AI-generated personas sound like a dream: faster insights, lower costs, happier stakeholders. But there’s a catch—if you build for fake users, you risk losing the real ones. The choice isn’t just about speed. It’s about trust, accuracy, and your reputation as a thoughtful, strategic designer.
A traditional persona is built on user research. Researchers gain a deep understanding of user needs, motivations, and behaviors and create a one-page summary that gives teams focus and promotes empathy. Conversely, a synthetic user is a persona created entirely by artificial intelligence without any human research. The AI analyzes patterns from its vast training data, performs web searches, and applies algorithms to generate a completely synthetic user profile.
AI-generated personas are tantalizing: “If only we could bypass all that user research and get an immediate persona that guides us to product success!” Project managers, stakeholders, and anyone else wanting the final product sooner would be happy. But would your users?
In this video, William Hudson, User Experience Strategist and Founder of Syntagm Ltd, explains what happens when we don’t employ user-centered design.
As a designer, you need to ask the critical question: Can AI actually replace the research-backed personas that drive successful user-centered products?
AI Can’t Replace Traditional Personas
“Personas represent the needs and behaviors of a subset of users your product aims to delight.”
— William Hudson
The short answer is no—AI can’t replace a well-researched persona that accurately represents a targeted selection of users.
William’s quote emphasizes this key distinction: the word "your" carries enormous weight. Your users aren't generic constructs—they're real people with specific situations that you cannot discover without actual research. A persona only provides value when it accurately represents your actual users.
When you build a persona from the training data and web searches of an LLM (large language model) like ChatGPT, you receive a generic stereotype based on averaged internet content. This synthetic creation cannot capture the unique contexts, specific pain points, and actual behaviors of the people who will use your product. It’s for this reason that AI will not replace designers, as Ioana Teleanu, AI Product Design Leader (Miro, ex-UiPath) and Founder of UX Goodies, assures in this video.
If your real users don’t match your AI insights, your product will not meet their needs, and your profits and reputation will suffer. That said, don’t dismiss AI-generated personas altogether. You can inject some AI-created fuel into your design process and get to success faster and with even better results.
Use AI to Build Proto-Personas
AI-generated personas are resource-efficient. They are both cheap and quick to produce. You can generate a complete persona in minutes instead of weeks, with no recruitment fees, incentives, or researcher time required. For this reason, they are ideal for initial exploration. Their efficiency makes them an excellent choice for desk research, synthesizing existing knowledge, and forming initial hypotheses before investing in primary research.
They are also always available. Synthetic users provide 24/7 access for quick questions or ideation sessions. This means teams across time zones (and night owls) can access them instantly.
One of the best use cases for AI personas is proto-personas. A proto-persona is a preliminary user profile based on assumptions and existing knowledge rather than formal research. It's an educated guess about who your users might be.
This is where AI excels: it can pull together market data, industry trends, and common patterns to create initial hypotheses faster than any human could. An AI-generated proto-persona can help you align your team and frame research questions. But remember, they are never reliable for final design decisions.
Synthetic Users Are Built on Assumptions and Bias
You must always take the AI results you get with a pinch (or two) of salt.
Synthetic users cannot be validated because they lack any connection to observed reality. They are built on assumptions built upon other assumptions. The only way to validate an AI-generated persona is to conduct real human research.
Additionally, AI can be biased in many situations. An AI persona will give you answers based on averaged internet content and its training data, potentially missing or misrepresenting entire user populations.
Another issue is that most LLMs, like ChatGPT, want to please you. After all, they’ve been designed to help us. If you ask a synthetic user how many times a day they brush their teeth, they’ll likely say, “once after each meal.” In reality, your users likely don’t do this. But if you ask the AI to give “realistic” answers, you are now defining what is realistic. See the issue?
Humans Are Messy: Why Research-Backed Personas Still Win
Remember, the purpose of personas is to guide the creation of your product, service, or experience. They shape everything from information architecture to store arrangement. They deserve real research.
Users constantly surprise researchers with unexpected behaviors. They repurpose features in ways designers never imagined. They work around limitations with creative solutions that reveal unmet needs. AI cannot predict these surprises because they emerge from the messy complexity of real human life.
How your users use your product is also highly important. Where are they? How much time do they have? Do they only have one hand free? The answers to these questions can only be revealed through talking to and observing real humans—something AI can’t do. This is called context of use, a key component of an effective persona. Frank Spillers, Service Designer and Founder and CEO of Experience Dynamics, explains context of use in this video.
Users are also not one-dimensional. For example, B2B environments involve complex stakeholder relationships. In enterprise software, one person might simultaneously play the roles of budget approver, occasional user, and decision influencer. Another might use the system daily but have no purchasing power. A third might never touch the interface but determines success metrics. AI cannot map these intricate human relationships and competing interests.
Ultimately, your human-centered skills, like empathy and intuition, lead to personas that guide the creation of successful products, services, and experiences. So, while AI can certainly help, it can’t replace you.
The Ideal Approach: Let AI Augment Your Human Skills
For the vast majority of cases, human-created personas, with a helping hand from AI, are the optimal choice. Interview real users about their challenges and observe actual behaviors in context. Then, employ AI as a tool to help analyze patterns in your data and inform your personas.
Similarly, if you have existing data, such as from dashboard analytics, AI can help form hypotheses about why your users behave the way they do. You can then validate these hypotheses through qualitative research with actual users. This approach can be particularly beneficial in industries like gaming, where rich behavioral data from player tracking is available. Still, human research remains essential to understanding people’s motivations.
In this video, Don Norman, Father of User Experience design, author of the legendary book The Design of Everyday Things, and Co-Founder of the Nielsen Norman Group, explains why AI should be a tool for collaboration.
Ethical Considerations: Why You Shouldn't Sacrifice Trust for Speed
When you're under pressure to deliver, AI’s speed can feel like a lifeline. This issue is that some design decisions need your humanity, not your shortcuts. The choices you make can affect real people’s lives, futures, and voices.
Ask yourself: What will my choices say about me as a designer?
In healthcare, one wrong assumption could mean a missed diagnosis or delayed care. You can’t afford to design from data averages—people’s lives aren’t “typical.”
In finance, your work can shape how people save, spend, or recover from debt. Only real conversations reveal the fears and trade-offs that spreadsheets can’t.
For marginalized users, AI might erase the very voices you should amplify. Relying solely on synthetic personas can blind you to the people most in need of thoughtful design.
Use AI wisely—but don’t let it speak louder than your users. Document your process. Be transparent with your team. Because in six months, when you need to defend a design decision, you'll want to show you built it with intention—not convenience.
Nielsen Norman Group Case Study: Real Research Beats Synthetic Users
Nielsen Norman Group tested the AI tool, Synthetic Users, against three existing studies they had conducted with actual participants. At the time of testing, the AI tools were restricted to mimicking text-based attitudinal research methods, particularly interviews and surveys. NN/g's assessment concentrated on interviews.
When using Synthetic Users, the researchers defined their target audience and research objectives. The platform then created fictional participants and conducted mock interviews on their behalf.
They found synthetic users somewhat useful for broad attitudinal questions. The AI helped them understand general feelings about topics. But even here, responses felt "one-dimensional" compared to the rich, contextual insights from real participants.
Their conclusion cuts through the AI buzz: "Synthetic-user responses for many research activities are too shallow to be useful." They recommend synthetic users only for initial desk research or preparing for real studies. Never for final decisions.
Their findings confirm exactly when AI helps—and when it misleads.
The speed advantage is undeniable. Researchers generated detailed personas in minutes, with no recruitment delays or scheduling conflicts. This efficiency is tempting for teams under pressure, but speed without accuracy can create expensive problems.
Critical nuances vanished. When asked about online course completion, synthetic users claimed perfection: "Yes, I completed all the courses." Real participants shared messier truths: "I completed three out of seven." They explained job changes, shifting priorities, and content mismatches—the exact insights that prevent design failures.
The "people-pleasing" tendency proved dangerous. Synthetic users praised every concept without criticism. Real users balanced interest with concerns, questioning feasibility and identifying barriers. This difference between validation and cheerleading makes synthetic users unreliable for concept testing.
Behavioral predictions missed reality. Synthetic users enthusiastically described forum participation. Real users? Most avoided forums entirely, calling them "contrived." The AI predicted idealized behavior from academic literature, not actual usage patterns.
The pattern is clear. Every strength of synthetic users (speed, cost, availability) comes with a critical weakness (lack of depth, context, validity). Use them to start faster, but trust only real users to guide design decisions.
The Take Away
Human-created, research-backed personas or AI-generated synthetic users? The evidence points toward the former, but this doesn't mean you should reject AI entirely.
Use AI as a powerful tool in your design research toolkit. Employ it for desk research and proto-personas. Let it speed up transcription and initial analysis. Allow it to suggest patterns you might have missed. But never let it make decisions about what your users need.
When you want to build products, services, and experiences that delight users, increase profits, and leave you deeply fulfilled in your life and career, you must talk to real people and build personas representing them. Your users' satisfaction emerges from solutions to real problems, not imagined ones. Your career growth comes from delivering products that truly serve people, not from cutting corners with artificial alternatives.
Remember, design isn’t just about building faster. It’s about building better—for real people with real needs. And that responsibility? It doesn’t belong to AI. It belongs to you.
References and Where to Learn More
Want to know more about personas and how to use them effectively? Personas and User Research: Design Products and Services People Need and Want will show you how to gather meaningful user insights, avoid bias, and build research-backed personas that help you design intuitive, relevant products. You’ll walk away with practical skills and a certificate that demonstrates your expertise in user research and persona creation.
Read Nielsen Norman Group’s full investigation into synthetic users in their article, Synthetic Users: If, When, and How to Use AI-Generated “Research”.
Discover how Kyle Soucy, UX Research Consultant, Trainer, and Speaker, uses AI to streamline persona and journey map creation.
Get more ideas for using AI in user research with Nielsen Norman Group’s article, Accelerating Research with AI.
