Over the past several months, lawmakers on both state and federal levels have worked to pass legislation regulating social media algorithms (especially TikTok). Among other things, the debates have revealed (or reminded us) that a lot of folks fail to understand various aspects of the internet, social media, and especially AI-based recommendation algorithms. But that doesn’t stop them from having an opinion on it.
In fact, a few months ago I came across an interesting Facebook post from a pastor of an evangelical church in Texas. He’d been using a popular online Bible website when he noticed a pop-up ad for a camouflage Let’s Go Brandon t-shirt (“Let’s Go Brandon” being code for insulting President Biden). He posted a screenshot of the ad along with a message lamenting the product itself, suggesting that the political sectarianism exemplified by the slogan was why young people were leaving the church.
As the post went semi-viral, people showed up in the comments to work out exactly how the ad had gotten there and what it meant. The entire exchange provides an interesting look into how media shape conceptions of religion and religious groups, and especially the relationship between algorithmic targeting and identity.
Here on RD, several people have written about AI and religion, from the possibility of an internet-connected human superorganism to the way ChatGPT chooses which religious figures deserve respect. In addition to thinking about how AI will reshape communities, beliefs and practices, it’s also interesting to think about how people’s conceptions of and discourses on AI, whether accurate or not, are already changing religious relationships and identities.
After all, it’s not only algorithms and computation, but especially the way people imagine and perceive algorithms and computation, that shape the way we think about ourselves and our communities. They also shape (and reshape) the way people think about “the media” itself, raising concerns about democracy in the digital age.
In the original Facebook post, the evangelical pastor at first expressed dismay at why a Christian website would advertise a political t-shirt, especially one that included such an inflammatory message and euphemism for profanity. Commenters quickly explained that it most likely wasn’t the website’s choice, but rather a personalized, targeted ad based on searches he had performed for similar products in the past—that the ad was a reflection of his own browsing activity. Programmatic advertising uses artificial intelligence and machine learning algorithms to make decisions about which ads to show to specific users.
When he insisted that he’d searched for nothing that would lean rightwing, he bemoaned that “the internet” must think that evangelicalism in America is synonymous with this type of political message. Ads have become a modern day mirror, he said, curated by our demographics.
Can we look into this computational mirror to say “beware” or “behold,” assuming that the algorithm knows us better than we know ourselves? Indeed, “Let’s Go Brandon,” as noted above, is a popular rallying cry for Trump supporters, and White evangelicals did make up a substantial portion of Trump supporters in both 2016 and 2020. The advertising algorithm may well have been effective in targeting the appropriate demographic, and perhaps we could read this as a reliable method of confirmation of the overlap between Trump-style, right wing politics and Christianity.
On the other hand, more than just “evangelicals” (however defined) read the Bible online, and simply because some Trump supporters identify as evangelical or Christian doesn’t mean they do, in fact, read the Bible. Thus others in the comments had different reactions, expressing anger that the company selling the t-shirts was “targeting” Christians. There were discussions of how cookies work, and how perhaps those who created the ad might have included “Bible” in the code, for example.
Far from the algorithm being a perfect mirror or predictor of our interests, for some this incident revealed that the people controlling the code have a lot of power. If one thinks that marketing agents put “Bible” in the code, then they might think of them as predatory, or presumptuous about U.S. Christians and their political affiliations. They might become angry that companies have such power in shaping the public’s perception of Christianity and, indeed, in shaping Christians’ own ideas about their communities’ identity and values. Perhaps the ad would act as confirmation bias, or serve to normalize slogans, tones, and political attitudes like “Let’s Go Brandon” among Christians.
The whole conversation underscores the complex relationships between vendors, websites, consumers, and advertising platforms like Google Ads (as was the case in this example). But this particular example highlights yet another layer to the conversation: those users reacted not only to the algorithmic profiling, but also to their perceptions of advertising, AI, and the online platforms involved. In this example, the pastor initially reacted to the (incorrect) idea that the Bible website had platformed this ad, leading to an initial anger at the people who run the site. After all, not all ads are targeted, and some platforms do pay companies to advertise, which can lead to strong reactions and boycotts of particular brands.
On the other hand, the user might become more alienated from (or, on the flip side, more enthusiastic for) the group or tradition if they think a Let’s Go Brandon t-shirt is an accurate depiction of “their” values. Some commenters jumped to lots of conclusions about “the media” and its “bias,” while others saw “the algorithm” as an all-knowing, accurate thing that revealed the connection between contemporary Christianity and right wing politics. Yet others reacted to the people or tech companies they imagined control the code—the ones who not only decide who sees ads and where, but also shape these types of associations between products and religious groups.
In other words, while the pastor himself was more focused on lamenting that the algorithm had again revealed (in his view) this affinity between evangelicals and right wing politics, the public discourse very quickly moved to identifying problems with “the media” and accusations of “anti-Christian bias,” “pushing ideas,” and the takeover of information by “big tech.”
No matter how the ad got there, whether it was targeted (it was), whether it was based on an estimation of the user’s demographic information or predicted based on browsing activity, what is interesting about the conversation on Facebook was how the group was imagining itself in reaction to whatever they thought had happened, highlighting the ways in which not only the technology itself, but also discourses on technology shape group identities. Conversations like this, even if clarifying for some, can for others lead to a further erosion of trust in “the media” and in information online. What are the implications for a growing distrust in media, information, and technology? The corpus of comments revealed a complex interplay of increased virtual interactions, while at the same time increased skepticism of online activity and information.
In the end, the pastor and the majority of those interacting with his post seemed in agreement that the Let’s Go Brandon t-shirt was not a product they enjoyed or wanted to purchase. The pastor (based on his post, comments, and the interactions with the post) clearly surrounds himself with evangelicals who, at the very least, find a camouflage Let’s Go Brandon t-shirt to be careless in tone, and thus he expressed some surprise—or at least lamented being reminded—that there are evangelicals out there whom this ad could successfully convert. Advertising pun intended.
