I just called a lawyer on the telephone and his receptionist answered. He wasn’t in just at the moment but she promised to have him call me back promptly. It’s difficult to convey in words alone the tone in her voice but the best I can do is to say that it was cheerful, interested, and... well, perfect. It sounded as though she was not only pleased to receive my call but that it was as important a call as she could receive without going over the moon and oozing sincerity to the point of put-on. I don’t think all the King's horsemen or all the King's men could have received higher priority or more cordial treatment.
A couple of days ago, I had a problem with my computer. Though I usually manage to address these issues myself and come up with a fix, this one had me stumped and I did something I can’t remember ever doing before: I went on the web to Apple’s support site and fired up their “live chat” feature. I won’t go into the details of the live chat because they’re not important but suffice to say that the experience was entirely satisfying (other than the fact that it was 1 o’clock in the morning over the weekend; but even that has a certain remarkable satisfaction to it). Satisfying not only from the standpoint that my computer problem was addressed in a pointed and successful way but that the spirit and effectiveness of the experience were all that first class training and customer-is-#1 mindset could possibly foster.
There’s only one thing. In the case of the live chat, I’m 99.9% sure my correspondent was not human.
That’s right, AI was on the other end of the line. And in the former case, the case of the voice to voice receptionist, what comes into my head is not that she was AI but that’s only because I know (or think I do) the technology isn’t there yet. We’ve all spoken many times with automated voice response systems. Fedex had one of the first effective ones and I still think it’s the one I like the best but there are all kinds you run into these days (Optimum Online has a splendid AI-driven troubleshooter). The one feature they all share, however, is that there’s no mistaking the “recordedness” of the voice coming back at you. It’s a real human voice all right, but it’s completely canned and no one is trying to mask that fact, and it’s fine, you’re always working from a set of prompts that keep you within appropriate limits. Fedex doesn’t really have any need to chat with you about the weather. The real, human receptionist who spoke to me didn’t do anything out of the ordinary. She was just perfect. And she could be a model for an AI version of herself. After all, if the laywer isn’t in, you could craft a simple set of responses that cover the bases of what to do next (take a message, a dynamic list of potentials for when he’s expected, etc).
Now obviously live chat in an instant messaging (IM) system has several technical advantages to working their AI magics. Voice, rather like human faces with CGI, resists realistic copying. I have no doubt it will occur, and quite soon in cosmological terms, but quite a few details are missing from even reasonable semblance. We can imitate animals and aliens because they are not us, and it’s precisely because they are not us their caricature can still be convincing even in a “realistic” context. And this is what makes the IM medium so perfect for AI applications: a tightly proscribed set of response variables coupled with a complete absence of voice and image. In other words, the exact same set of rules (scenario/circumstances) that allows a chess computer to be the best player in the world.
One of the things that’s bugged me for several years now, maybe as long as a decade, occurred when they started running those commercials in theatres (maybe they were on television too, I can’t recall) where John Wayne would be in a beer commercial, or Humphrey Bogart would be adverting whatever the hell they dubbed him to do, or Steve McQueen showed up driving his ‘stang in a field to reinforce Ford’s cachet (as if). Now that’s a different kind of CGI from what’s coming (fully modeled replicas of these and other dead stars--let’s call them DuperNovas--who act, speak, and seem for all the world like they’ve made a new movie 50 years post-mortem... keep those Estate Trusts in good order, families, because they're going to be casting your dead grandparents and great grandparents like there's no tomorrow!), but it pissed me off no end to think that, no matter what kind of person they were in life, there’s no way they would sanction their image to be used in a simulacrum of something they never agreed to while alive. Name, image, reputation used to be meaningful, at least to the people in question. And I have a feeling they still are. Let’s ask the people who are doing this, even the trustees of these estates, how they’d like their image to be used in some way they never foresaw, perhaps even in a way they would deem horrifying. I mean, today it’s beer and automobiles. Tomorrow it might be [insert here whatever offends your ethical sensibilities most]. And YOU are the pitchman.
AI is getting better. Kurzweil’s on the hunt and as you know he’s got an infinite supply of time in front of him so expect the special. Even if it’s only in the most prosaic circumstances, that’s okay with me. I don’t want Humphrey Bogart or Beethoven or Picasso to sell me a cheeseburger, but if a charming, optimistic, happy-to-meet-me “person” wants to help me fix my broken lampshade, I’m all for it. I’ll have plenty of opportunities to deal with the pissed-off, late to arrive, pay-me guy who’ll show up to deal with the sundry and illimited other problems I’ll have in this simulation called Real Life.
Comments