Howard S. Smith | Will robotics advance on the battlefield?
'I, robot' author says military is the most likely place for artificial intelligence machines
- By Patrick Marshall
- Feb 20, 2009
Howard S. Smith, a Massachusetts Institute of Technology-trained engineer and artificial intelligence expert, recently published a new book, “I, robot” (Robot Binaries & Press), a techno thriller that serves as a modern update to the original “I, Robot” by Isaac Asimov in the 1940’s. Smith was the founder and president of Optimal Robotics, which patented, designed, built and installed the first supermarket self-checkout machines, which were originally called “service robots.” Smith currently offers consulting services in the area of robotics and artificial intelligence through Robot Binaries & Press Corp.
GCN: How did you get involved with robotics?
Smith: I’ve been interested in artificial intelligence and robotics my whole life.
My engineering degree is from MIT, though I studied biomedical engineering as opposed to AI. About 15 years ago I wanted to do a project and I was looking at what would work. Developing robots requires so much money for research and development that it’s critical to find the right market.
The problem is to build something like Rosie the robot of “The Jetsons” that would come and clean up your kitchen. But it’s not going to work because it would take billions and billions of dollars to make and there’s not enough of a market for it. So you look at what would work.
Another market was replacing cashiers. At the time, automatic banking machines existed. IBM and a few other large companies said, “Well, now that we have bank machines we’re going to try and replace the cashiers in supermarkets and other retail stores.” I thought, wow, they have no idea what they’re getting into. You need a lot of intelligence to do what the cashier does. It may be low on the totem pole compared to the company president, but you’re making a lot of decisions when people are checking out.
I thought that would be a good application because the market is big enough. It’s a good application. The first working supermarket checkout robots were mine.
GCN: How were the check-out machines received?
Smith: There were demonstrations when the first machines went in. Unions said they were going to take away jobs. But it turned out it did a lot of the tedious work and it allowed people to do a lot more interesting work in the store. It turned out to be a big success. There are loads of the machines everywhere. Ultimately, the technology was sold to NCR and Fujitsu.
My original machines actually did a little bit more than the machines they have now. I had an optic system in there so it could actually look at things for pattern recognition and such. They also had a personality. If you, say, were buying 25 Hostess Twinkies the machine would ask you if you really needed that. But supermarkets tend to be very conservative organizations and I had to take out the personality module.
I had these megalomaniac dreams of placing my machines throughout the whole world and using that as a starting point. Unfortunately, since the 1990s there hasn’t been much improvement in the machines, which is a bit disappointing to me. I did the welding. I did the soldering. I did the electronics. I did the software. I did the marketing. But I guess I was a bit weak in the financial aspects of raising capital for the company. I guess that was my downfall there.
GCN: You have mentioned that even though you didn’t want to develop military robots you could have made money more easily in that sector.
Smith: Yes. But that was something I didn’t want to do.
It takes so much money to make these machines that the military route is the route that will fund them.
For the military, it’s not a cost-sensitive device. For western societies, politically no casualties are acceptable. It’s not even a question of being cost-effective. Regardless of the cost, robots will be bought and they will be used. If you have a machine that will kill other people there is an unbelievable market for it.
GCN: You’ve said there are that there are already approximately 6,000 robots in the use in the Middle East.
Smith: Yes. But those are just the ground ones. That doesn’t even include the ones in the air.
Building air robots is much easier that ground robots. [To control their movement,] it’s just X,Y and Z – up, down and left-right.
GCN: So a human still has to control the robot’s movements?
Smith: Actually, they do have a fair amount autonomy. It’s not like flying one of those planes you can buy at ToysRUs in your back yard. Because you are so far away from the machine, there are delays in giving them commands and the planes actually need to have a bit of intelligence. They can do a lot of stuff autonomously. With the ground robots, too, we’re finding that the more autonomy you give them the better results they get.
The other thing, of course, is that people started mounting guns on them, weaponizing them. But the official doctrine is that there is still a person in the loop always.
GCN: Should we be worried about terrorists getting their hands on robots and using them against us?
Smith: Not really. Life is cheap for terrorists. Why use robots? They are expensive machines.
GCN: Will the ability to employ robots make it easier for us to undertake wars?
Smith: I think it will. When I was in going to university the doctrine of mutual assured destruction, or MAD, kept people from killing each other. The analogy is bombing. It’s easy to bomb people because in the worst case a pilot will get shot down. But the risks are low.
Would a Western society be willing to send in a division of robots even if the cost was, say, a billion dollars? Probably yes. The political cost of losing a billion dollars worth of robots is much less than that of hundreds of casualties.
And it’s a part of the doctrine now. There is $125 billion being allocated by the [Defense Department] for the next 10 years. One-third of troops are desired to be robotic within 10 years. Huge amounts of money are being spent right now.
GCN: What are the next developments we should expect?
Smith: For military robots it’s a question of giving them more autonomy, and of having groups of them together.
For non-military ones, it depends on markets. I think we’ll start to see the Japanese model [of developing robots] to assist in nursing duties. But we’re talking a couple of years for those things to come out.
GCN: Is there anything policymakers should be considering?
Smith: We need to be aware of the subject so that when it starts to happen in a couple of years, do something about it rather than it just coming under the radar. That’s what happened with nuclear weapons. They just happened so fast that nobody had a chance to react to them.
Robots will make war much easier. If machines were a little bit farther along you probably wouldn’t have all the outrage about Iraq. You still have some leftist groups upset, but probably most of the population wouldn’t care. That’s going to happen in a few years. It’s a horrible idea that we create these machines, but from a military point of view the machines work.
GCN: Are robots covered under any international standards for conducting wars now?
Smith: No. It’s slipping under the radar right now.
It’s going to happen fast. In the 1960s and 70s, right after we landed on the moon, we really expected AI and robots to go fast. It didn’t happen, but it doesn’t mean it won’t.
Patrick Marshall is a freelance technology writer for GCN.