(Photo by Robin Worrall/Unsplash)

MY CHILD doesn’t have a smartphone or use social media. Until they turned 13, I thought that meant they were protected from its influence. I was wrong.

I’m watching the culture of social media reach my child secondhand, both as a mother and the wife of a public school teacher. Online vocabulary travels offline, including increasingly precise categories young people use to understand differences in attention, personality, sexuality, gender, politics, and mental health.

Some of this can be enormously valuable. A teenager with undiagnosed autism can discover an explanation for their differences. A gay or transgender teenager can discover that others feel the same way. But social media doesn’t simply provide access to information. It continuously classifies our behavior to predict what to show us next, including content we never asked for.

Last week’s landmark multistate settlement with Meta will limit how much time teenagers can spend on its platforms, but it largely leaves personalized feeds intact, requiring only that Meta give parents and teens the option to turn them off.

Massachusetts should go further by passing S.3175, “An Act Protecting Children from Addictive Social Media Feeds.” The bill would disable certain algorithmically personalized feeds for minors across all social media platforms, while still allowing teenagers to search for content, follow creators and groups, and otherwise tell the platform what they want to see. In other words, it would preserve teenagers’ ability to explore while limiting the platform’s ability to direct that exploration.

Algorithmically personalized feeds can be addictive by design, but there is another reason to think twice about them: They may also influence how teenagers come to understand themselves.

Developmental psychologists have long studied “social essentialism” —our tendency to see categories as revealing deep and meaningful differences between people. Research with children has found that categorical language can make characteristics seem more intrinsic and categories more rigid, as though they represent distinct kinds of people. Put simply, a label can make an attribute feel less like something about you and more like something you are. This is especially consequential during adolescence, when forming a sense of self is a central developmental task.

That should give us pause in an information environment built around classification. Online platforms infer demographic characteristics, interests, and other attributes to decide what content and advertising to put in front of us.

Research on TikTok has described how recommendation systems can turn users’ behavior into increasingly specific profiles and then reflect those inferred identities back to them through the content they see. The algorithm may therefore be putting teenagers into ever-smaller boxes at precisely the age when they should still be forming—and revising—ideas about themselves.

Imagine a teenager wondering whether she has ADHD. She watches several ADHD videos. The algorithm hasn’t learned that she has ADHD; it has learned that she engages with ADHD content, so it supplies more.

Troublingly, fewer than half of symptom claims in the 100 most popular #ADHD TikToks aligned with clinical diagnostic criteria. Beyond the risk of inaccurate self-diagnosis, researchers studying autism have documented how users who encounter diagnostic frameworks online can reinterpret their lives through them.

Obviously, TikTok doesn’t cause ADHD or autism. But algorithms can shape the path of exploration and self-reflection. Repeated exposure can make one characteristic disproportionately present in someone’s information environment, making it seem increasingly central to who they are.

This can also run counter to one form of social progress: making our differences matter less. I grew up believing equality meant gender should tell us less about a person, not more. Girls should be allowed to be masculine without needing to explain themselves, and boys should be allowed to be feminine without needing to explain themselves.

Not every difference needs to define your identity. Algorithms may undermine that progress by increasing the salience of the very characteristics Americans have spent generations fighting over, deepening cultural divisions that cut across shared economic interests.

That should concern both sides of our culture wars. Progressives worried about algorithms pulling teenage boys toward increasingly rigid ideas about masculinity and conservatives concerned about the growing list of identity categories may be seeing different manifestations of the same underlying system. We don’t have to agree about the categories to agree we should be suspicious of the machinery that makes them more salient.

The goal shouldn’t be to prevent teenagers from exploring or changing their identities, but to protect their freedom and intellectual independence as they do so. Teenagers should remain free to search for information, encounter unfamiliar ideas, and decide for themselves what those ideas mean for who they are, without an algorithm constantly mediating that process.

We cannot expect teenagers to protect that freedom on their own. And parents cannot solve this individually. I keep my child off social media, but their peers use it, shaping what they discuss at school and how they understand themselves and one another. That is why the House should pass S.3175.

The bill would put reasonable limits on how platforms can direct minors’ attention while preserving their ability to choose what they explore. And it can do so without requiring teenagers to surrender their privacy. Its age-assurance system must include a method that requires neither government identification nor biometric data and allows users to remain anonymous to the platform.

At precisely the age when young people are still figuring out who they are, an algorithm optimized for engagement should not be shaping the possibilities. Particularly when that system is susceptible to manipulation and capable of fragmenting us along increasingly granular lines.

Elise Rapoza is a New Bedford resident and senior research associate at the MassINC Policy Center.