Originally published on July 12, 2016
Only a few weeks ago, I met my first bot swarm.
That morning, I posted a link on my Facebook Wall to one of Donald Trump’s campaign speeches. Cluttering my friends’ Facebook feeds with political commentary is one of my more shameful addictions, but the costs of indulging the habit are small enough that, at this point, the behavior is almost automatic. This time the costs were higher. Like blowflies drawn to a carcass, the bots descended on the link, sharing it, liking it, and leaving one comment after another beneath it.
A representative example was one Yenj Romin, who reposted my comments on ‘her’ own profile. Ms. Romin had been on Facebook for approximately nine weeks. Those who visit her profile will discover that all of her posts are written in capital letters. They proclaim things like “KILL ALL THE MUSLIMS” and “EVERYONE HATES JOHN MCCAIN.” She had four friends, and her profile picture is a drawing of the words “I’m voting for Trump.” This March she enrolled at Barstow (CA) high school, Barstow Community College, and began her gainful employment in the U.S. Army—all at the same time.
Barstow, California is home to a Marine Corps Logistics Base.*
Romin was probably the most colorful bot attracted to the post, but hardly the only one. I wasted a day deleting memes posted by ‘people’ I had never heard of, who do not follow me, and who do not have any friends who are my friends, or even any of my follower’s friends. Some of these accounts were quite old—the oldest (a HillaryBot) had been posting things since 2013, and had just enough pictures of neat family vacations in Miami to allay suspicion.
“In the election bots of today we see small foreshocks of the social shockwaves that will ripple through our society in the future.”
The central bot of Maria Cecire’s “Massively Open” is far more sophisticated than the bots that defaced my Facebook wall in May. No one could ever imagine falling in love with an election bot (or, for that matter, their sexually-charged cousins on Tinder). Cecire imagines a world where that is no longer the case. That world may seem far away, but in the election bots of today we see small foreshocks of the social shockwaves that will ripple through our society in the future. Most discussions in the popular press on the impact of advances in AI technology focus on the economic side of these shocks, returning to the well-tread ground of automation, inequality, and economic displacement. Cecire calls our focus to the social fallout of these same trends. What happens to a social landscape hijacked by bots? How is trust possible in a society where anyone can be a bot in disguise?
The question is a surprisingly relevant one that cuts to the heart of contemporary debates on how the Internet is changing our communities and our willingness to trust the people we meet in them. Though some would like to blame the Internet for all of the country’s woes, America’s failing social capital predates the Internet and social networks like Facebook by decades. In the Internet’s earliest days, many hoped that the web might be the key to reversing these trends. Silicon Valley visionaries, confident that their works alone had the power to mend every dent, dry every tear, and fix every ill, advanced a hopeful array of arguments about the transformative power of technology.
By now these arguments are probably long familiar: the Internet makes coordination possible between parties who would not even been able to communicate without it; blogging, forums, and social networks have created entirely new platforms for social capital to bloom; and the Internet has lowered the transactions costs of friendship itself, so that loneliness never had to be determined by location again.
Two decades have passed since these heady visions of the future were first described. The science is in and the verdict is out: the Internet has not brought about a new age of American community.
Affairs have plodded along much as they did before. The Internet may make international friendships possible, but the people who compose the average Jane’s online network are more or less the same as those who live close to her. The Internet increases the number and strength of ‘weak’ social ties with acquaintances, but does not make a meaningful difference in the number or intensity of the ‘strong’ social bonds we form with trusted friends and family members. Thus, using the Internet improves most measures of social capital, but only on the margins (and far less than fast-disappearing activities like weekly church attendance do). These benefits also follow class lines. Millennials with well-to-do parents are more engaged in associations, volunteer work, and friendships than their parents were at their age; yet those with working class origins are even less engaged than their parents. For America’s poor, the Internet has done nothing more for the strength of their social networks than the television did.
“The marginal benefits Internet use has on social health should not be discounted—meaningful Internet communities have formed, and what is more, formed on nothing more than ID tags and small, pixelated avatars.”
These humdrum findings are uninspiring in the eyes of the technologist, whose lofty visions did not materialize. However, the marginal benefits Internet use has on social health should not be discounted—meaningful Internet communities have formed, and what is more, formed on nothing more than ID tags and small, pixelated avatars. These social bonds forged on digital social networks have not changed the world, but they have proven something remarkable: you do not need to meet someone in person in order to trust them.
In time, bots may put an end to that.
Cecire captures the psychological costs of a bot-dominated social sphere in the character Zoe, a battered down teacher struggling to find purpose in her work. The only bright spot in the MOOC classes she manages are the few “students who get it.” These few students are all that keep her “feel[ing] like [I’m] doing something that isn’t just stamping out implements for a mindless machine.” But in a world where anyone online might be a bot in disguise, the student who ‘gets it’ may have been a mindless machine all along. Zoe faces this revelation with shock:
“What’s real about a bot?” Zoe cried. “He wasn’t even a real student, just a piece of a giant degree scraping scam. He was in my course — in all of his courses — just to get an authenticated degree, not to learn anything. All so his scammer overlords could harvest the wages of whatever remote job he’d eventually steal from a real student. What was the point?” ….Zoe felt her eyes fill again. “I thought I was getting through to someone,” she said softly. “But I was just talking to myself.”
Zoe’s tears are fictional, but her frustrations have clear analogues in the real world. In her cry there is an echo of the very first dude fooled by a Tinderbot into flirting for ten minutes with a piece of code. As AI technology progresses, it is a feeling we will all become more familiar with.
“Zoe’s tears are fictional, but her frustrations have clear analogues in the real world.”
These feelings have consequences. A social network bombarded by bots will soon be a barren one. This is partly because of the sheer tedium of blocking, deleting, or ignoring one bot-message after another. A high volume of stupid bots is enough to drive anyone away from a platform. But there is a second hazard, one that becomes more apparent the smarter the bots become. I might reduce this to a general hypothesis: the more sophisticated bots haunting a social network are, the less trusting its members will be.
The logic behind this hypothesis is easy to grasp. How many false students would you have to put your heart into before you stopped teaching with earnestness? How many political debates would you have with robots before you decided to stop discussing politics with strangers altogether? How many times would you be fooled by dating bots before you give up on online dating altogether?
Bots are not yet sophisticated enough to cast that kind of pall on most online communities they frequent. But the bot hordes of today are sufficiently adept to lend a peek into the likely tenor of things to come. Look no further than the 2016 elections, whose primaries were filled with accusations that electioneers on one side were using bots to drown out the other. Perhaps these claims are true. Perhaps they are not. The truth here did not really matter. What mattered is the accusation. The mere intimation that the other side relies on robots for its strongest support poisons discussion. It is the possibility of astroturfing, not its proven use in every case, that sows these seeds of distrust. The more sophisticated these bots become, the more distrustful we will be. It is not hard to imagine a future in which any honest Internet debater will be tarred with the accusation they are nothing more than a sophisticated discussion bot.
How we will solve this problem remains to be seen. It is still a problem far on the horizon, most visible in fictional tales like Cecire’s. But it is a problem we will one day have to confront: how is civic discourse possible when you cannot be sure the people talking with you are human?
Tanner Greer is a writer and analyst currently based out of Taipei. His research focuses on the evolution of East Asian strategic thought from the time of Sunzi to today. He blogs at The Scholar’s Stage, and can be followed on twitter at @Scholars_Stage.
*It is also the closest city to the Army run Fort Irwin. But if our bot had truly been assigned there then that is where ‘she’ would be. Without a family of her own there is no reason for ‘her’ to be living 45 minutes off base.