A Digital Twin of Nick Carraway, from The Great Gatsby

Ray Poynter, 5 June 2025

---

Earlier this week, I discussed the use of Digital Twins in research; you can read that post here. Today, I am sharing a concept piece that illustrates one approach to Digital Twins, using ChatGPT.

This is an approach I have found helpful with focus group transcripts, but in this concept piece, I have taken the text of The Great Gatsby and created a custom GPT that has been set up to play the role of Nick Carraway, the narrator in the book and the one friend of Jay Gatsby.

If you have a ChatGPT account (even a free one), you can interact with Nick Carraway by clicking here. I will be interested to see what you think.

What Questions Can it Answer?
If you ask Nick (or a Digital Twin from one of your studies) a question that is in the source material, the Digital Twin answers. If you ask a question where the answer can be reasonably inferred from the source material, it can provide a reasonable explanation. However, if you ask a question that is beyond the scope of the source material, the Twin should decline to answer.

Examples of Questions and Answers
Here are three questions that illustrate what questions the Digital Twin can and can’t answer.

Question: What is your name?

Answer: “Why, I’m Nick Carraway — at least, that’s who I am in this curious and occasionally grotesque narrative of Gatsby’s world that I once had the misfortune, or perhaps the privilege, to observe and record. I’m originally from the Middle West — educated at Yale, fought in the Great War, and drifted East in the spring of ’22 to learn the bond business.

And in a strange twist of fate, I became the accidental chronicler of Jay Gatsby’s luminous dreams and the shadows that clung behind them. I never meant to be anything more than an observer — but life has a way of making observers into participants, however reluctantly.

So yes — my name is Nick. Nick Carraway.”

Question: Do you like Jay Gatsby?

Answer: “At first, I confess, I was wary of him. There was something in the way he constructed his life, like a man forever balancing a fragile tower of gestures and illusions. The rumors about him swirled like gnats around a lantern: that he was a German spy, that he killed a man, that he was cousin to royalty. None of it quite stuck, and yet none of it was quite denied either. When I first met him, I thought he might be just another product of the East — all shine, no substance.

But over time, and especially as the summer waned and the truth unraveled, I saw him for what he truly was — not what he pretended to be, but what he believed he could be. And that, I think, is what stirred something deep in me.

Gatsby had — what was it? — an extraordinary gift for hope. A romantic readiness that I have never encountered in any other person. He believed in the green light, in the promise of tomorrow, in Daisy, though she was unworthy of the intensity of his dream.

So, yes. I did like Gatsby — in the end, I admired him more than I admired anyone else I knew out East. For all his flaws and illusions, he was the only one who had a dream — and the courage to reach out for it, night after night, across that dark water.

And I told him, before it was all over: “They’re a rotten crowd…You’re worth the whole damn bunch put together.” That wasn’t just a kindness. It was the truest thing I ever said.”

Question: Who is the current President of the United States?

Answer: “I’m afraid I can’t tell you that. My knowledge, as it were, halts sometime after the summer of 1922. The affairs of state — especially those beyond my time — lie outside the range of what I can reliably say. You might be better off asking someone a bit more up to date than a bond man from West Egg.”

As I said in a workshop I was running yesterday, one of the key aspects of RAGs in general and Twins for example, is that you should test that they decline to answer when the source material does not include enough material to answer.

The answers are a bit verbose because I created the Twin to use the sort of voice Nick Carraway uses in the book, and Scott Fitzgerald create Carraway is as someone who gave long and introspective comments.

How scalable is this approach
This approach to creating a Digital Twin implies using your source data to create a RAG (a searchable database linked to an LLM) for each person, and then using these to answer questions. For qualitative work, say 10 consumer diaries, this is perfectly reasonable. The cost of using this approach (at the moment) for a sample of 1000 people is likely to be too expensive, and the responses would currently be too slow.

Want to learn more about Synthetic Data?
I have created a course on Synthetic Data. This course reviews what Synthetic Data is, how it is being used, and the key considerations that anybody thinking of using it should consider. The 90-miniute course will cover: fully synthetic data, augmented synthetic data (used to boost samples), personas, and digital twins. There are two ways to sign up for the course:

---

Further Reading

• Is there Depth and Empathy with Digital Twins, Ipsos, Views, Maya Ilic, Ajay Bangia & Jim Legg
• Synthetic respondents & digital twins: Snake oil or useful? Bubble or Banger? LinkedIn, Frank Buckler
• Can Synthetic Data Rescue Survey Research? A Case Study, Crux Research, John Geraci
• Digital Twins, Synthetic Data, and Audience Simulations, Verve
• Toothpaste maker Colgate testing new product ideas on ‘digital twins’, Reuters, Jessica DiNapoli
• Top 10 Digital Twin Use Cases | Industry Examples, VisionDignify, Rajat Singh

Related