When I so much as think those words, let alone write them, I hear the voice of President Andrew Shepherd (and I hear Sorkin’s touch with words) in The American President. He was talking about “America,” he says. The line goes like this

America isn’t easy. America is advanced citizenship. You gotta want it bad.

“Wanting it bad” in the context of Shepherd’s re-emergence in his own presidency, means doing the things that need to be done; in particular, it means publicly opposing Senator Rumsen’s bid for the presidency.

There are, of course, other things it might mean and I am after one of those other things today. I am aware that I run the risk of sounding like Huxley’s John, the Savage, [1] but I really think that “good enough” is not really good enough any more. I think that performances are not enough to sustain the relationships we are capable of. I think that is as nearly true about Alexa as it is about my friend Alyx. [2]

Judith Shulevitz has an amazingly good article in the November 2018 issue of Atlantic. At the end, reflecting on the meaning for us of the rise to prominence of voice-activated “personal assistants:”

If I have learned anything in my years of therapy, it is that the human psyche defaults to shallowness.

You can set that remark against a rich background, which she develops in her article, or you can perch it precariously on top of a shallow one. Here’s the shallow one.

We cling to our denials. It’s easier to pretend that deeper feelings don’t exist, because, of course, a lot of them are painful. What better way to avoid all that unpleasantness than to keep company with emotive entities unencumbered by actual emotions?

And here’s the deep one.

Evolution has not prepared me to know. We’ve been reacting to human vocalizations for millions of years as if they signaled human proximity. We’ve had only about a century and a half to adapt to the idea that a voice can be disconnected from its source, and only a few years to adapt to the idea that an entity that talks and sounds like a human may not be a human.

Or, to say the same thing another way, evolution has not prepared us to treat personal assistants like Alexa as the things we know they are. You can know that Alexa is just a machine at one level of our minds and not know it—in fact strongly deny it—at another level.  “The family” here gathers around a voice-activated assistant in very traditional probably borrowed from the early radio era.

The school children Sherry Turkle recruited to try out cute little interactive robots [3] came up with a way to split the difference. The robots, which they understood were only on loan to them, were, the children decided, “alive enough” to have a relationship with. Not a bad first formulation, I think.

It recognizes the internal response they have formulated in “response” to the performance of the humanoid little robot. This response is entirely genuine. It is just like the responses they formulate to respond to their parents and their teachers and their classmates. That won’t be good enough for very long.

Shulavetz understands the work it takes to continue to be aware that Alexa is only a machine simulating affiliation with her. She understands the work that has gone into making the machine a tolerable entity—both personal enough and impersonal enough—but I don’t think she really deals with why people would do all that work. And I don’t think we will.

That’s where President Shepherd’s warning comes into play: You’ve got to want it bad. Will we “want it bad?” Will we continue to do all the work that keeps a buffer between us and “her?” [4] Probably not. The people who are building the algorithmic patterns that would be called “character” in a human being, understand that people want their personal assistants to be acceptable as well as desirable.

An example of the “acceptable” criterion is that Google Assistant, [says Emma Coats, who is designing the “character” part] said, “should be able to speak like a person, but it should never pretend to be one.”

Speaking like a person enables us to relate closely to her. But she will have to help us recognize the distance that is really there, and that is where the “never pretend to be one” comes in. Coats means that the presuppositions of the assistant’s responses can’t be too personal. That would destroy the buffer.

For instance, if you ask Google Assistant, “What’s your favorite ice-cream flavor?,” it might say, “You can’t go wrong with Neapolitan. There’s something in it for everyone.” That’s a dodge, of course, but it follows the principle Coats articulated. Software can’t eat ice cream, and therefore can’t have ice-cream preferences.

Note that the evasive answer—You can’t go wrong with Neapolitan”—avoids implying that she herself has a preference. On the other hand, it is something that might be said by someone who did have a preference and wanted to point to the range of acceptable options rather than to her own choice within that range.

But there is nothing that will stop us from emotional attachment to Google Assistant (or Alexa or Siri) if we continue to formulate personal responses that signal emotional attachment in other—in interpersonal—settings. And if the ratio of pleasant emotional performances we experience tips strongly in favor of the robotic over the personal, will we not come to prefer those kinds of responses? And will we not come to demand them from our friends; from people, in other words, who actually do have a favorite ice cream flavor.

Will not the “acceptable range” kind of response come to be “the way we relate to each other now” and will not there be pressure to bring the responses of our friends into line with “the way we say it now?” This brings us to actual people being required by other actual people to emulate a verbal style [5] that was developed to maintain a buffer between real people and robots. How weird is that?

I see that path of adaptation as clearly as if I had already seen it happen and was just reflecting on it. It is hard for me to see what will prevent it. There is no reason why the software designers would want to prevent it. In building the buffer, they have done their part. If it is not going to happen, we are going to have to define what real relationship is and we are going to have to want it bad.

We have not yet had any reason to demand real relationship as if it were something we could have if we demanded it. “Real relationship”—meaning relationships with beings like ourselves—has been the default category. When there were only humans, all relationships were “real.”[6] And now that there is another kind of interaction that meets many of our needs—our needs for emotional resonance as well as for information—we need to have a reason to demand relationship.

Not to presume them; to demand them. To reject as inadequate “simulations” that are not real, no matter how real they might come to feel.

We will have to want it bad. But I don’t think we will.

[1] “In fact’, said Mustapha Mond, ‘you’re claiming the right to be unhappy.’
‘All right then,’ said the Savage defiantly, ‘I’m claiming the right to be unhappy.’
[2] I will say a little more about Alyx, who, when I met her, was the receptionist at the Holladay Park Plaza Senior Center. This essay is actually about Alexa.
[3] Turkle’s exploration of this difficulty has always struck me as insightful. Now it is beginning to seem prescient. See her Alone Together for the studies her concern is based on.
[4] Or him. Shulavetz has chosen a young man’s voice for her personal assistant.
[5] And this style can be fine-tuned to each of us. Shulavitz thinks that
“…[we] no longer feel entirely comfortable with feminine obsequiousness, however. We like our servility to come in less servile flavors. The voice should be friendly but not too friendly. It should possess just the right dose of sass.” 
[5] Not to say they were all authentic.