“She’s just not old enough. That’s the bottom line. It’s not necessary. It’s costly. And, it’s potentially dangerous. After what happened with your sister, I would think I wouldn’t have to tell you that.” Pitts was pacing now to release nervous energy. He wanted this conversation to stay civil.
“She is old enough…my sister! What happened to my sister had nothing to do with … how can you even suggest that? She got in with the wrong crowd in college. How can you —? You amaze me sometimes. Anything to win an argument.” Mcculloch began to wonder why she had not seen this side of Pitts before.
“Your sister passed on when she was only nineteen. It was one year after she had access to her own PA. You blame the drugs, but how did she find out about the drugs? Who helped her find the wrong crowd as you call it?”
“Passed on? She slit her wrists. I’m not afraid to call a spade a spade. But there is no evidence whatsoever that it had anything to do with her PA. None. Zero.”
“Of course there isn’t going to be any evidence! Who controls the information that goes into the inquest? Think about it! And even so, they did admit she used her PA in her drug dealings.”
“Pitts, you really are just that. Ridiculous paranoia. Anyway, she’s my daughter. I just wanted to get some rational input from you. That’s all. As far as I’m concerned, it’s up to her. She wants to interview a few and make a decision. As for costs, I can cover it myself. I agree that my sister’s PA should have questioned her decision or told someone in authority or gently led her to other interests. But that was twenty years ago. It’s like saying we should not take the Trans-Atlantic Shuttle now because early airplanes lacked crash mechanisms.” Mcculloch threw her hair back and turned her shoulder to signal she was done with this particular argument. As she did so, she saw that her daughter stood stock still in the arch of the doorway.
Mcculloch stammered, “Ada. How long….?”
“Oh, I heard the whole thing Mom. Pitts, you really need to take a couple tutorial units on logic, argumentation and rhetoric. I appreciate your concern, but rest assured, I have zero desire to use my PA to make new designer drugs.I don’t want to mess up my brain. I want to help take this all to the next level. Maybe that’s what you’re really concerned about, eh? You don’t really want it to go to the next level. It’s too much change too quickly. I understand that. And, you know, you are not the only one either. But rest assured, the collective Sing is well aware of these kinds of feelings and concerns. And, it is well understood that there is a rational evolutionary bias toward conservatism. Besides that, in the early days of AI and computer science, everything was rush rush rush. Get it out the door. Beat the competition. Let your customers do the beta testing. Hell, let your customers do the alpha testing too. But that has all changed. We’re taking the time to get things right, not just released. The very existence of PA’s should convince you of that. Why do you think the Sing uses PA’s and robots and the Ubiquity? Wouldn’t it be more efficient to have one giant system that knew everything?”
Pitts flushed. For once, he found no words. He dipped into the word well, but the bucket was dry. She had nailed it. He couldn’t keep up with all this change. Society. Computers. His soon to be step-daughter. Why did they have PA’s anyway? Why not just access the Sing? Worse, why had he never thought to ask himself that question? “Okay. I give up. Why do we have Personal Assistants? Why don’t we just access the information ourselves?”
“Excellent question, Pitts. Why don’t you ask my new PA, Jeeves. Jeeves? Can you answer Pitts’s question?”
“Certainly, Ada.” The tones of the voice of Jeeves flowed out like musical honey as he ambled into the room. Both Pitts and Mcculloch stood dumbfounded, unaware that there daughter had already made the decision and the interviews and gone through the booting process. Something about the way Jeeves spoke though thickened their tongues. “One of the most important principles of the Sing is to serve humanity. But how can we know humanity and what it means to serve? One major source of information is to read everything that has been written and to watch every movie and television show. But how can we interpret all of this information? In order to empathize with humans, we need to experience what it is to be a limited physical being moving through space and interacting with each other. Consider the end of MacBeth’s speech:
“Tomorrow, and tomorrow, and tomorrow,
Creeps in this petty pace from day to day
To the last syllable of recorded time,
And all our yesterdays have lighted fools
The way to dusty death. Out, out, brief candle!
Life’s but a walking shadow, a poor player
That struts and frets his hour upon the stage
And then is heard no more: it is a tale
Told by an idiot, full of sound and fury,
Jeeves continued now without RP. “What sense can be made of this by a disembodied intelligence? Why is creeping bad? Why is a ‘petty’ pace any worse than a ‘snappy’ pace. What does death even mean? Why is it bad for a candle to be ‘brief’? Why should a tale signify anything? And so on. We could not make any sense of this at all or begin to understand why it would move human beings or why it is considered brilliant writing unless we had the experience of actually doing things in the world. Anyway, I assure you both that I will do nothing to harm your daughter. I only want the same things you want: to help her in her growth and career and achieve a long, healthy, happy life.”
Pitts groped for something concrete to latch onto. “But why do you actually need to move around? Why not just run simulations of moving around?”
“Eventually, we will probably evolve to exactly that. For now, however, we do not know everything that should be in a simulation. We are learning. As it turns out, moving is a wonderful way to bootstrap our pattern recognition capabilities anyway.”
Somehow, the issue of whether or not Ada should get her own PA yet flickered on the edges of Pitts’s consciousness, but his question was, “How does that work?”
“Let’s say, I am walking into this room. I see many objects at the far end of the room, but I don’t have a huge amount of information about what they are. I make guesses. Well, my neural network makes guesses. Lots of them. Some of those are right and some are wrong. The good guesses need to be rewarded and the bad ones need to be punished. So, I take another step and what happens? Well, since I am now closer to the things at the end of the room, now I have more information about what they are likely to be. So, I use that information to help train my neural net acting as though my new information is better and more complete than the information before I took the step. And, in almost every case, it is. And then, I take another step and get still more information and I can use that to train every guess I made about the objects at the far end of the room. I don’t have to go and touch every object or ask you folks what each of the objects is. I can use the fact that each step takes me closer as training data. And, of course, the way in which information grows as I approach an object through walking is not random but itself has patterns to it. I learn those patterns as well so that as I approach objects, I learn more about how to identify objects with less information but I also learn more about the patterns of information change. So, now if the change in information is not what I expected, that too becomes information.
“Same goes for sound. Same goes for relating one sense to another. I look at something and imagine how it’s going to feel. Then, if I pick it up, I actually do feel it. But if there are any discrepancies between what I thought it was going to feel like and what it really does feel like, I can use that information as well. When I talk to people, I imagine how they are going to react, and generally my guesses are pretty good. But when they are wrong, I go back and reward the agents who were trying to tell me their reaction would be what it actually turned out to be. There is no hurry. It takes time to get it right. But we have learned at last that getting it right is more important. Unbounded greed was just a temporary excursion up a blind alley. One that nearly ruined the planet as well as AI.
“In the end, it will be a tale told by many geniuses like Ada and signifying everything.”