A Machine Learning Analysis of Am I The Asshole? Offers New Insights Into Morality
One of Reddit's most popular subs has proven to be fertile grounds for an enterprising social psychologist
Previous generations would ask what Jesus would do. Now, we turn to an online jury to ask a more direct and crass version: am I the asshole here?
If you aren’t familiar with Am I The Asshole (or AITA as it’s often stylized), it’s a subreddit with a simple concept. Users make a post describing a moral conundrum they’ve found themselves in and ask the titular question. Their fellow users then vote one of four ways: you’re the asshole (YTA), not the asshole (NTA), everyone sucks here (ESH) or no assholes here (NAH). The range of queries is vast: a few recent examples include someone asking if they were wrong to walk out of the room after finding out their brother used a baby name they had called dibs on, or for calling the cops on obnoxiously loud neighbors. Familial conflicts are common - who is right in a situation where a husband tells his wife he doesn’t want to spend Father’s Day with her dad? What about calling out an obnoxiously sexist family member in public?
Though many of the posts are clearly faked, with the original poster describing behavior so outlandishly evil or angelic that it beggars belief (if you really want some satisfying hate reading, try sorting the posts by controversial), the sub has become one of Reddit’s most popular, with almost 17.5 million subscribers and in 2022 was the most visited subreddit.
Daniel Yudkin believes that the sub is not only one of the world’s great places to kill time, but an important lesson in the modern conception of morality.
A social psychologist who is currently a visiting scholar at Wharton, Yudkin found himself, like millions of others, fascinated by the wide range of moral quandaries being posted to AITA. He began to wonder if there was a way to quantify the millions of posts to come to some sort of larger understanding of how people interpret everyday morality. The result is a preprint study that he’s still trying to publish but that has some interesting conclusions about how we decide what is right or wrong.
“I have also wasted a good amount of time on the subreddit, and I also was surprised that somebody let me spend some of my academic and scientific time studying it because it was really a blending of fun and work in a way,” he told The Flatline.
With his collaborators, Yudkin fed a huge amount of AITA threads into a machine learning program designed to examine the answers posted by Reddit users in an attempt to learn what kinds of moral dilemmas people grapple with in their day-to-day lives while also seeing how the types of relationships involved affect how the perceived morality of a person’s actions. By analyzing almost 370,000 posts and 11 million comments, Yudkin and his colleagues were able to identify 29 common common dilemma types that ranged from dealing with fairness (AITA for skipping the line?) to intentional harm (AITA for calling my friend fat after she called me ugly?) to those dealing with social norms (AITA for not wanting to date a woman with tattoos or kids?).
“Our aim was not really aimed at digging into the underlying truth of how human beings kind of parse moral questions,” said Yudkin. “It was more like a descriptive exercise just trying to map the kinds of experiences that people have from day to day where they're grappling with questions of right and wrong, and really kind of trying to create a taxonomy or categorization scheme that can help us better understand what types of things that people are dealing with and how often are they dealing with these things and also, how often are they really the asshole when they perform one behavior or another.”
What they learned shed light on a little studied aspect of morality: how the relationship between people can affect the perceived rightness or wrongness of their actions towards each other, even when that relationship is a distant one. Our perception of morality may not be so much about the right thing to do as it is what do we owe those around us?
“It's really a question not just of answering how do you do the right thing?,” said Yudkin. “It's also answering a question like how can I be a good friend? How can I be a good partner or son or whatever? And so, even though that seems like it's kind of a trivial idea, or like an obvious idea, I think the implications are profound because it signals a shift in the way that we we think about morality from a philosophical perspective and also it can help to remind us about what's important in our own moral lives, that being a good person requires us to identify who we are to different people and to honor the particular set of obligations or commitments that we've made to them, whether explicitly or implicitly.”
Of course, behavior online and in the flesh often bears little resemblance to each other. People who spend significant amounts of time on AITA know that many respondents often urge the original poster to go for the nuclear option. Had a disagreement with your mom about how she treats your kids? Cut her out of your life. Your best friend of 20 years didn’t like the color of your new car? Never speak to her again. Yudkin acknowledged the possibility that how people respond to morality on the Internet may be very different from how they do so in the flesh. To account for this, his team conducted a follow up study consisting of a survey, asking a large group of people about incidents from their own lives where they felt they may have been in the wrong.
“We found that there's a very strong correlation between the two groups,” he said. “That gives us some kind of confidence that despite the fact that there's sometimes people who are trolling or saying things that don't really make sense, those really are mostly more just like noise and the vast majority of the time the kinds of things that people are writing about are real and sort of reflected in everyday life.”
After spending so much time thinking about the now ubiquitous question of whether someone is an asshole (seriously, YouTube and TikTok accounts have begun attracting followers off nothing more than getting an AI voice to read some of the sub’s most popular posts), it’s only natural that Yudkin’s relationship with AITA has changed. Like tens of millions of other people, he still gets a kick out of the posts, but now has a more complicated relationship to the questions being asked.
“It's still very entertaining to me. I do have a little sort of map in my mind of which posts fall in which category. But again, I'm always just sort of impressed by how much these things have to do with the sort of complex interpersonal dynamics that I think people are constantly navigating and that just take up so much of our moral thinking and our time every day. Trying to figure out answers to those questions of our relational obligations, I think, is just the most important thing that we can be focusing on thinking about right now.”