The Sparrow - Mary Doria Russell [14]
"Dr. Yanoguchi, I have been thinking about the AI program," Quinn began. "I know my job is pretty mechanical and I understand that it makes good business sense to automate what I do, so I’ve begun thinking about going back to school for a Ph.D., and it occurred to me that you and ISAS might be interested in the topic I hope to use for my thesis." Jimmy paused, brows up, looking for permission to continue. Yanoguchi nodded, apparently relieved that Quinn was not there to fight. Pleased with the sincerity of his own performance, Jimmy warmed to his topic. "Well, sir, I would like to attempt a little pilot project, a comparison of an AI astronomy program with the human subject it was based on. I’d like ISAS to use a first-rate AI analyst to develop the program. Then I’d do a side-by-side comparison of the program’s data handling with my own, for perhaps two years." Yanoguchi sat up a millimeter straighter. Jimmy smoothly amended his proposal. "Of course, a year or even six months might be enough, and then I could work up a grant proposal. I might be able to come back to work here, on grant money, later on."
"Mr. Quinn," Yanoguchi said at last, "it could be argued that the results of such a comparison would be suspect because the subject held back critical information."
"Yes, that’s true, sir. But that might be true of anyone who resented being the subject of an AI analysis, sir. I’m sorry, Dr. Yanoguchi, but it’s common knowledge that most people do hope the programs will fail. I think that the use of a really good AI analyst would mitigate the possibility that the subject is holding back. Plus, since I’d be using the data myself in my thesis research, I’d have a personal motive to make sure the results were reliable." Yanoguchi said nothing but he didn’t exactly frown, so Quinn continued. "It seems to me, sir, that it might be in ISAS’s interests to have some kind of hard comparative data, to judge each AI program, wouldn’t it? To see if a program misses things that humans pick up? And if that’s not so, then the Institute can go on using artificial intelligence to eliminate low-level jobs like mine, knowing that it’s truly as competent as the people it was based on. It’s just one more aspect of the system that could be nailed down properly, sir." Jimmy waited a few moments and then said thoughtfully, "Of course, it’s just a little pilot project. If it doesn’t work out, you’ll only have gambled six months’ extra salary for me. If it comes to something, it would reflect well on Arecibo ..."
And on Masao Yanoguchi. Who said nothing. Jimmy forged ahead.
"If you have no objection, sir, I wonder if we could get Sofia Mendes to do the analysis. I’ve heard she’s very good and—"
"Very expensive," Yanoguchi pointed out.
"But I have a friend who knows her and he says she might be willing to do the project for the publicity. If her program beats me, her broker could use that to command higher fees. Maybe we could work something out with him. If she wins, ISAS could double the usual fee?"
"And if she loses, the broker gets nothing?" suggested Masao Yanoguchi thoughtfully.
It’s worth considering, Jimmy urged Yanoguchi mentally. Very little downside risk. Take a chance, he prayed. But Jimmy didn’t expect an answer and didn’t press for one. Yanoguchi would never say yes until he’d gotten a consensus about the project from everyone in ISAS and maybe even beyond the Institute. A lot of people had a lot riding on artificial intelligence. And that was the beauty of the thing: the longer the Japanese took to make a decision on this, the longer he had a job. And if they said yes, he’d be around for the months it took the vulture to pick his brains and then for at least another six months to do the comparison. If he beat the program, he’d be