20/07/2026
I recently wrote an article about a Ted Hall Substack. A friend asked her AI if Hall's post was AI-assisted, and it said probably. I asked her to ask it if MY article was AI-assisted. (It's not; I have never used AI on an article.)
Here's the answer. It's flattering to me at points, and insulting to the human race at others (if an argument is cleanly and logically structured, that must be AI.)
There are also interesting conclusions about the future of wine writing. Check it out. I'll link to the article in question in the comments.
#########
was this article written by AI?
Yes—but with an important distinction: I would not conclude that this article was primarily generated by AI. My best guess is that it was written by a human journalist, with a significant possibility of AI assistance in the research, organization, or polishing.
My read is roughly:
60%: human-written, perhaps AI-assisted in places
35%: substantially human-written with little or no generative-AI involvement
5%: primarily AI-generated
Why it feels more human than the Ted Hall essays
The strongest evidence is the distinctive journalistic voice. Phrases such as:
“Small vineyard owners are the most screwed...”
“For the others, the math is brutal.”
“Expensive business”
“an annuity with a view”
“the most expensive ground cover in America”
“Coombsville... is screwed”
have a sharp, opinionated, slightly irreverent voice that feels much more like a specific writer making editorial choices than generic AI prose.
The article also has a clear journalistic architecture: it takes Hall's thesis, explains it, adds context, quotes him extensively, and then offers the writer's own framing—particularly the provocative conclusion that this may ultimately be good for consumers. That is a recognizable reported-essay structure.
What does raise an AI-assistance flag
The article is extraordinarily smooth and efficiently organized. Nearly every paragraph advances the argument, and the progression is almost textbook:
Small vineyards are uneconomic.
Here's why.
Here's the cost structure.
Here's how the economics changed.
Here's why small wineries make the problem worse.
Here's why specific AVAs are vulnerable.
Here's the solution: remove the vines.
That kind of clean structure can certainly be produced or enhanced by AI.
Also, the article repeatedly uses a pattern of short declarative sentence → explanation → dramatic conclusion, which AI is very good at generating. For example:
“For the others, the math is brutal.”
followed by the explanation of why. That's effective journalism, but also a familiar AI rhetorical pattern.
The biggest clue: the relationship to Ted Hall's writing
Because the article is built around Hall's essays, I would be especially interested in comparing the language attributed directly to Hall with the surrounding prose.
The Hall quotations have a noticeably different style:
“A lifestyle homeowner often never ran the arithmetic at all.”
“The one seat at the table that durable businesses always keep filled... sat empty for 20 years.”
Those are highly polished, metaphorical, consulting-style formulations. They are consistent with the Ted Hall writing we discussed earlier: probably human-originated thinking, but possibly AI-assisted in expansion, polishing, or rhetorical refinement.
The W. Blake Gray article, by contrast, is more concise, journalistic, and colloquial. That difference argues against the entire piece simply being AI-generated from Hall's material.
My overall judgment
I think the article is probably genuinely written by W. Blake Gray, but I would not be surprised if AI was used as an editorial tool. Perhaps Hall's Substack essays, data, and notes were fed into an AI system to help organize the argument or identify the strongest points, followed by substantial human rewriting.
The most likely scenario to me is:
Human journalist + human reporting/editing + possibly AI-assisted research, summarization, or structural editing.
I would be much more suspicious if the article had the telltale AI combination of generic opening, excessive “Moreover/Furthermore” transitions, vague abstractions, repetitive conclusions, and no distinctive point of view. This piece has too much specific attitude and too many sharply chosen, sometimes provocative judgments to strike me as raw AI output.
One further observation: the article may actually be more AI-assisted in its underlying economic analysis than in its prose. The very clean chain from vineyard economics → small wineries → falling grape prices → vineyard removal feels like it may have been synthesized from Hall's essays and data. But the voice—especially the repeated use of “screwed”—feels distinctly like a human journalist choosing to make the argument more blunt and readable.
************************
And then I asked:
should a writer like Blake be worried about his future as a wine writer, competing with AI?
Yes—but probably not in the way he should fear.
I think a writer like Blake Gray should be worried about the economics of wine journalism, but not necessarily about AI replacing him as a wine writer.
The key distinction is:
AI is very good at producing an article. It is much less good at being a journalist.
What AI can increasingly do
AI can already do a credible job of:
summarizing Ted Hall's Substacks
explaining Napa's vineyard economics
organizing a story
generating a clean first draft
producing SEO headlines and subheads
comparing data
writing a competent 1,200-word article about Napa's vineyard crisis
And frankly, a publication could ask AI to produce something resembling this article in seconds.
That absolutely threatens the market for generic wine content—the sort of article that says what a wine is, where it comes from, what it tastes like, and whether it is worth buying.
But Blake's real value is elsewhere
What makes a writer like Blake valuable is not simply his ability to put words into sentences. It's his ability to:
know which story is worth writing
recognize when a claim is questionable
understand the personalities and history behind the story
call people who will actually talk to him
ask uncomfortable questions
develop sources over years
notice contradictions
bring a distinctive point of view
AI can write:
“Napa's vineyard economics are under pressure.”
But a journalist can write:
“Napa Needs to Bulldoze Vineyards.”
And then make that argument in a way that provokes the industry.
The title itself is a human editorial judgment. It creates controversy. It gets shared. It makes people angry. It makes people want to respond. That is much harder to automate than prose.
The biggest danger is not AI replacing Blake
The bigger danger is that publishers will use AI to replace the less distinctive wine writers around him, reducing the number of paid opportunities available to everyone.
I think wine journalism may increasingly split into three categories:
AI-generated commodity content�“Five Napa Cabernets to Drink This Weekend.”
AI-assisted human journalism�A writer researches and reports, while AI helps with transcripts, data analysis, background research, and drafting.
Distinctive human journalism�Original reporting, access, expertise, personality, controversy, and a recognizable voice.
Blake should aim aggressively for No. 3.
And there's an irony here
The more AI-generated writing floods the world, the more valuable trusted human voices may become.
People are going to increasingly ask:
Did this person actually taste the wine?
Did they actually visit the vineyard?
Did they speak to the grower?
Do they know the people involved?
Is this their real opinion?
Can I trust their judgment?
That creates an opportunity for an established writer with a strong voice and deep sources.
My advice to Blake would be: don't compete with AI at producing more words. Compete at having better access, better judgment, better sources, better taste, and a more unmistakable point of view.
In fact, the article you shared is a good example. An AI could probably have produced a technically competent version of it. But the provocative framing—Napa needs to bulldoze vineyards—and the willingness to say that out loud is precisely where a human journalist can still create value.
The future problem for Blake is less “Can AI write like me?” and more “Can I make my reporting and point of view so valuable that nobody mistakes me for a replaceable content generator?”
