AI: A Shortcut to ?

Written By Jonathan Krogh

‍Dear Skeptical Scholars:

A few weeks ago, I wrote a musing which referenced the Omaha Gospel Tabernacle, the congregation of my childhood. Over the decades they have morphed from a series of evangelistic tent-revival meetings into a more respectable and permanently housed congregation rebranded as Christ Community Church. I was looking for reference to the first radio broadcast produced by founding pastor R.R. Brown, when radio was the hot new communication tool. I knew that in the 1970s they had received an award for being the longest consecutive radio broadcast from the same station (WOW). Armed with this information, I typed my query into AI-equipped Google. From my prompt I was informed that the Omaha Gospel Tabernacle was founded in Zion, Illinois, as an early African American congregation in north Omaha with connections to the African Methodist Episcopal denomination. I also learned that Christ Community Church returned to its roots by restoring the name Gospel Tabernacle in the late 1990s. Thank you, AI, none of this is true.

I was reminded of these phantom facts while preparing for today’s Musing. I wanted to introduce you to Aetheria, a remarkable woman from 4th century Spain who embarked on a three-year journey to Jerusalem. Copied fragments of her travel diary were discovered in the 1880s in a Tuscan monastery. The translation from the early Latin revealed in great detail the earliest descriptions of Holy Week celebrations at the sacred sites. After reading a few articles and skimming translations of her Pilgrimage, I began to find phantom information which seems to have worked its way from AI to Wikipedia and beyond. A few articles claimed that the document had been found in a folio known as the Codex Aretinus, found in a monastery library in Tuscany. Except at some point the documents seemed to be conflated with the Codex Argenteus, a 6th century illuminated copy of a 4th century translation of scripture into the Gothic language. Spell check dutifully changes Aretinus into Argenteus. At that point, I gave up. Instead, I decided to offer this musing about the joys of AI in academic research.

Scholars of ancient manuscripts have long noticed how the history of hand-copied documents can be traced by following the occurrence of errors. A simple transposition of letters would be repeated in subsequent copies of that edition. It is possible to follow the mistakes back to their presumed source by tracing the mistake. Later scribes would sometimes attempt to correct the error based on their ability to spell correctly or the use of comparative documents; but not every ‘correction’ resulted in a return to the earlier word. Sometimes whole sentences would be reworked so that the ‘mistake’ made more sense. The fact is no one is working from an original manuscript for any verse in scripture. To some degree the entire Bible has been crowd-sourced throughout the centuries. The same seems to be happening with AI, but at a vastly accelerated pace as each new mistake is posted and requoted as ‘source material’ for the next iteration of large-language analysis.

Consider taking this multiplication of error rate into the technology of modern warfare and ask, “What could possibly go wrong?” It’s one thing to mislabel the source of a 4th century diary for a pastor’s musing about radio preachers; it’s something else entirely when setting targets for cruise missiles. Is that a school or a military warehouse? Let’s see what AI thinks.

Thinking slower thinking may save us all, I remain,

With Love,
Jonathan Krogh
Your Pastor

Jonathan Krogh
Next

Tuning In - Tuning Out?