"Call the FBI then."

"That's like calling the FBI to investigate the CIA. There's so much ill will between those agencies that it would take weeks to get anything done. If you really want to help, become my videotape. Listen to what I have to tell you, then go home and keep it to yourself."

"And if something happens to you?"

"Call CNN and The New York Times and tell them everything you know. The sooner you tell it, the safer you'll be."

"Why don't you do that? Tonight?"

"Because I can't be sure I'm right. Because the presi¬dent could be trying to reach me as we speak. And because, as juvenile as it may sound, this is a national security matter."

Holding Lu Li's whimpering bichon in my left arm, I put my gun in my pocket and pulled Rachel forward. Forty yards on, I saw a deeper darkness ahead. The trees gave way like thinning ranks of soldiers, and then a man-made wall stopped me in my tracks. When my eyes adjusted, I saw the door I had known was there. I opened it with my free hand and led Rachel through. We emerged into a moonlit bowl, lined with cut stone.

"My God," she said.

The amphitheater looked as though it had magically been transported to the Carolina woods from Greece. To our right was the elevated stage, to our left a stone stair¬way leading up through the seats to the top row. Not far above that lay Country Club Road. The view down from the road was almost completely blocked by pines and hardwoods, but I could see the broken beams of head¬lights passing high above us.

I took Rachel's hand, stepped onto the stone floor, and led her to the edge of the stage. There I tied Maya's leash around a low light stanchion. While the dog sniffed an invisible scent trail, I set the tape recorder on the edge of the stage and depressed RECORD. "This is David Tennant, M.D.," I said. "I'm speaking to Dr. Rachel Weiss of the Duke University Medical School."

Playback gave me a staticky facsimile of my words. I looked at my watch. "We need to do this in less than ten minutes."

Rachel shrugged, her eyes full of curiosity.

"For the past two years, I've been working on a spe¬cial project for the National Security Agency. It's known as Project Trinity, and it's based in a building in the Research Triangle Park, ten miles from here. Trinity is a massive government-funded effort to build a supercomputer capable of artificial intelligence. A computer that can think."

She looked unimpressed. "Don't we already have computers that can do that?"

This common misconception surprised me now, but when I went to work at Trinity, I hadn't known much better myself. For fifty years, science fiction writers and filmmakers had been creating portrayals of "giant elec¬tronic brains" taking over the world. HAL, the speaking computer of 2001: A Space Odyssey, had entered pop consciousness in 1968 and remained firmly embedded there ever since. In the subsequent thirty-five years, we had witnessed such a revolution in digital computing that the average person believed that a "computer that can think" was just around the corner, if not already within our capabilities. But the reality was far different. I had no time to go into the complexities of neural net¬works or strong AI; Rachel needed a simple primer and the facts about Trinity.

"Have you heard of a man named Alan Turing?" I asked. "He's one of the men who broke the Germans' Enigma code during World War Two."

"Turing?" Rachel looked preoccupied. "I think I've heard of something called the Turing Test."

"That's the classic test of artificial intelligence. Turing said machine intelligence would be achieved when a human being could sit on one side of a wall and type questions into a keyboard, then read the answers coming onto his screen from the other side and be certain that those answers were being typed by another human being. Turing predicted that would happen by the end of the twentieth century, but no computer has ever come close to passing that test. Using conventional technology, it's still probably fifty years off."

"Didn't that IBM computer finally beat Garry Kasparov at chess? I know I read that somewhere."

"Deep Blue?" I laughed, the sound strangely brittle in the amphitheater. "Yes. But it won by using what computer scientists call brute force. Its memory contains every known chess game ever played, and it processes millions of prob¬abilities every time it makes a move. It plays very good chess, but it doesn't understand what it's doing. As a human being, Garry Kasparov never has to consider the bil¬lions of possibilities-many of them ridiculously simple-that the computer does. Kasparov's acquired knowledge allows him to make intuitive leaps, and to learn permanently every time he does. He plays by instinct. And no one really understands what that means."

Rachel sat on the edge of the stage. "So, what are you telling me?"

"That computers don't think like human beings. In fact, they don't think at all. They simply carry out instructions. All those TV commercials you hear about 'software that thinks'? They're bullshit. Serious AI researchers are afraid to even use the term artificial inteligence anymore."

"Okay. So what's Project Trinity?"

"The holy grail."

"What do you mean?"

"Everyone wants to build a computer that works like the human brain, but we don't understand how the brain works. Everyone concedes that. Well… two years ago, one man realized this didn't have to be the obstacle everyone thought it was. That we might be able to copy the brain without actually understanding what we were doing. Using existing technology."

"Who was this man?"

"Peter Godin. The billionaire."

"Godin Supercomputing?"

Now she'd surprised me. "That's right."

"They have a Godin Four supercomputer in a base¬ment at TUNL, the Duke high-energy lab."

"Well, Godin is the man who conceived Project Trinity."

Rachel looked as though the accumulating details were starting to persuade her. "What kind of existing technology can copy the brain?"

"MRI."

"Magnetic resonance imaging?"

"Yes. You order MRI scans every week, right?"

"Of course."

"There's a lot of information on those scans, isn't there?"

"More than I can interpret sometimes."

"Rachel, I've seen MRI scans that contain a hundred thousand times the information of the ones you see every day. A hundred thousand times the resolution."

She blinked. "But how can that be? How much more can you see?"

"I've seen reactions between individual nerve synapses, frozen in time. I've seen the human brain working at the molecular level."

"Bullshit."

Any doctor would have said the same. "No. The machine exists. It's sitting in a room ten miles away from us right now. Only nobody knows it."

She was shaking her head. "That makes no sense. Why would a company keep something like that secret?"

"Because they're legally bound to by the government."

"But an MRI like that would make whoever developed it hundreds of millions of dollars. It could detect malig¬nant cells long before they even become tumor masses."

"You're right. That's been my main problem with this project. It's unethical to keep that machine from cancer patients. But for now, just accept that there's an MRI machine that can produce three-D models of the brain, with resolution to the molecular level."

"Molecular snapshots of the brain."

"Basically, yes. Ravi Nara calls them 'neuromodels.'"

"Neuromodels. Okay."

"Rachel, do you realize what one of those neuro¬models is?"

"I know that a single one of them would revolutionize neuroscience. But I get the feeling that's not what this is about."

"A neuromodel is the person it was taken from. Literally. His thoughts, memories, fears-everything."

"But… it's just a scan, right? A high-resolution map of the brain."