Skip to content

The Darkness of AI

Robert Rackley
Robert Rackley
2 min read
The Darkness of AI

Casey Shutt considers an article on AI by Paul Kingsnorth for Mere Orthodoxy. Kingsnorth sees demonic forces at play within technological advancement in general and AI in specific. Shutt expands upon the concerns expressed by Kingsnorth in his own piece. He hones in on the sense of real foreboding that plagues some who work with the technology.

The Godfather of AI, Geoffrey Hinton, seems genuinely haunted by AI’s mysterious power and the astonishing speed at which it is advancing. Hinton said in a 60 Minutes interview that humanity does not know what it’s doing with AI, and he fears that we’ll get it wrong, something we can’t afford to do. Scott Pelley followed up, asking why. Hinton replied simply: “Because they might take over.”
And Hinton’s not alone. Technology reporters Casey Newton and Kevin Roose describe what they call “AI vertigo,” that is, the dizzying possibilities that could flow from AI technology and the unease it produces in its creators. In Newton’s reporting on the topic, he has found that those working with AI often have AI nightmares. Even Sam Altman, chief executive at OpenAI and as optimistic as they come regarding the technology, admitted feeling “very strange extreme [AI] vertigo” at different moments, especially surrounding the launch of ChatGPT-3.

Perhaps the most disturbing element of the piece is when Shutt introduces a woman called "Loab," a creation of AI who seems to be cobbled together from the fragments of nightmares interrupted by the blessed relief of waking hours. I had to stop reading the story of Loab only part of the way through, so unsettling was the tale and the accompanying images that were generated by AI.

However, not content to stop at scaring us with Loab, Shutt brings the devil himself into the picture. Pointing out that the devil is known for his dishonesty, Shutt points to the fact that most of the concerns about AI center around deceit.

It is striking that when it comes to most of our AI fears, deception is the common denominator. In academic settings, concerns abound as to how the technology might be used by students to deceive their teachers into thinking AI generated work was student generated. Similar worries can be found in creative enterprises like music, visual arts, and writing. And let’s not forget that one of Sydney’s “dark fantasies” is to spread misinformation. Whether it’s misinformation, deepfakes, AI-generated work presented as one’s own, a faux romance with AI, deception is the common thread. The fingerprints of the “Father of Lies” seem to be all over the technology.

In tying the supernatural to novel technology, Kingsnorth and Shutt are playing with ancient fears. Of course, it isn't safe to assume that the notion of evil has simply become outdated. It makes sense to view evil as a shapeshifter, adapting to the new ways in which it can infiltrate our lives more easily.

FaithTech

Robert Rackley

Orthodox Christian, aspiring minimalist, inveterate notetaker and paper airplane mechanic.


Related Posts

Members Public

The Christian Dissonance

I’ve known for a long time how Donald Trump’s hateful rhetoric can affect people. One of the first incidences of the impact of his speech was when, at the beginning of his last successful presidential campaign, two men beat a homeless Mexican man they thought was an immigrant

Members Public

The Return of the Discman

I don’t really feel like a gadget guy anymore. I don’t get hot and bothered by the newest origami laptops and I skip past the thousands of posts about the latest iPhone model in my feed reader. But man, I love music and have to admit to a

The Return of the Discman
Members Public

The Blogging Learning Curve

My favorite blogging platform, Blot, seems to be defunct. There are major issues with template editing. When I try to edit a template on my iPad, the device heats up, gives me a browser error and then reloads the page. Numerous users have reported that the Google Drive integration is

The Blogging Learning Curve