is just fallacious. So if you are a government regulatory agency, and you list the set of rules that everyone has to follow, you are more likely to cause trouble than you are to cause people to really follow new rules. If there is a problem, then people are not likely to want to talk about it openly if they have to report it and file official forms. But if we really don’t have a grasp of the kinds of problems that keep repeating themselves how things will get better? So, in a sense, the regulations make things worse instead of making them better. What I would like to see is a machine smart enough so that whenever there is any problem whatsoever, it has expert advice to offer you right then. Just in time. We can get that expert advice from people in the shipping industry. The machine itself does not have to know what it is saying. It just says, oh I see you have this problem because I know what you are trying to do, which ship it is, and what the conditions are. Listen to this. It’s going to be very helpful. It is like having your best friend be very wise and always watching over our shoulder. People have been around 200.000 years or so in their current form. We are actually not very different from our cave men ancestors. What did the caveman do when he was in a risky situation? He asked his friends. That’s what we have to do. So we need to get the computer to be like that, to have all the knowledge of shipping and be able to tell you the right thing at the right time. And that does not mean having 100.000 manuals because no one is going to read them. We need the collective wisdom of the shipping industry to be at everyone’s fingertips when needed. Technology can make this happen

Q: Has this idea of yours been successful in many other industries?
A: I have worked in many industries. The thing that’s interesting about shipping and my work with Dimitri Lyras is that he wants to push a little further than we’ve done before. I’ve been doing this stuff for 30 years. The U.S. army is one of my big clients. There are plenty of folks I’ve done this stuff with, but we are pushing the envelope. Shipping is a good industry to do this in, and the reason is that it’s bounded. You are dealing with a world where people have been doing the same thing for a long time. There’s a lot of information available and it is not an awfully complicated subject. People know a lot of things about shipping but don’t share their knowledge with each other. So even in a company, people don’t really know who the expert is to go to, or the expert may not be available when you need advice, even though the expert knows exactly what your needs are. As an example, imagine booking an airline ticket on line using Expedia or Travelocity. These programs assume that the reason you are using them is that you want to find the cheapest fare. Suppose I go there to see if the place I’m going is fun or suppose that I go there because I ‘m trying to find which airline has the most pleasant business class seats. These online travel sites make one assumption about who you are. Now, your friends know who you are, in a different way so they might tell you to take the trip because they think you would like the experience because its very pretty there. Every time I am use one of these programs it is as if it’s the first time I was ever on the site. The site doesn’t learn about me. But, all that can change. Software can get easier to use if it knows your world and it knows you. Shipping is going to be a place where we introduce these ideas.

Q: How can this help ship owners save money?
A: Harnessing expertise saves money. We rely upon expert advise. But, you can’t keep bothering the same person with the same question. If a thousand people are constantly asking the same question, the expert doesn’t have time for everyone But you can get the best experts in the world and ask them once and get the answers into the machine and then they can be delivered a thousand times without bothering the expert again. So you save money by having instant access to expertise without any complications. More efficient use of consultants on issues you might actually need advice for, say, an accident or something important, will help you save money. Getting good advice just in time always saves you money. If you are buying a house and you have someone very smart in the housing market to advise you on your purchase, that saves you money.

Q: How can you persuade people to write or talk about their experiences in order to organize a database, especially in an industry as secretive as the greek shipping industry?
A: It really isn’t that difficult to get experts to share their stories. People like to tell you what they know, even Greeks. If one person doesn’t want to share his advice on how to deal with an engine problem then another will. In the end we will have a machine that actually knows a lot about shipping. That’s what we’re doing. If you don’t want to add your piece to it, you can just use it. I think over time, people will add their pieces to it because they’ll begin to understand that information about how they had an accident is not actually such terrible information to share with people. But they don’t have to share their secrets.

Q: The machine is getting smarter and smarter?
A: Over time, the machine will know more about shipping than any one person could possibly know.

Q: Will the machine replace humans in the future?
A: No. Making machines intelligent is not about replacing humans. Humans are cheaper than computers. We could have an expert on everything on every ship if people could do that but they can’t. We need to create special purpose intelligences that are not like any person and can do what no person could. The problem is that humans who are knowledgeable about something are hard to find and may not be there to tell you when you want to hear it. So they write long books or articles which not everybody reads or remembers. We don’t worry that books will replace humans. Here, we are replacing books, not people. There is a lot of information in books but that information is hard to access. Computers can have a lot of information too. The problem is making that information find you rather than the other way around. As a decision maker, I want to be fully informed at any given time. The old technology was books and manuals. Before books there was word of mouth. Now the new technology is a machine smart enough to know what you need to know. It’s not going to make the decisions for you. But, it will help you know what has worked before and what has not and it will get smarter over time.