Athens

John Lennox's Advice on Writing: AI, God, and Why Machines Cannot Author

- Moritz Wallawitsch

John Lennox is an Oxford mathematician, philosopher of science, and Christian apologist. He is the author of 2084: Artificial Intelligence and the Future of Humanity, Can Science Explain Everything?, Cosmic Chemistry, and Determined to Believe?. He has debated Richard Dawkins, Christopher Hitchens, and Peter Singer on the existence of God. He holds three doctorates. He is currently writing his autobiography.

His writing advice comes from David Perell's How I Write podcast, his Mind Matters interviews, and lectures on AI and theology.

Words Come from Minds

Lennox opens with a claim that sounds theological but is actually an argument about authorship. "In all our human experience, words come from minds." The word EXIT above a door is four letters long, and no one doubts a mind produced it. The human genome is 3.4 billion letters long, and it functions precisely as a word with meaning. To attribute mind to four letters but not to 3.4 billion is, in his words, "nonsense."

This is not a detour into apologetics. It is the foundation of how Lennox thinks about writing. If words come from minds, then the act of writing is inseparable from the act of thinking. A machine that produces text without a mind behind it is producing something that looks like writing but is not writing. The artificial in artificial intelligence is real.

The Machine Does Not Think

Lennox uses AI the way most people use spell-check. He finds ChatGPT "quite useful for collecting ideas" because "it will bring to the screen knowledge that I don't have." But he follows every use with verification because the machine "may be hallucinating and inventing that knowledge because it wants to please me."

The distinction he insists on: AI simulates intelligence. It does not possess it. "It doesn't think. It's simply a machine. It's sheer computing power, predictive computing power." He gets tired of telling people this. The machines are not conscious, nor are they likely to be, "for a very simple reason. Nobody knows what consciousness is."

This matters for writers because it reframes the anxiety around AI-generated text. If a machine cannot think, it cannot author. It can produce sentences that resemble authored work. But authorship requires consciousness, intention, a mind choosing this word over that one for reasons the author may not fully understand. Lennox quotes Rosalind Picard: all AI systems are "no more alive than Microsoft Word."

A pastor who watches a late-night film on Saturday and gets a sermon from ChatGPT in thirty seconds before bed will produce something technically accurate. "There is no spiritual power in a machine." The doctrine may be correct. The sentences may parse. But something essential is missing, and every reader will sense it.

How Could This Be Misunderstood?

When Lennox sits down to write, he asks himself two questions. The first: "How can this be understood?" The second, far more useful: "How could this be misunderstood?"

He gives an example. You want to write about God as father. You do not realize that in your audience is a young woman whose concept of father is someone who comes home drunk and abuses her mother. Her concept of father will not map onto yours. If you do not anticipate the misunderstanding, your writing fails not because it is wrong but because it is incomplete.

This is a universal principle dressed in theological clothing. Any writer addressing any audience faces the same problem. The reader does not share your context. Your words will be filtered through their experience, not yours. The question "how could this be misunderstood?" forces you to write from the reader's position, not your own.

Structure and Thought Flow

Lennox spent fifty years studying with David Gooding, a classicist and world authority on the Septuagint. Gooding taught him two things about ancient literature that changed how he approached his own writing: structure and thought flow.

Ancient writers did not label chapters 1, 2, 3. "They were more sophisticated and therefore much more interesting." They divided their writing with repeated phrases. In Matthew's Gospel, the phrase "and it came to pass when Jesus had finished these sayings" appears four or five times. These are division markers. Between them, you look for repetition of ideas and the flow of thought from one section to the next.

Gooding's principle: "Most of the failure to understand what scripture means is not taking enough time to see what it says." The meaning question comes second. The observation question comes first. What words are repeated? What patterns emerge? Why is this section here and not there?

Writers outside the biblical tradition can use this identically. Before asking what your essay means, ask what it says. Trace the repeated ideas. Map the structure. The meaning will surface from the observation, not the other way around.

Don't Be So Proud

"Don't be so proud as not to allow people to read what you've written." Lennox sends his autobiography drafts to what he calls "the very best editor" at his publishing house. The process has been "very painful but very useful."

What has the editor taught him? "Simple things. The danger of being episodical." If you are a person who has done a lot of speaking, you default to listing events. "I went to Leipzig and spoke on X. I went to Berlin and spoke on Y. It becomes episodical and there's no depth. There's only length."

The fix is specificity. Not "I spoke at Austin and then Denver and then Portland," but the one thing that happened at one of those stops that has depth and texture. The editor is "ruthless with that kind of stuff" and tells him to cut the rest or put it in an appendix.

Explanation Has Many Levels

Lennox keeps returning to a distinction between explanation levels. A car engine can be explained by physics. It can also be explained by Henry Ford. "Science no more conflicts with God as an explanation for the universe than Henry Ford conflicts with physics and automobile engineering as an explanation for the automobile."

Even Newton, he notes, did not claim the law of gravity explained gravity. Newton uttered the phrase hypotheses non fingo - "I do not make hypotheses." He could give a formula to calculate gravity's effects. He could not say what gravity is. "No one even now knows what gravity is."

For writers, this is a warning against premature closure. When you think you have explained something, you have probably explained it at one level. The reader who says "I get it" may have gotten one layer. The writer who says "I've explained it" may have delivered one frame. The best writing acknowledges that explanation is never complete. It opens doors rather than closing them.

The Feynman Principle

Lennox cites Feynman: "Always bend over backwards to understand and criticize your own work because the easiest person to fool is yourself." He applies this directly to his writing process. Before publishing, he pressure-tests his arguments by imagining the strongest objections.

This connects to the misunderstanding question. How could this be misunderstood is the reader-facing version. The Feynman principle is the self-facing version. Together, they form a two-sided check: have I fooled myself into thinking this is clear? And even if it is clear to me, will it be clear to someone who does not share my assumptions?

Read Good Literature

"If ever I think I've attained anything in writing, I just read a Lewis book and that brings me right down to size." C.S. Lewis taught Lennox that metaphors stand for reality. "My heart is broken" does not refer to the literal pump. "The car was flying down the road" does not mean the car had wings. These are metaphors for real things, not literal things.

Simple, obvious, and yet Lennox finds that most people have never articulated it. They confuse literal with real and dismiss metaphorical language as imprecise or fantastical. In fact, metaphorical language is often the most precise tool available for conveying experience that literal language cannot reach.

Key Takeaways

  • Words come from minds. Writing without a thinking mind behind it is not authorship.
  • AI can collect and arrange ideas. It cannot think. Use it as a research tool, not a writing tool.
  • Ask "how could this be misunderstood?" more often than "how can this be understood?"
  • Study structure before meaning. Observe what the text says before asking what it means.
  • Avoid the episodical trap. Depth over length. One vivid scene beats ten listed events.
  • Explanation has levels. Do not close the door when you have only opened one.
  • Submit your work to ruthless editors. The pain is the point.

Lennox's insistence that authorship requires a mind - not a processor - is the strongest case for why writers who worry about AI are asking the wrong question. The machine cannot replace what it cannot do. What it can do is help you see your own prose more clearly, the way Lennox's editor helped him see his. Athens is an AI writing tool built on that principle: the mind is yours, the edits are visible, and every change is a suggestion you accept or reject.

This post draws from Lennox's appearance on How I Write and his Mind Matters interviews on AI and 2084. For another thinker skeptical of AI replacing human thought, see Ezra Klein's writing advice.