Zara managed to ooh and aah in all the right places, but it was clear to both of them that she was struggling against her attraction to Zhu more than she was paying attention to the things he was showing her. After all, that was the point, wasn’t it? She kept that uppermost in her mind so she wasn’t too ashamed of herself for the fight she had to put up to not give in to the drug’s effects. And she kept solving number problems in her head.
Before her talk, she had him take her to the ladies’ room. She threw up like she did every time before she gave her talk. From experience she knew, once she got started, she would be fine, but the idea of standing before colleagues, others interested in AI work, always made her feel incredibly sick. She knew if Zhu was aware she was ill, he would think she had something to hide. He would never consider it nerves. She carefully rinsed her mouth and ate the strong peppermint candy she always carried before rejoining him.
“I’d like to take you on a tour of our city,” Zhu said as he brought her to the auditorium where they’d set up a podium for her. Her briefcase was there, sitting right beside the glass of water provided for her.
“I’d love that.” She’d be long gone, thankful she’d escaped with her life.
He took her straight to the podium, and Zara immediately slipped into her role. She hated everything about her life but this—talking about what she loved and believed in with those interested. That, more than anything else, always allowed her to escape the horrible shyness that made her the worst traveler ever. She had developed the character everyone saw and believed, and she hid behind her. Once she got past her nerves, she could settle into explaining the program and why it could be so helpful on so many levels.
Zhu stood to one side. Close. Beyond the lights she could see half a dozen men with automatic weapons at the entrances. She pretended not to, but it was a very definite fight to keep her heart rate normal.
At her introduction, conducted by another very charming man in a suit, the applause was enthusiastic. She wondered if Cheng had threatened all of them—applaud her loudly or my goons will shoot you.
“Good evening. My talk is called the VALUE system, the program you’d love to have as a partner. I think you’ll see why in just a moment …” She trailed off and scanned her audience. She’d given her talk dozens of times already and knew it was cutting-edge. They would be hanging on every word if they were really interested in artificial intelligence and what it could do for them.
She reached out to the machines on the first floor. The computers. Touching them with her energy, that psychic gift Dr. Whitney had so carefully enhanced. She could talk to machines, and they listened with rapt attention just as these people were listening. She had the ability to serve as a wireless conduit between the remote computers and her wireless hard drive. She instructed the remote computers to transfer their data from every one of the computers, floor by floor, and store it in the PEEK-carbon nanotube hiding the SSD in her brain.
“AI game-playing systems since the 1960s have been fixated on winning. Every twenty years there is a quantum leap in AI programs’ ability to win. Arthur Samuel built the first self-learning program in 1959, a program that learned how to play checkers increasingly better over time. The program reached a respectable amateur level status of play by the 1970s. Fast forward twenty years, and in 1997, you could watch the deep learning program, Deep Blue, beat the reigning world chess champion, Garry Kasparov—an amazing accomplishment! Fast forward another twenty years, and in 2017, you see Google’s deep learning program, AlphaGo, beat the reigning world Go champion.”
It took time to transfer the amount of data stored in the computers in Cheng’s facility. It would take as long to destroy every hard drive to ensure the man had none of the data on the GhostWalker program given to him by the treasonous senator Violet Smythe. Zara kept her voice even and calm, so that later, when Cheng and Zhu compared it with other speeches she’d given, there would be no difference. Inflections would be the same. She wasn’t under undue stress. She couldn’t possibly be the reason they lost the data on every computer. She was incredibly thankful for her mind’s ability to work on solving number problems. In doing so, it had lessened the effects of the drug they’d given her enough for her to control the systems in her body.
“But here’s one thing we have yet to see … What about a program that could learn to intentionally lose when playing a little boy, so that boy could experience winning? What about a program that could learn how to propose ‘win-win’ solutions for itself and someone else? What about a program that knows that ‘you can’t always get what you want’ and learns how to ‘get what you need’ by making good trade-offs given limited, competing resources—time, money, people, materials, et cetera?”
The idea had been talked about for years. For trade, such a program would be invaluable. It was expected that there would be a breakthrough sooner or later, but to be able to stand in front of them and announce it had been accomplished was exciting. Every. Single. Time. She had to be careful to never lose sight of why she was really there. She needed the information in those computers. She’d done this so many times, but she’d never had to destroy the hard drives. Most businesses or universities had no idea she’d taken anything out with her when she went because she only gathered information; she never left evidence that their computers had been touched. Destroying the hard drives of every computer in the building would definitely raise alarms.
“In this talk, I’m going to describe a program, the VALUE system, which integrates an entire suite of learning techniques, some old and some new, to do just that. The VALUE system integrates: the inverse reinforcement learning techniques of Russell and Ng for learning the value of others, our earlier deep learning techniques for creating and refining negotiations and compromise in a two-party circumstance, our new supervised learning techniques for reformulating design spaces based on human guidance with acceptable trade-offs.”
She launched into her talk, trying not to get lost in the excitement of the artificial intelligence world and the endless possibilities that always consumed her mind when she allowed herself to become fully immersed there. She had a job, a much more important one, in terms of serving her country, saving lives and getting out of there alive.
As each of the computers gave up its data, its hard drive destroyed itself, wiping out all documents, making certain no trace remained. It was a big facility, but she was used to delivering her talks while making the data transfers. She was certain the flow of information to her would never be detected, so she was never nervous. It was a matter of instructing the machines in any chosen building to cooperate. She didn’t need to hack in, or figure out passwords. She just needed a wireless environment. Destroying the hard drives after was a much riskier thing to do, and she’d never done it before. That left footprints. No one could prove she had anything to do with the losses, but she was there. On-site.