This was a really good story, though I was expecting something more to happen after the reveal. I'm a programmer, and I was thinking about all the little details that one would explicitly have to put into a program to make it seem humanlike. It had seemed in the early stages of the story that the programmers who made Sandy were basing it on the
Turing Test for determining if something is an AI, which is an incredibly flawed concept in that it does not test the intelligence of the subject but of its ability to mimic humanity, with all of its inherent flaws. To pass the Turing Test, a machine would typically have to simulate human thinking lag, would have to refuse to do too difficult math problems, would simulate lack of knowledge, and simulate emotions that would make it irrational.
Some of Sandy's flaws seemed very humanlike, but to a point that was well beyond acceptable bounds for its situation, which really made me wonder who had designed the damned thing, and had gone to so much effort to simulate humanlike behavior even at the detriment of functionality.
Then the jig was really up when Sandy passed the image rotation test. It's not inconceivable for an AI to eventually be able to solve that, but the fact that that was the way in which the society in this story is able to shut out bots from their systems implies that this is not trivial. At that point I was wondering if Liu had screwed up and not considered this idea, or whether the AI developers had surpassed the verification developers so that they had no currently workable system.
So the reveal was very satisfying when it suddenly all came together. Manuela presumably has the Internet at her hands, so she can quickly look up trivial information very quickly, like when they were watching baseball. But although she could quickly look up the rules to chess, that does not mean that she could instantly LEARN how to play chess like an AI presumably could. The mistakes she makes, the sleeping on the job, it all makes sense.
It also all makes economic sense in the context of how this situation could come together. This company will make a killing selling these robots as the pinnacle of technology, taking a huge portion of the market share. All you need to do this is fine motor control of the robot, a motion capture system for control, and a telecomm system able to handle the load. It will do so very cheaply, and probably before true AI technology has been fully developed. If they play their cards well, their secret may take a while to get out. Maybe they're developing AI on the side, or maybe they're just skipping the R&D expense since they have a solution in place already. At the same time, the company is taking a big risk in that they will at some point have their pants sued off when one of their workers messes up and kills someone. Our protagonist is right that a waiver wouldn't protect the company from a legal battle if the robot ducked into traffic.
One way that the company's scheme might be revealed is if they enter some area that doesn't get a wireless signal, either someplace rural or underground. Since the robot is controlled remotely, it would just stop at some point.
Another way they might be caught is by running afoul of the FCC. Any consumer product that uses wireless communication has to pass certification by the FCC. If they are not admitting that these are remote controlled, then maybe they don't point out the wireless signals they send (if they did, how could they justify a sustained signal?).
I think Mr. Liu is encroaching on Mike Resnick's territory.
Lately Resnick's stories have seemed, to me, to have such an effect. I can just see too clearly when he's trying to manipulate my emotions. I can see the man behind the curtain too clearly. I find that Liu is much more effective at this, at least for me, considering stories like this one and The Paper Menagerie. So if there were such a thing as claiming emotional territory, I'd say that it's Liu's territory already.