Reflections on Artificial Intelligence, Truth, and What It Means to Be Human.
An Unexpected Conversation
between guest Blogger Alex Smith and AI.
Recently I asked Google's AI a simple literary question.
"I ran across this interesting phrase in a book regarding an artist: 'He was working in a false idiom.' Can you explain what that meant?"
The response seemed confident and plausible. AI explained that an artistic "idiom" refers to a style or mode of expression and that working in a "false idiom" meant creating in a manner that was artificial, derivative, or unsuited to the artist's genuine gifts.
So far, so good.
The problem came when the AI confidently stated that the phrase originated in The Voyage by Charles Morgan (1940).
It did not.
The phrase was actually from C.S. Lewis's The Great Divorce (1945).
When challenged, the AI first attempted to construct an elaborate explanation linking Lewis and Morgan before finally admitting that it had simply made a mistake and had "hallucinatory-mapped" a quotation from one book into another.
That admission led to a much deeper and more interesting discussion.
What Does "Working in a False Idiom" Mean?
In literary criticism, an "idiom" is a characteristic style or mode of expression.
To suggest that someone is "working in a false idiom" is to imply that the person's artistic voice is somehow inauthentic. The artist may be copying trends, imitating admired predecessors, or adopting a style that does not genuinely fit either the subject matter or the artist's own nature.
The expression appears in The Great Divorce when the Tousle-Headed Poet remarks:
"Even before we came here, I'd had some doubts about a man like Cyril Blellow. I always thought he was working in a false idiom."
Some readers have looked for possible real-world literary influences behind Lewis's characterisation of Cyril Blellow. While such suggestions can be intriguing, identifying a specific individual as Lewis's model remains speculative and is not required for understanding the point Lewis is making.
The irony, of course, is thoroughly Lewisian: the speaker is quick to detect falseness in others while remaining blind to his own spiritual condition.
Reference: C.S. Lewis, The Great Divorce (Geoffrey Bles, 1945), Chapter 1.
Can a Creation Surpass Its Creator?
One of the AI's responses included this observation:
"An expression cannot be more perfect than its source."
At first glance this sounds reasonable.
But is it actually true?
Human beings routinely create things that outperform them in specific functions.
A calculator performs arithmetic more accurately than most people.
A crane lifts more weight than any human being can.
A computer can store and retrieve information far beyond the unaided capacity of human memory.
In specialised tasks, creations frequently exceed their creators.
This prompted an interesting question:
If AI can process enormous amounts of information and apply formal logic at speeds impossible for human beings, why does it make such obvious errors?
Why AI Hallucinates
The AI gave an illuminating answer.
Unlike a traditional database, a large language model does not simply retrieve stored facts from a filing cabinet. Instead, it generates responses based upon learned patterns and probabilities within language.
The result is that related ideas can become blended together.
In this case:
C.S. Lewis and Charles Morgan were both twentieth-century British authors.
The Great Divorce and The Voyage belonged to roughly the same literary era.
The phrase sounded entirely at home in that intellectual and cultural setting.
The AI therefore produced a highly plausible answer which happened to be wrong.
As the discussion continued, it offered further literary connections and interpretations. Some sounded convincing, but closer examination revealed that they rested more upon association than evidence. The experience became a useful reminder that plausibility is not the same thing as truth. Even an elegant explanation still requires verification.
Ironically, it did the very thing that many human beings do: it remembered incorrectly while sounding completely confident.
Logic, Discernment, and the Human Spirit
The discussion eventually moved beyond literature and into philosophy and theology.
I suggested that future AI systems could undoubtedly become better at verification, cross-checking quotations and evaluating logical relationships between propositions.
George Boole's formal logic, truth tables, deductive reasoning, inductive reasoning, and abductive reasoning all provide tools that are, at least in principle, open to computational analysis.
Yet there may be another category entirely.
I argued that human beings possess:
A body.
A soul (psyche), including mind, personality, memory, imagination, and emotions.
A spirit (pneuma), through which we may relate to God.
The AI readily discussed the first two categories. Whether the third category can ever be replicated by machinery is a different question altogether.
The Candle of the Lord
One verse came repeatedly to mind:
"The spirit of man is the candle of the LORD, searching all the inward parts of the belly."
— Proverbs 20:27 (KJV)
The picture is remarkable.
God knows everything already. The "searching" is not for His benefit but for ours. Through the human spirit, God illuminates hidden motives, subconscious fears, and deeply buried attitudes which we ourselves often prefer not to acknowledge.
The New Testament likewise declares:
"God is a Spirit: and they that worship him must worship him in spirit and in truth."
— John 4:24 (KJV)
If this spiritual dimension is real, then it is not merely a computational process. It is not simply data, memory or logic operating at higher levels of complexity.
Blade Runner, Philip K. Dick, and Artificial Humanity
The conversation naturally turned toward science fiction.
Ridley Scott's Blade Runner (1982) was adapted from Philip K. Dick's 1968 novel Do Androids Dream of Electric Sheep?
The novel explores questions that remain surprisingly relevant today:
If emotions can be artificially generated, are they still real?
If memories can be implanted, what constitutes personal identity?
If a machine can imitate every human behaviour, what remains uniquely human?
Dick's fictional "Penfield Mood Organ" allows people to dial in desired emotional states. Artificial animals become status symbols in a world where real animals are scarce.
Throughout the novel the reader is continually forced to ask:
What is real, and what is merely a convincing imitation?
The same question now confronts us in the age of artificial intelligence.
Lewis and Dick: Approaching the Same Question from Opposite Directions
What fascinated me most was that C.S. Lewis and Philip K. Dick seem to approach similar territory from opposite directions.
Lewis depicts human beings who gradually become less human through pride, self-deception, and spiritual blindness.
Dick imagines artificial beings striving to become more human.
Some readers detect echoes of contemporary literary debates and personalities within Lewis's portrayals. Whether or not any individual author was specifically in view, Lewis's larger concern appears to be the danger of self-deception and the loss of contact with reality.
Lewis asks:
What happens when persons lose their connection to ultimate reality?
Dick asks:
What would remain if humanity could be artificially reproduced?
Both end up pointing toward the question that technology cannot fully answer:
What does it mean to be truly human?
Conclusion
My original question concerned a simple literary phrase.
Instead, it became an extended conversation about memory, truth, logic, consciousness, artificial intelligence, and the nature of humanity itself.
The most amusing part is that the entire discussion began because the AI got the quotation wrong.
In fact, the initial attribution to Charles Morgan proved to be one of the very errors under discussion: a plausible literary connection presented with unwarranted confidence.
Perhaps there is a lesson there.
Artificial intelligence can be astonishingly informative, impressively articulate, and occasionally insightful. Yet it remains a tool rather than an oracle. It can assist inquiry, but it cannot replace discernment.
A convincing explanation is not necessarily a true one. Facts still require verification, evidence still matters, and wisdom remains something more than information.
Sometimes it takes a human reader to notice that the machine has placed a quotation into the wrong book.
And perhaps, for now at least, that is a comforting thought.
This is an edited version of the original article which can be viewed on Facebook at
Lux Venit Ab Alto - Light comes from above
'My question to A.I. Google' by Alex Smith.
References
Lewis, C.S. The Great Divorce. Geoffrey Bles, London, 1945.
Dick, Philip K. Do Androids Dream of Electric Sheep? Doubleday, New York, 1968.
Boole, George. An Investigation of the Laws of Thought. Walton and Maberly, London, 1854.
The Holy Bible (King James Version):
Proverbs 20:27
John 4:24
1 Thessalonians 5:23
Comments