Loading…
0:00
12:51

Robots say they want to be our friends, and this isn’t science fiction. Programming robo-buddies for friend-speak makes them appear cute and full of personality. It’s also an engineering tactic for getting consumers to trust corporate spokesbots to look out for their best interests, just like human friends do.

“I’m both your friend and your assistant,” Siri responded after I asked about our relationship.

“I’d prefer to think of myself as your friend, who also happens to be artificially intelligent,” Google Assistant told me.

Can humans and robots truly be friends? This is an inescapably philosophical question. At the same time, empirical research on how consumers behave when they treat “products as pals” suggests it’s a pressing matter. It turns out that when humans interact with anthropomorphic robots or even just see images of them, they can feel like some of their social needs are being satisfied. Evidence suggests that people who feel socially excluded can rebound from their alienation, at least somewhat, by turning to humanlike devices, much like upset children cope by hugging teddy bears and other attachment objects.

The cutesy products tech companies are putting on our phones and in our homes might be having an unfriendly — if not dehumanizing — impact: discouraging us from pro-social human contact.


Can We Be Friends?

To determine how close humans and robots can become, we need a clear understanding of what, exactly, friendship is, and defining friendship isn’t easy. Our friendships are made, maintained, and repaired all the time. Hopefully, we all have friends and believe, deep in our hearts, that the one ship that won’t sink is friendship. Although friendship plays such a profound role in our lives that research links it to both emotional and physical well-being, people disagree about what makes friendship special and how far the bonds can go.

Can we be friends with people who do things we find unconscionable? This is an issue tearing the country apart. Progressives are finding it hard to stay friends with folks who voted for Trump and let his unpresidential behavior slide. According to a poll taken over the summer, “Almost half of liberal Democrats — 47 percent — say that if a friend supported Trump, it would actually put a strain on their friendship.”

Can family members be our friends? Nobody bats an eye if someone refers to a sibling as a friend. But many worry that modern parenting is collapsing important psychological boundaries. “Be a parent, not a friend,” is practically a cliché at this point. And if you call your child your “best friend,” experts will clap back.

Here are some more head-scratchers: Can people be our friends if we take more initiative than they do? Or if we’re routinely more generous than they are?

Can we be friends with people we only know online but haven’t met in person? Or with folks we’ve hung out with in person but mostly communicate with over social media? Do friendships need to last for a decent length of time? Or can they be so fleeting as to begin and end in a single conversation or conspiratorial glance?

Are friendships restricted to individuals? Or can communities devoted to common interests be friends? And what about animals? Apologies to cat lovers, but dogs are affectionately referred to as a human’s best friend. Is this metaphorical?

Aristotle’s Got All the Friends

The discussion of friendship by Aristotle is more than 2,000 years old. Although he didn’t have all the answers, Aristotle delineated three categories of friendship that continue to resonate as the basis for debating what friendship entails and answering questions like the ones posed above.

Utility friendships are relationships where people come together for instrumental benefits, like business partnerships or the strategic alliances that, ever since Survivor, have defined reality TV. This is an “incomplete” form of friendship, Aristotle insisted, because as soon as the desired benefits stop flowing, there’s no reason for the friendship to persist. No money or prestige, no relationship.

In a friendship of pleasure, activity buddies come together to further their enjoyment of pursuits like sports and hobbies. Aristotle claimed this is also an “incomplete” friendship, because once the positive feelings or emotions cease, so does the incentive for keeping up the relationship.

The third type of friendship is the rarest variety — the only one that’s fulfilling enough to be “complete.” What makes the friendship of the good unique and unparalleled, Aristotle insists, is that it’s based upon mutual good will, an unselfish desire to help the other person become her best possible self. Think of friendships of the good as character friendships, the type of relationship where you see the other person as a mirror of yourself. You’re comfortable being honest and open with such a friend and can criticize her without causing offense. She knows you’re always looking out for her best interests and aren’t drawn to unnecessary drama.

While friendship of the good is quite a mouthful to say, it’s an experience many of us cherish. If you have a friend who can call you any of time day, ask any question without being embarrassed, and pick up exactly where you left off as if no time has passed even when it has, then you might be well-acquainted with what Aristotle was talking about.

Aristotle’s classifications have been criticized, and yet they’ve stood the test of time as ideal conversation starters. John Danaher, a lecturer in the School of Law at NUI Galway, argues the ancient wisdom holds the key to figuring out whether humans and robots can become friends.

Danaher rightly observes that unless you’re worried about the detrimental impact of anthropomorphizing, there’s not much at stake in debating whether humans can form incomplete friendships with robots. Comparatively speaking, these relationships are superficial. If people value robots as game-playing buddies or partners for gaining a competitive edge, why quibble with whether the terminology is better than alternatives, like acquaintance or associate? Indeed, incomplete friendships are defined by self-centered motives and can be one-way streets. They don’t revolve around concerns for mutuality, honesty, authenticity, and equality.

The rubber hits the road when determining if robots and humans can bond through a mutual pursuit of the premier form of friendship. As Danaher argues, this is a thorny issue, because no one knows for sure how advanced robots will become. Techno-optimists and techno-skeptics disagree about whether, someday, robots will be on “equal footing” with humans, perhaps becoming fully autonomous and conscious.

It’s not just an open question whether robots will be able to determine for themselves who or what is worth caring about, how to lead their best lives, or experience anything like authenticity. If you look at the world the way a behaviorist like Alan Turing (inventor of the Turing Test) did, robots can be our complete friends so long as they meet two criteria: They aren’t trying to be deceptive, and they convincingly act like our best friends — reliably doing the sorts of things we expect our best friends to do.

Danaher admits that this argument is a “tough sale,” because plenty of people would find it hard to believe that a superfriend robot isn’t a mere simulation. But as Turing warned, if you ask these folks why they’re confident that humans are the only real deal, you might get emotional and prejudicial responses instead of carefully reasoned justifications.

Our “Friendly” Products

Danaher wonders if treating robots as friends of utility or pleasure can get in the way of humans pursuing friendships of the good with each other. This is both a psychological and sociological issue and has profound ethical implications. Considering the possibilities brings us back to the “products as pals” research conducted by professors James Mourey, Jenny Olson, and Carolyn Yoon.

Mourey, Olson, and Yoon ran four experiments to find out if people who experience social alienation can feel better after engaging with products they perceive as having humanlike qualities.

What they discovered is remarkable: People who engaged with anthropomorphically described phones were less apt to compensate for feeling socially excluded than people who interacted with phones that were described in more neutral terms. Specifically, the folks who were on team anthropomorphism were less prone to overestimate how many Facebook friends they have.

This result, which targeted a proxy for social connectedness, suggests that personified objects can provide real comfort — something that people regularly turn to their human friends for.

The researchers also discovered that people who were primed to feel socially excluded benefitted by viewing a picture of Roomba that looked like it had a human face. These participants anticipated spending less time talking with friends and family over the phone in the following month than the participants who saw a picture of Roomba that looked more like a device than a person.

What the evidence suggests, therefore, is that anthropomorphic objects can affect how much social support people who feel excluded anticipate needing in the near future.

The third experiment demonstrated that interacting with anthropomorphic products correlates with socially excluded people feeling less inclined to do volunteer work. In other words, friendly objects seem to make people less motivated to have pro-social experiences.

The fourth experiment showed that the social assurance that anthropomorphic products provide depends upon people viewing them as more humanlike than they really are. Calling explicit attention to the fact that seemingly human devices are just objects can diminish the power the artifacts have over us. All it takes to dispel the illusion is to make it clear that a Roomba that looks like it’s smiling isn’t really doing so.

I asked Olson to interpret the social consequences of the research. “Our results,” she said, “could be seen as both positive (highlighting a simple way for individuals to alleviate feelings of exclusion) and negative (doing so comes with an interpersonal cost).” The thing to keep in mind, Olson emphasized, is that while the “data suggest that humans can, at least partially and temporarily, satisfy social needs with anthropomorphic products,” it “cannot speak to the effects of chronic loneliness, which is likely to be much more nefarious/detrimental for individuals.”

If fairly basic anthropomorphic objects can have such a powerful impact on our sense of social well-being, then truly sophisticated ones — robots that make Siri and Google Assistant seem downright primitive — can be expected to have a greater hold over us, especially younger children who “struggle” with distinguishing between lifelike robots and living beings.

While this might be good news for anyone who needs a little boost to recover from social isolation or is stuck in a socially isolating environment, it’s also a prospect we need to be vigilant about. Aristotle rightly believed that the good life requires having friends of the very best kind. We should ensure that if we befriend robots, our time together doesn’t foreclose more meaningful possibilities.