One day before we sent the September print issue of America to press, we relaunched the americamagazine.org website. This column will be one of the first things published directly to the new site.

(Though the coincidence of press day and launch day was not ideal timing, it marks an improvement over the last website relaunch in 2017. That one had been deliberately timed to coincide with the print magazine’s redesign but also coincidentally wound up happening the same week that we moved out of the building housing the magazine offices and the Jesuit community. This time was easier.)

Because we were preparing the website at the same time as we were reviewing a print issue with two excellent essays on artificial intelligence, I found myself reflecting on the nature of what is sometimes called “knowledge work”—the kind of work that A.I. is posed to disrupt or perhaps displace.

In this print issue, Michael O’Connell applies insights from David Foster Wallace to the question of A.I. and the humanities, and Nathan Schneider reflects on the fact that interacting with chatbots depends on the quality of the questions we ask. He notes that these models are trained on the results of humans asking and answering questions, with websites like Stack Overflow, Quora and Reddit providing some of the most valuable sources behind the apparent intelligence of a ChatGPT, Gemini or Claude.

While working on the website, I have spent plenty of time searching the internet for snippets of code to iron out one small issue or another and hints for which box to check in a complicated settings page. I frequently found myself reading Stack Overflow (much of which is dedicated to such technical questions) or landing on a Reddit thread tangentially related to my query. And Google being what it is these days, almost every search I made had an “A.I. overview” at the top of the results.

Many of these overviews were helpful. Two directly produced working code, which felt like magic; in another case, the overview led me into a conversation with a chatbot that mostly “wrote” the function I needed. A few of the A.I. overviews were hallucinatory, which is to say that they made up answers that sent me down rabbit holes and wasted hours of time, but which would have been extraordinarily useful if the technological features that the A.I. had “imagined” had ever been coded by the human engineers who built the website software. In a sense, what the A.I. had hallucinated was a feature request that unfortunately had never been handed off to a human for implementation.

What was profoundly more helpful and reliable was the expert and tireless assistance of the teams behind our website project. These included designers from Goji Labs, who also built our mobile app; engineers at Newspack, our new website platform, who migrated our content from the old site and configured the new one; and of course my colleagues at America Media. Particular thanks are due to Zac Davis, who heads our digital team and led the website redevelopment project, and Jai Sen, who has worked with us for years as a technology consultant.

Maybe one day—though I doubt it—it will be possible to instruct a chatbot or an A.I. agent to “migrate this website for me,” then go to the beach for a few days and come back to find everything working. Even if A.I. companies manage to make such miracles routine, however, behind them will be the accumulated knowledge, the questions and answers, of human beings who knew what they were doing. There is simply no substitute for someone who understands not only what you have said but also what you are trying to do and is sharing that project with you. 

Nor is there any substitute for the deep and textured knowledge of a whole body of work or field of inquiry that comes from long experience and disciplined attention. This sort of expertise can be technical, and many of the engineers we worked with have it in spades, but it can also be humanistic. Moving the whole website, with more than 40,000 individual pieces of content, was a powerful reminder of the decades of writing, editing, learning, reflection and prayer that are embodied therein.

I am sure that those decades of generous effort have been included in the corpus of text used to train the chatbots we talk to these days, in part because ChatGPT can mimic my own prose style fairly well. If that means the chatbots are marginally more reasonable in discussing Catholicism, we should probably count that as a win, even if they do not buy a subscription.

But I repeat that the basis of such knowledge is the chain of human questions asked and answered by people with reason to trust one another. We are honored by the trust our readers and subscribers have placed in us for more than a hundred years as we pursue our mission, in the words of our founding editorial announcement, “to broaden the scope of Catholic journalism and enable it to exert a wholesome influence on public opinion, and thus become a bond of union among Catholics and a factor in civic and social life.” We plan to keep it up for the next century as well.

This article appears in September 2025.

Sam Sawyer, S.J., is the editor in chief of America Media.