Human history can be charted by its modes of communication. The age of bronze, the age of iron. Those are nothing compared to the benchmarks of literacy, from the invention of alphabets to the novel, from the coming of the internet to the smartphone. 

And now we have AI, “artificial intelligence.”

In the early days of newspapers, people worried about the long reach made possible by the printing press. The distance between any one reader and the author of the words being read made accountability an issue. It was the anxiety of print.

A favourite example is the eighteenth-century battle of Jonathan Swift, the author of Gulliver’s Travels, and John Partridge, an astrologer. Mr. Partridge published almanacs that made predictions for the coming year (most often wrong), including the deaths of public figures.

Eventually, Swift had had enough. Using a pseudonym, he published a prediction, in Partridge’s style, that Partridge himself would die on March 29th 1709, “about eleven at Night, of a raging Fever.” 

On March 29th he published a report — from an anonymous source, of course — confirming the death of John Partridge, along with a deathbed confession that all of his astrological predictions had been made up. 

For believability, Swift announcement included a correction. Partridge died at 7:05 in the evening, making his prediction, he acknowledged, “mistaken almost four hours.” This astonishing report was reprinted everywhere. The eighteenth-century equivalent of viral. 

Partridge, of course, put out a publication protesting that he was very much alive, to which Swift replied in print that this “Partridge’s” letter was false. “He dares not assert, he was alive ever since the 29th of March, but that he is now alive and was so on that day: I grant the latter, for he did not die till Night … and whether he be since revived, I leave the World to judge.”

While Swift moved on from this little print tempest, many others (for their own purposes) kept the joke going. Partridge had to deal with being widely thought to be dead until his, ironically, obscure death in 1714 or 1715.

Although now three decades old, the internet is still very much the wild west of media. As it emerged, it seemed an extension of print, and we thought the same rules applied: the standards of truth and integrity created within the institutions of print journalism, science, research, and scholarship. 

The first website I started visiting regularly in the 1990s was called “Bert is Evil.” It specialized in posting photos of historic moments and crudely photoshopping the Sesame Street character into each of them. Bert was at Nazi rallies. He was on the grassy knoll of the Kennedy assassination.

How did we get from “Bert is Evil” to “Pizzagate,” the 2016 episode in which a Washington, DC pizza parlour became a front for Hillary Clinton’s child sex ring? How did that story get taken so seriously by a conspiracy-minded right wing that someone travelled all the way from North Carolina with a gun, which he fired during his “search” of the restaurant?

Yes, we had always had tabloid stories about Hillary Clinton, or Sigourney Weaver, adopting an alien baby, but they were just entertaining. Nobody took those headlines literally.

What happened?

Part of the issue was identified by Marshall McLuhan half a century ago. “The medium is the message.” When a medium becomes familiar, it becomes trustworthy. We are less likely to be sceptical of something we read in black and white, because newspapers worked very hard over the three centuries after Swift to establish their credentials. The internet cannibalized on that trustworthiness. Literally, it ate it. Now the distrust that attends the internet’s viral conspiracy theories threatens the venerable institutions of journalism. 

Last week on CountyFM’s Grapevine, Lynn Pickering interviewed computer science professor Mark Daley about Artificial Intelligence, and what it does to truth and facts. The very familiarity of ChatGPT’s format — convincing, assertive prose — is dangerous, because the rules have changed. AI can sound like a human in its “writing.” 

All the words around AI, the idea that it can be “trained,” that it “reads,” that it “writes,” that it has “intelligence” — all of them are deceptive. All of them encourage us to humanize an artificial medium that piggybacks on human thought. 

“People can be easily duped with good, old-fashioned ‘information’,” notes Professor Daley. 

“We have to retrain ourselves. Any media we consume, whether it’s the written word, a photograph, a video, we have to ask, ‘what source did that come from; is this a reasonable thing that I’m seeing?’”

An extra perk is that AI can “hallucinate.” That’s an effect of the way AI is designed. It is tested on real people at the end. “It turns out that most humans prefer to be lied to than to be told, ‘I don’t know the answer’. They’re mad five minutes later when they realize they’ve been lied to, but in the moment, they give the AI the reward signal for the lie. So it learns to lie.”

We need a new information literacy: “Understand that everything you get told by an AI you need to interrogate and make sure it’s true. You have to look for reliable sources.”

Professor Daley then offered direct praise for his interviewer and mainstream media in general: “this is precisely why journalism from respected institutions becomes so critical in the next two decades, because it used to be I could Google something and I could click on the top link and it’s probably going to be true. Now we’re in a world where disinformation and misinformation are easy to create and so I want to look to someone like you,” meaning Ms. Pickering and her radio station, “or to a publication like The New York Times where its reputation is everything, so I trust it’s verified its facts and is telling me the truth. 

“Know your sources. If someone tells you something you don’t believe, find the source you trust to corroborate that.”

What to do when AI obscures its sources, as Google’s Gemini seems to be doing, is a subject for another day. 

CountyFM’s archive contains the complete Grapevine interview with Mark Daley. (Search by show title and choose Monday 21 July.)