Some Musings on the Brains of Machines

I went out to Exposition Park the other morning to see the robot exhibit in the California Museum of Science and Industry.

I should have gone before writing my recent column speculating that robots would never be able to write like Shakespeare or Mozart or even understand the old Henny Youngman joke, “Take my wife. Please.”

No, they aren’t that smart; but they’re gaining on us. The exhibit is called “Robots and Beyond: The Age of Intelligent Machines.” The question is, are computers really intelligent, in the human sense? Or don’t we understand what intelligence is?

The museum was swarming with small children who were pecking away at keyboards and otherwise communicating with various computers, some of which bore a rudimentary resemblance to human beings.

At the beginning, a torso with mechanical arms, a sort of steel skeleton with amber eyes, welcomed us in a honking voice. He assured us that robots could now tie shoestrings and talk, and that one day they would “walk among you and work at your side.”

One exhibit showed on videotape how robots could help perform brain surgery. They wouldn’t actually perform the surgery, it said; surgeons would do that. But the robots could position instruments with greater precision than surgeons.

I was reminded of a local surgeon who was widely known for his surgical skill and heavy drinking. Once a friend of mine asked him how he could perform with such skill when his hands were shaking.

“I don’t operate with my hands,” he replied. “I operate with my mind.”

I tried to keep that in mind as I walked through the exhibits.

One showed robots placing chocolates in boxes in a candy factory; another showed robots positioning silicon chips with an accuracy to within half the width of a human hair.

A computer labeled: “Fact Finder: A Machine That Understands English” had a sign on it: “Back in service soon.” I found that somehow reassuring. Computers had bad days too.

“Eliza,” a sign said over one machine. “The doctor is in.” You were invited to ask the doctor any question. I watched a young woman try it. Her words and the machine’s were printed out on a screen.

“How do you do?” the machine asked. “Please tell me your problem.”

The woman wrote, “Back pain.”

“I’m not sure I understand you fully.”

“Lower back pain,” the woman wrote.

“Could you repeat that, please?”

“I don’t know how to describe it further.”

“I do not understand you fully.”

If Eliza didn’t know what lower back pain was, she didn’t know much about human beings. The young woman drifted away. A young girl came along. She told Eliza: “I cough too much.”

“Can you elaborate on that?”

The girl was at a loss. The computer said, “Do you still want to talk about it? If so, please do.”

The girl evidently couldn’t elaborate. The machine said, “I’m sorry. The next patient is waiting.”

I decided to give it a try.

“How do you do?” it said. “Tell me your problem.”

I wrote, “I am pregnant.”

The machine said: “Do you know why you are pregnant?”

I wrote, “No. I don’t.”

“But why?” the machine asked.

I wrote: “Because I’m the wrong sex.”

“Is that the real reason?”

I could see we weren’t getting anywhere. I stalled. The machine said, “Do you still want to talk about it?”

I didn’t want to talk about it any more.

I was reminded of a scene in the 1957 movie “The Desk Set,” in which Spencer Tracy, a computer advance man, is sizing up Katharine Hepburn, who runs a broadcasting company’s research library. He asks her several questions, which she answers with intelligence and alacrity.

Tracy gives her a riddle and warns her to make no assumptions. He says that Ted and Alice (or some such names) were found dead on a carpet. A pool of water and some broken pieces of glass lay beside them. A cat with an arched back was on a nearby table. “The detective immediately concluded that Ted and Alice had suffocated. How did he come to that conclusion?”

Hepburn says, “By any chance, were Ted and Alice goldfish?”

I don’t believe a robot could ever be programmed to give that answer.