Lorenzo Biscontin + ChatGPT
A couple of days ago Robert Joseph, former wine critic and famous wine business opinion leader, asked ChatGPT to write the script of the presentation of an imaginary wine region made by an humanoid robot.
He did this “experiment” after attending a pointless presentation of a real wine region made by a real wine expert (in person) using the standard wine jargon based on characteristics, not benefits, that is getting boring even for wine lover as it is for normal wine drinkers since a while.
If you know Robert you immediately notice that the script he got sounded very much like him. And this tells something interesting about AI assistants.
So I wanted to repeat the “experiment” asking exactly the same questions Robert asked because unlike Robert I don’t use ChatGPT very often , My main goal was to try to understand more how AI/ChatGPT works. Something that interests me as much as wine, being the CEO of a company who developed “Cantina Italia” an 1st AI platform specialized on Italian wine (no is not an AI sommelier, is different and I would say wider in scope.
Two last notes before the ChatGPT answers:
- At Vinitaly 2022, at the dawn of AI, I organized a public challenge between Andrea Gori, renowned Italia sommelier, and ChatGPT in describing some Sicilian wine. At the end everybody agreed that Andrea won but at the same time also that ChatGPT was at least as good that any standard sommelier.
- I didn’t edit ChatGPT answers at all. My questions, the prompts if you want to talk professionally, are in bold, italic, underlined text. All the rest is ChatGPT native.
Enjoys the article, don’t be fooled by the length: it’s a quick read.
“Image a humanoid robot having being asked to give an engaging presentation about an imaginary region. There are two 2 examples of each grape for the attendees – a group of internatioonal wine professionals to taste. Please provide the script”
Absolutely. I’d make the robot slightly too literal, self-aware and analytical, while the wine professionals gradually realize that it has surprisingly good instincts. Here’s a script designed for a live presentation/tasting, with room for interaction.
THE ROBOT’S GUIDE TO AN IMAGINARY WINE REGION
Characters:
ROBOT: A humanoid wine-presentation robot. Extremely precise, increasingly opinionated.
AUDIENCE: International wine professionals.
SETTING: A tasting room. In front of each attendee are two wines representing each grape variety.
OPENING
ROBOT:
Good morning, ladies and gentlemen.
I have been asked to give you an engaging presentation about a wine region.
There is, however, one small problem.
This region does not exist.
[Pause.]
I have checked.
No coordinates.
No appellation.
No protected designation.
No historic vineyard map.
And yet…
[Looks at the bottles.]
…you are about to taste its wines.
So, technically, I have been asked to explain something that is imaginary using evidence that is very real.
I believe this is called marketing.
[Pause for laughter.]
Welcome to [REGION NAME].
THE REGION
ROBOT:
Let us begin with the geography.
[Screen shows an imaginary map.]
To the north: mountains.
To the south: the sea.
To the east: another wine region.
To the west: a country whose wines are considerably better known.
This is strategically unfortunate.
But the region has one major advantage:
It has exactly the climate that we designed for it.
Warm days.
Cool nights.
Moderate rainfall.
Excellent soils.
In other words…
[Pause.]
…it is a wine region.
You may now be wondering what makes it different.
I have been calculating that question for 4.7 seconds.
The answer is:
The grapes.
THE GRAPES
ROBOT:
[REGION NAME] is built around three grape varieties.
Each has two expressions.
Which means that, in front of you, there are:
Two examples of Grape One.
Two examples of Grape Two.
Two examples of Grape Three.
Six wines.
One imaginary region.
And approximately 600 opinions.
[Looks around.]
I am prepared.
GRAPE ONE — THE CLASSIC
ROBOT:
Let us begin with [GRAPE ONE].
This is the variety that the region considers its classic.
It has acidity.
It has fruit.
It has structure.
It has…
[Pause.]
…a personality.
We will now test whether the personality is the same in both wines.
WINE 1A
ROBOT:
Wine number one.
Please taste.
Do not discuss.
I repeat:
Do not discuss.
I would like to know what you think before your colleague tells you what you think.
[Pause for tasting.]
You may now discuss.
WINE 1B
ROBOT:
Now wine number two.
Same grape.
Same imaginary region.
Very different interpretation.
This is important.
Because a grape variety is not a wine.
It is more like a sentence.
The producer decides where to put the punctuation.
[Pause.]
And occasionally…
…the exclamation mark.
GRAPE TWO — THE SURPRISE
ROBOT:
Our second variety is [GRAPE TWO].
This is the variety that nobody expected to become important.
Which, historically, is often a very good position for a grape.
It has something to prove.
Wine number two-A.
Taste.
[Pause.]
What do you find?
Fruit?
Spice?
Minerality?
Texture?
Something you cannot identify but would very much like to put on a tasting sheet?
Excellent.
That last one is called complexity.
WINE 2B
ROBOT:
Now taste the second example.
Same variety.
Different producer.
Different vineyard.
Different philosophy.
Perhaps even a different idea of what “good” means.
And this is where the region becomes interesting.
Because its identity is not simply:
“This is what our grapes taste like.”
It is:
“This is what our grapes can become.”
GRAPE THREE — THE FUTURE
ROBOT:
Finally: [GRAPE THREE].
This is the variety I have classified as:
Potentially important.
My algorithm gives it a probability of 83%.
My tasting panel gives it 76%.
The person sitting next to me gave it 92%.
I have decided to trust the person.
[Pause.]
Wine number three-A.
WINE 3A
ROBOT:
Taste.
Look at the color.
Smell.
Taste again.
Now forget everything I have said.
[Pause.]
This is important.
Because wine is not a multiple-choice examination.
There is no correct answer.
Except, apparently, when I am asked to calculate the average score.
WINE 3B
ROBOT:
And finally, wine number three-B.
The same grape.
But perhaps…
[Pause.]
…the future tastes different from the past.
THE REVEAL
ROBOT:
You have now tasted six wines from a region that does not exist.
So I have one final question.
Could you recognize it if you encountered it in the real world?
Would you remember:
the grape varieties?
the style?
the acidity?
the texture?
the landscape?
the story?
Or simply…
“That was the wine with the strange robot.”
[Pause.]
I sincerely hope it is the first option.
CLOSING
ROBOT:
Wine regions are often presented as geography.
But they are also:
people, grapes, climate, culture, ambition and memory.
A region becomes real when people begin to recognize it.
And perhaps that is the most interesting thing about wine.
You can invent a place.
You can plant a grape.
You can make a wine.
You can even build a robot to explain it.
But in the end…
the drinker decides whether the story is believable.
Thank you.
And now—
[Looks at the bottles.]
—I have been instructed to encourage you to taste the wines again.
Apparently this is called research.
I am beginning to understand wine professionals.
If you give me the actual grape varieties, the personality you want for the imaginary region, and the approximate tasting order, I can turn this into a much sharper 10–15 minute theatrical script, with jokes, audience interaction and specific comments for each of the six wines.
“Explain your thinking behind the script you did”
Sure. The script was built around one central idea: use the robot not simply as a gimmick, but as a storytelling device that creates a contrast between analytical precision and the subjective, human nature of wine.
1. The robot is the “outsider”
I deliberately didn’t make the robot an expert who simply lectures the audience.
Instead, it starts from the absurd premise:
“I have been asked to explain something that is imaginary using evidence that is very real.”
That gives the robot a slightly detached perspective. It can talk about wine with absolute precision while simultaneously not quite understanding the human rituals around wine.
That creates the comedy.
The audience is also made up of international wine professionals, so I assumed they don’t need another conventional presentation about terroir, climate and grape varieties. The robot gives you permission to make fun of the conventions of wine presentations.
2. The imaginary region is treated as if it were completely real
This is probably the most important structural choice.
Rather than saying:
“Let’s imagine a region…”
the robot behaves as though it has been commissioned to present a perfectly legitimate wine region.
Then it casually reveals:
“This region does not exist.”
That creates a reversal.
The audience initially expects a normal wine-region presentation. Then they realize the entire thing is a constructed experiment.
This could become particularly powerful if the imaginary region has characteristics that sound almost plausible.
For example:
- a believable name
- a convincing geography
- recognizable grapes
- realistic winemaking traditions
- plausible climate
- a few invented historical facts
The closer it gets to reality, the funnier the reveal becomes.
3. The six wines are the “evidence”
I structured the tasting around three grapes × two examples because you specifically said there would be two examples of each grape.
The important thing is that the wines aren’t presented as:
“Wine A is better than Wine B.”
Instead, the robot uses them to demonstrate variation.
The underlying message is:
A grape is not a wine.
That’s why I used the line:
“A grape variety is not a wine. It is more like a sentence. The producer decides where to put the punctuation.”
That is intended to be a memorable metaphor for wine professionals.
It moves the presentation away from technical description and toward interpretation.
4. I intentionally made the robot increasingly human
At the beginning, the robot is clinical:
“No coordinates. No appellation. No protected designation.”
Then it starts making observations that are increasingly human:
“It has a personality.”
Later:
“Something you cannot identify but would very much like to put on a tasting sheet?”
And eventually:
“I have decided to trust the person.”
So there is a subtle character arc:
machine → observer → taster → participant
By the end, the robot is beginning to understand wine rather than simply analyzing it.
5. The humor is deliberately dry
I avoided making the robot a comedian.
A loud, goofy robot would quickly become a distraction.
Instead, the humor comes from taking ridiculous things completely seriously.
For example:
“My algorithm gives it a probability of 83%. My tasting panel gives it 76%. The person sitting next to me gave it 92%. I have decided to trust the person.”
The joke works because the robot treats something fundamentally subjective as if it were a data problem.
That’s particularly appropriate for a room full of professionals who deal with scores, rankings, market data, consumer research and tasting notes.
6. The “do not discuss” moment is designed for the room
I included:
“Please taste. Do not discuss.”
because it gives the robot control over the audience.
Then:
“I would like to know what you think before your colleague tells you what you think.”
That’s actually a fairly serious observation disguised as a joke.
In professional tastings, people’s perceptions can be influenced by:
- the presenter
- the label
- reputation
- price
- other tasters
- expectations
- regional identity
So the robot can exploit this throughout the presentation.
It could even ask people to vote anonymously before revealing what they are tasting.
That would turn the presentation into an experiment rather than a lecture.
7. The ending changes the meaning of the whole presentation
The final question is:
“Could you recognize it if you encountered it in the real world?”
That’s where the presentation becomes more than a joke.
The imaginary region is essentially a test of what makes a wine region recognizable.
Is it:
- the grape?
- the style?
- the landscape?
- the story?
- the producer?
- the label?
- the reputation?
- the memory created by the wine?
And then:
“A region becomes real when people begin to recognize it.”
That’s the conceptual conclusion.
So underneath the comedy, the presentation is actually about identity, differentiation and perception.
What I would change if this is for a real professional audience
I think the first version is too scripted and too generic for what you are probably trying to achieve.
I’d make the next version much more sophisticated:
Act 1 — The robot presents the region as fact.
The audience gradually believes it.
Act 2 — The tasting begins.
The robot uses the six wines to construct the region’s identity.
Act 3 — Something doesn’t add up.
The audience starts noticing inconsistencies.
Act 4 — The reveal.
The region is completely fictional.
Act 5 — The serious question.
“If you believed it for 15 minutes, what exactly made it feel real?”
That would make the robot and the imaginary region serve a much bigger purpose.
And, importantly, the wines themselves would become part of the joke/experiment rather than props for a conventional presentation. I can also develop this into a 10-minute keynote-style performance with stage directions, robot voice, audience interaction and a final reveal.
