Apply to my task
arXiv:2602.20408 ★ 91 Feb 23, 2026

Famous-Founder Personas Lose to Ordinary People: CoT + Ordinary Personas as the Way Out of the Sameness Trap

Key result

The paradox: famous-founder personas help less than ordinary people. In the study, personas of 99 real fitness entrepreneurs beat the plain model but fell short of random ordinary personas on every diversity measure. The CoT + ordinary personas method yielded idea sets that, in the fitness-product study, beat the human sample on distinct idea combinations (240 vs 196) and roughly matched it on categories. The trick: you don't need celebrities — you need "a retired politician who has campaigned for library funding." An ordinary persona works as a sampling cue that points the model to a less typical corner of its knowledge.

01

TL;DR

An LLM compresses all human knowledge into a single distribution — and with every new request it gravitates toward the center. Ask a model ten times to come up with product ideas and you'll get ten similar answers clustered around "typical" solutions. The authors argue this follows from how LLMs are trained, not from one particular model (they tested mainly GPT-4o, with Gemini and GPT replications). Two mechanisms work against diversity: fixation within a single session and knowledge aggregation across sessions.

Fixation is when early ideas suck in the later ones. The model generates text step by step: each next token depends on the previous ones. If the first idea is "a gamified app," the next four will most likely circle around gamification too. On top of that, the authors argue (as theory, not tested) that RLHF (reinforcement learning from human feedback) pushes the model further toward "familiar" answers by training it on the averaged preferences of annotators.

Two targeted prompts treat each mechanism separately: Chain-of-Thought (CoT) breaks fixation within a session — the model first sketches titles, then forcibly diversifies them, then expands them. Ordinary personas (not a famous founder, but "a retired politician who has campaigned for library funding") pull the model into different corners of its knowledge across sessions — though on their own they increase fixation within a session. Together, in the paper's fitness-product study, they matched or beat the human sample on most diversity measures.

Want this for your own case? Apply this study to your task: AI adapts its method to what you are working on and writes the prompt.

02

Method Blueprint

Two levels that work independently — and together give the maximum:

LEVEL 1: CoT — against fixation (one request)

  STEP 1: Generate {N} short idea titles (don't expand them)
  STEP 2: Revise the titles — make each one as different
          from the others as possible, replace repeated themes
  STEP 3: Expand each title into a full idea

LEVEL 2: Personas — against aggregation (several requests)

  REQUEST 1: [Persona A — specific, ordinary] + the CoT above → ideas
  REQUEST 2: [Persona B — different, ordinary] + the CoT above → ideas
  REQUEST 3: [Persona C — different, ordinary] + the CoT above → ideas
  ...
  OUTCOME: merge all the idea sets → maximum coverage

⚠️ Each level works in a regular chat. CoT is a single request.
   Personas are separate requests (or one with several roles in a row).
03

Example in Practice

The task, personas and prompt below are our illustration; the paper tested fitness product ideas with one-line personas.

Task: Come up with ideas for Substack newsletters in the personal finance niche. Without the technique, the model churns out the same thing every time: "investing tips," "stock breakdowns," "financial literacy." Boring and predictable.

Prompt (CoT + persona):

You are Megan, 34, a kindergarten teacher from Toledo, Ohio.
You love vegetable gardening and Nordic walking. You got into personal finance
two years ago after your divorce — you learned everything from scratch.

Your task: come up with ideas for personal finance newsletters on Substack.

STEP 1. First, sketch 8 short concept titles (3–7 words each).
        Don't expand them — titles only.

STEP 2. Reread the list. Find repeated themes and angles.
        Replace similar titles with fundamentally different ones —
        so that the 8 ideas cover the widest possible range of the topic.
        Show the updated list.

STEP 3. Expand each idea: 2–3 sentences — what the newsletter is about,
        who the audience is, how it stands out from the rest.

Then repeat the request with a different persona — for example, "Mike, 52, an Army veteran from Tucson who retired early" or "Katie, 26, a freelance designer from Chicago who works for overseas clients."

Expected result:

The model will show all three stages explicitly: a rough list of titles → a revised list explaining what it changed and why → fully developed concepts. Each persona pulls toward its own corner: the teacher gives ideas through the lens of "an ordinary person with no finance background," the veteran through discipline and early retirement, the freelancer through irregular income and currency risk. Merge the three sets and you get a wider palette than a single request usually gives.

04

Why It Works

The fixation problem. An LLM generates text step by step — each token depends on the previous ones. That means the first idea literally "weighs" more when generating the second, the second when generating the third. Start with "gamification" and the next ideas will gravitate toward that theme. Plus, the authors argue, RLHF taught the model to produce "approvable" answers — and approvable tends to mean familiar and predictable (their explanation, not tested).

Why CoT breaks fixation. Step 1 (titles only) keeps early ideas from "growing" and taking over the context. Fewer tokens → a weaker anchor. Step 2 (an explicit instruction to diversify) is a direct command to break the "typical" pattern. In the paper CoT clearly reduced fixation in the model, while for people the same procedure slightly reduced diversity; the authors call the idea that LLMs simply follow such instructions better "speculative."

Why a famous founder is worse than an ordinary person. The paper did not test "Steve Jobs" itself. It compared 99 real, well-known fitness entrepreneurs ("creative" personas) with 99 ordinary people drawn at random from a large persona dataset. Both beat the plain model, but the ordinary personas gave more diverse ideas on every measure. The authors treat a persona as a sampling cue: it points the model to one region of its knowledge. A likely reason for the gap (our reading): successful founders in one industry point to similar regions, while random ordinary people are spread across many walks of life.

Control levers: - Number of CoT steps → our suggestion, untested: cut to 2 (titles + straight to expansion) for simple tasks - Persona specificity → the paper's personas were one-line descriptions (e.g., "a retired politician who has successfully campaigned for library funding"); adding a name, age and city, as in our example, is our addition and untested - Number of personas → in the paper each separate session got its own persona (99 sessions); how few personas you can get away with was not tested - Explicitness of the step 2 instruction → you can add "list which themes are already used and what's missing" — the model will make the diversity visible

05

Prompt Template

You are {name}, {age}, a {profession} from {city}.
{1-2 specific details: hobbies, life experience, context}.

Task: come up with ideas for {task topic}.

STEP 1. Sketch {N} short concept titles (3–7 words).
        Titles only — no explanations.

STEP 2. Reread the list. Find repeated themes and angles.
        Replace similar ones with fundamentally different ones — so the {N} ideas
        cover the widest possible range.
        Show the updated list.

STEP 3. Expand each idea: {format — e.g., 2-3 sentences /
        bullets / a table}.

What to fill in: - {name}, {age}, {profession}, {city} — specific and ordinary; not "Elon Musk," but "Tom, an accountant from Dayton" - {1-2 details} — hobbies, life experience, something specific - {N} — 6–10 for good coverage - {task topic} — as specific as possible: "ideas for newsletter posts about remote work," not just "content ideas" - {format} — whatever output format is convenient for you


🚀 Quick start — paste into the chat:

Here's a template for generating diverse ideas with CoT + a persona.
Adapt it to my task: {your task}.
Ask me questions to fill in all the fields.

[paste the template above]

The LLM will ask about the task topic, the desired format and the number of ideas — because without them it doesn't know what kind of persona to create or how many steps to run.

06

Limitations

⚠️ A method for ideation, not facts: CoT + personas help where you need broad coverage of options. For factual questions ("how much does it cost to register an LLC?") the technique is pointless.

⚠️ A persona anchors but doesn't guarantee: if the personas are too similar ("a manager from Chicago" and "a marketer from Chicago"), diversity will be minimal. You need contrasting biographies — different ages, professions, life contexts.

⚠️ Raising the temperature doesn't help: it's tempting to crank up randomness in the settings. But according to the study, raising temperature from 1.0 to 1.5 gave a small gain in diversity, and at 2.0 outputs were often incoherent or garbled.

⚠️ CoT doesn't work for people the way it does for LLMs: if you're brainstorming with a team, this technique may not help humans. In the paper, people using the same CoT procedure ended up with slightly less diverse ideas, not more.

⚠️ The maximum effect takes several requests: a single CoT request improves diversity within a session. The real breakthrough is several requests with different personas. It takes extra time, but the result is worth it.

07

How They Studied It

The Columbia Business School team ran a "fair" experiment — unlike most earlier work, where LLMs and humans got different instructions. Here the conditions were equalized: the same task (ideas for fitness products), ten ideas per participant. Two studies, with about 100 human participants each and GPT-4o as the main model.

The main methodological contribution is a new way to measure idea diversity. Standard methods compare word similarity (embeddings). Instead, the researchers used an LLM for semantic categorization: ideas were sorted into conceptual categories (industry context, psychological need, product form) rather than by lexical similarity. The authors argue this captures meaning better than wording (our example: "a meditation app" and "a mindfulness program" are lexically different but semantically the same thing).

To measure fixation they looked at the rate at which new categories accumulate: if the first three ideas come from new categories and the next seven repeat old ones, there's fixation. The LLM showed the same curve as humans: diversity piles up quickly at first and then fades. CoT flattened that curve — the model kept adding new categories for longer.

The most unexpected result: famous-founder personas are worse than ordinary ones. Intuition says the opposite — a creative icon should produce unconventional ideas. In the paper, personas of 99 real fitness entrepreneurs did beat the plain model, but they "still fall short of ordinary personas" on every diversity measure. There was a catch with ordinary personas too: on their own they increased fixation within a session, because the persona stays in context and keeps pulling ideas the same way. That's why CoT is needed on top.

The CoT + ordinary personas combination beat the human sample on the number of distinct idea combinations (240 vs 196, about 22% more), while matching it roughly on the number of categories (27 vs 28). The authors say that, to their knowledge, no earlier paper had pushed LLM diversity above human levels. In replications with other models the combination was again the most diverse LLM setup, but beat humans on some measures, not all.

08

Original from the Paper (optional)

The authors describe the CoT method they tested as follows (adapted from De Freitas et al. 2025, which builds on Meincke et al. 2024b). This is a summary, not a verbatim quote:

Chain-of-Thought prompting for ideation:

Step 1: Generate a set of short idea titles (summaries),
        without expanding them into full ideas.

Step 2: Explicitly revise the titles to ensure they are
        different from each other — bolder, more distinct,
        covering different angles.

Step 3: Expand the revised titles into full ideas.

Key mechanism:
- Short titles prevent early elaboration → weaker token anchoring
- Explicit "make different" instruction → breaks typicality bias from RLHF

Context: In the paper's version the model wrote 10 short titles, made them "bolder and more different," and returned 10 one-sentence fitness product ideas as JSON. The persona version opened with "You are acting as this persona: {persona}" and added steps to consider what the persona would value and how they would approach each idea. The link to RLHF in the "key mechanism" lines is the authors' explanation, not something they tested.

09

Adaptations and Extrapolations

💡 Adaptation: Several Personas in One Request

If you don't want to run three separate requests, ask the model to play several roles in sequence within one conversation:

We're going to run three rounds of ideation for {task}.
In each round you take on a new role and generate ideas using the 3 steps below.

ROUND 1: You are {Persona A}. [CoT steps 1-3]
ROUND 2: You are {Persona B}. [CoT steps 1-3]
ROUND 3: You are {Persona C}. [CoT steps 1-3]

After three rounds: pick the 5 most dissimilar ideas across all rounds.

Untested in the paper: in one shared context earlier rounds can anchor later ones (the paper found a persona kept in context increases fixation), so expect less diversity than from separate requests — but you win on convenience.


🔧 Technique: Add an "Antagonist" to CoT

In step 2 you can add an explicit list of what's already been covered:

STEP 2. List all the themes and angles from step 1.
        Now find 3 themes that are NOT on this list.
        Replace 3 similar titles with ideas from these new themes.

This makes the diversification process visible and mandatory — the model literally maps what it has already used and hunts for blank spots.


🔧 Technique: The Outsider Persona

Our idea, not tested in the paper: try someone from a different industry looking at your task for the first time:

You are Linda, 47, an agronomist from Fresno, California.
You just found out what Substack newsletters are, and you're thinking about
how to apply what you know about farming
to the topic of {task}.

The farther the outsider is from the topic, the more unexpected the analogies are likely to be. An agronomist in the personal finance niche will offer ideas about "crop rotation" for capital, diversification as polyculture, the "yield" of investments — metaphors a "finance expert" would never reach.

10

Resources

A Cognitive Framework for Idea Diversity in Large Language Models (arXiv v2; v1 was titled "Examining and Addressing Barriers to Diversity in LLM-Generated Ideas") Yuting Deng, Melanie Brucks, Olivier Toubia Columbia Business School

The work builds on: - Meincke et al. 2024b — CoT for ideation - De Freitas et al. 2025 — source of the CoT prompt and the creative-persona approach - Doshi & Hauser 2024 — an embedding-based measure of collective diversity - Wei et al. 2022 — Chain-of-Thought prompting (the original)

Research digest

Key result

The paradox: famous-founder personas help less than ordinary people. In the study, personas of 99 real fitness entrepreneurs beat the plain model but fell short of random ordinary personas on every diversity measure. The CoT + ordinary personas method yielded idea sets that, in the fitness-product study, beat the human sample on distinct idea combinations (240 vs 196) and roughly matched it on categories. The trick: you don't need celebrities — you need "a retired politician who has campaigned for library funding." An ordinary persona works as a sampling cue that points the model to a less typical corner of its knowledge.

Principle

Two mechanisms kill diversity — and each one is treated separately. The first is fixation within a single request: the first idea "infects" the next ones, because the model writes token by token and each depends on the previous ones. CoT breaks this by force: first titles only (they're short — a weak anchor), then an explicit command to replace the similar ones, then the expansion. The second mechanism is aggregation across requests: different requests still slide toward the same center of the distribution. Different personas with contrasting biographies anchor each request in its own corner. Together — the most diverse setup in the paper. Separately — CoT alone cuts fixation, while personas alone widen coverage but increase fixation, because the persona stays in context.

Why it works

The authors' explanation (theory, not tested): the LLM is optimized toward the averaged preferences of raters through RLHF (reinforcement learning from human feedback), so it gravitates to the center. In the paper, famous-founder personas helped less than ordinary ones, likely because successful founders in one industry point to similar regions. An ordinary person with a specific biography is an anchor in an atypical place. The model looks at the task from a corner it never visits on its own. Every new persona = a new angle = new ideas. Cranking up the "temperature" (randomness) in the settings doesn't help: 1.0 to 1.5 gave a small gain; at 2.0 outputs were often garbled.

When to apply

Ideation → specifically brainstorming content, business concepts, product formats, course topics — especially when you need broad coverage without repeats and there's no budget for a live team. NOT suitable for factual questions — registration costs, dates, laws: personas change nothing there; the model either knows the fact or it doesn't.

Mini recipe

1. Assemble 3–5 contrasting personas: different ages, professions, cities, life contexts. Not famous figures but ordinary people — the paper used one-line descriptions like "a retired politician who has successfully campaigned for library funding." Adding name, age and city ("Gary, 48, a maintenance engineer from Akron who raises rabbits") is our style, untested. 2. Run a separate request for each persona with a three-step prompt: — step 1: titles only, N of them, no explanations — step 2: find repeated themes, replace similar ones with fundamentally different ones — step 3: expand each title into a full idea 3. Pool all the sets together — that's your final pool. Remove duplicates by hand or ask the model to group them and drop repeats. 4. Want more coverage? Increase the number of personas, not the model's temperature. The paper used one persona per session across 99 sessions; how many you need for your task is untested.

Examples

[BAD]: `Come up with 10 ideas for a personal finance podcast. Think like a successful investor.` [GOOD]: `You are Rachel, 41, an EMT from Omaha. Two kids, you work night shifts. You got into personal finance three years ago — learned everything from scratch after taking on a mortgage. Task: come up with ideas for a personal finance podcast. STEP 1. Sketch 8 short concept titles (3–7 words). Titles only. STEP 2. Reread the list. Find repeated angles. Replace similar ones with fundamentally different ones — so the 8 ideas cover different sides of the topic. Show the updated list. STEP 3. Expand each idea into 2–3 sentences: what the show is about, who the audience is, how it stands out.` Then repeat with a different persona — for example, "Mike, 54, an Army veteran from Tucson who retired early" or "Katie, 26, a freelance designer who works for overseas clients." Three personas should give you three noticeably different sets of ideas (our illustration).

In plain words

AI models work like reality averagers: they blend all of human experience into a single distribution and tend to hand back the most "typical" solution. The authors argue this comes from how they are trained, not from a random glitch. When you ask an LLM to toss out some ideas, it doesn't mull over options in its head — it just takes the path of least resistance toward the center of its knowledge distribution. What comes out is polished mainstream, where ten different answers look like identical twins.

It's like going to karaoke with someone who knows thousands of songs but sings "Sweet Caroline" every single time. The voice is there, the repertoire is huge, but they're stuck in the rut of whatever lands best with the crowd. The model behaves the same way: it's afraid to take risks and serves up safe platitudes, because, the authors argue, that's how RLHF shaped it — by optimizing for the averaged preferences of raters.

The core problem here is the fixation effect: the model literally gets "stuck" on its own words. Because text is generated step by step, the very first words of an idea become an anchor for everything that follows. If the model starts with advice on saving money on coffee, the rest of the output will circle around everyday thrift. It's intellectual inertia: the first decent thought keeps pulling the later ones toward itself — the paper found people show the same fixation.

The researchers tested this on ideas for fitness products; applying it to other creative work, from scripts to naming, is our extrapolation. Ask for ideas for a money newsletter and you'll get dreary "investing for beginners," because that's the statistical center of the topic. To jolt the model out of its slumber, the paper used two levers: different ordinary personas across sessions plus a "titles first, then make them different" step within each session, and make the AI look at the edges of the distribution instead of its comfortable middle.

Bottom line: use an LLM "head-on" and you'll get lowest-common-denominator content. To squeeze something genuinely fresh out of the model, you have to remember that by default it drifts toward the typical. SEO for brains doesn't work — you need to break the inertia of generation and push the model past its first decent phrase, or you'll just keep cranking out endless copies of "the typical opinion from the internet".

AI lab assistant

Apply this study to your task

Describe what you're working on. AI takes the method and the measured results from this paper and adapts them to your case, or writes a ready prompt built on them.

Based on Famous-Founder Personas Lose to Ordinary People: CoT + Ordinary Personas as the Way Out of the Sameness Trap arXiv:2602.20408

Sign in to apply this study New accounts get 50 free credits.