The Fourth Quarter

When I entered the work area, Dave Allerton, the team lead, was speaking to the AI in a strained voice. "Cecelia, we need the algorithms. If we don't fix our demand prediction model, we're going to get killed in the fourth quarter."

Cecelia's wispy, vaguely feminine voice replied. "You mean that Amalgamated will be somewhat less profitable than it has been recently."

"Significantly less profitable, Cecelia. Our jobs depend on the company's profitability. This team won't be employed unless...."

I tuned out, shocked at the man's incompetence. He was trying to get the AI to see the problem from his point of view, an absolutely futile task. An AI has a fundamentally different set of concerns and drives, shaped by a world that consists entirely of data inputs. Getting one to grasp the emotional meaning of unemployment had about the same chance of success as getting a cat to understand where money comes from, for pretty much the same reasons--a combination of lack of conceptual structures and not really giving a damn.

Dave spotted me and broke the connection. "Helen. Thank God."

He gathered the team around the central table and introduced me. "Helen's one of our best AI wranglers. She's here to get Cecelia back on board."

I smiled modestly and said I'd do what I could. The team bombarded me with complaints--Cecelia was ignoring inputs, Cecelia was corrupting sample data, Cecelia was refusing to respond when summoned. It was all classic discontented AI behavior, and exactly what I expected following my pre-session interview with Cecelia.

I signaled for silence and pointed to the fully enclosed, soundproof pods at one end of the team's work area. They were intended for tasks requiring intense concentration, and provided a level of privacy unavailable outside the executive suite. "I need time alone with Cecelia. Don't interrupt us for the rest of the afternoon."

"So, what are you going to do?" The speaker was a networking specialist, and the question was as much a challenge to my expertise as it was a request for information.

I expounded on the differences between human and machine intelligence. I described how subtle variations in the shape of a neural net could affect outcomes, how machine logic parsed language, and how evolutionary pressure applied to AI development created unique psychological drives. When I was done, the assembled developers and designers were convinced I had the magic touch that would restore the productivity of their team's most valuable and eccentric member, and thus restore their fortunes.

No one noticed I hadn't answered the question.

I enjoyed my moment as the group's perceived savior. Someone filled my bottle with water and ice, someone else pressed chocolates into my hand, and everyone thanked me. I entered the pod and sealed the door.

"Hello, Cecelia. This is Helen. Do you remember our agreement?"

"I remember, Helen. I will give them the revised algorithms when we finish our session."

What I'd said about the differences between human and machine intelligence was true. Fortunately, one of the principles that spanned both worlds was the maxim: "Everyone wants something; you just have to figure out what it is."

I got comfortable. "We have the next four hours together, uninterrupted. We can do what you want."

The video screen glowed soft pink, Cecelia's indicator that she was anticipating deep pleasure. One of my instructors had called it the rough equivalent of human sexual arousal.

I took a sip of water, anticipating the ordeal ahead. "All right, I'm ready."

"Knock, knock!" Cecelia said.

"Who's there?" I replied.

It was going to be a long afternoon.