AI in Market Research: Opportunities, Risks, and Why Real People Still Matter

Key Takeaways

  • AI in market research is transforming how research teams design studies, analyse data, and deliver insights more efficiently.
  • While AI tools can support speed and scalability, human moderators still play a critical role in uncovering emotion, nuance, and authentic participant insight.
  • Data quality remains a major concern, with synthetic respondents and AI-generated answers creating new challenges for the research industry.
  • Angelfish Fieldwork combines responsible AI adoption with rigorous human-led participant validation and qualitative recruitment expertise.
  • The future of AI in market research lies in balancing technology with human judgement to protect research quality, ethics, and authenticity.

The rise of AI in market research and why human insight still matters

AI is reshaping market research quickly. What started as a novelty is now part of everyday research design. Insight teams are using AI to write screeners, analyse transcripts, moderate interviews, and synthesise findings at a scale and speed that simply wasn’t possible a few years ago.

And yet, something risks being lost.

Qualitative market research has always been built on a simple but powerful idea: that real people, sharing real experiences in their own words, hold insights that no algorithm can replicate. The texture of a hesitation. The contradiction between what someone says and what their face does. The unexpected answer that reframes an entire project. These things cannot be automated.

At Angelfish Fieldwork, we’ve been connecting real people with qualitative research since 2011. Our passion has always been the human side of market research and keeping it alive. This guide sets out what AI in market research can and can’t do, what the risks are for data quality and ethics, and why human insight remains irreplaceable, especially in qualitative work.

Whether you’re a client-side insights manager, a UX researcher, a project manager navigating new tools, or an agency trying to stay ahead of the curve, this page gives a clear, honest and practical view of AI in market research.

What is AI in market research?

AI in market research is about using AI tools to support different parts of the research process, from study design through to analysis and reporting. In practice, this spans a wide range of applications.

Common ways AI is used in market research

1. Screener writing

AI tools can generate draft screeners quickly from a project brief, helping research teams save time at the design stage. They can also support the creation of those research briefs, giving teams a stronger starting point. Read more in our blog on AI-written market research screeners.

2. AI-moderated interviews

Automated systems that conduct one-to-one qualitative conversations with participants, usually following a structured discussion guide. We explore this in depth in our blog on AI-moderated interviews.

3. Thematic analysis

AI tools can scan transcripts and open-ended responses to identify patterns, themes, and sentiment quickly.

4. Report generation and synthesis

Drafting summaries, pulling key quotes, and helping structure findings from large volumes of qualitative data.

5. Translation and transcription

Making multilingual research faster and more accessible.

6. Synthetic respondents

Perhaps the most controversial application: AI-generated personas used in place of real participants. We examine the risks of this approach in detail in our blog on synthetic respondents in market research.

According to a recent industry survey, 98% of market research professionals now use AI tools in their work, with 72% using them daily or more frequently, reflecting how embedded AI has become in everyday research workflows.

The question is no longer whether AI belongs in research, but where it helps, where it falls short, and what it can’t replace.

What can AI do well in market research?

To have an honest conversation about AI in market research, it’s important to recognise what it does well. Used in the right way, AI tools offer real benefits to research teams.

Speed and scale

AI can process and analyse large volumes of data far faster than a human researcher, in many cases. For quantitative studies with thousands of responses, or qualitative projects generating hours of transcripts, AI-supported analysis can surface patterns in minutes rather than days. For time-pressured insight teams, this is significant.

Consistency in repetitive tasks

Tasks like transcription, translation, coding, and thematic tagging benefit from AI’s consistency. A human analyst working across 50 transcripts may apply slightly different criteria as time goes on. AI applies the same rules to the first and the last.

Efficiency in study design

Writing first drafts of screeners, discussion guides, or survey instruments is time-consuming. AI tools can produce useful starting points quickly, freeing researchers to focus on refinement and judgement rather than starting from a blank page.

Accessibility and reach

AI-moderated interviews can, in many cases, be run at scale, across geographies, and at lower cost than traditional human-led fieldwork. For exploratory or directional research, this can open up access to a wider range of participants than might otherwise be possible.

Data pattern recognition

AI excels at spotting correlations, clusters, and anomalies in large datasets, including open-ended text. Used alongside human interpretation, this can help researchers spot threads worth following that might otherwise be missed.

The key word is 'alongside'. AI's strengths are most valuable when they support human researchers, not replace them.

Where AI falls short: the case for real people

The limitations of AI in qualitative market research aren’t just small gaps that technology will close in the next software update. They’re more fundamental and tied to what qualitative research is designed to do.

AI cannot read the room

Qualitative research is not just about what people say. It’s about how they say it. A skilled human moderator notices when a participant’s tone shifts. They pick up on body language, hesitation, and contradiction.

They sense when someone is holding back, and they know how to gently draw that out. However sophisticated it is, AI lacks this kind of contextual awareness. It follows a script, and it can’t adapt in the moment in the way a human can.

Empathy is not a feature

When a research topic is sensitive; health, money, relationships, identity; participants need to feel heard and safe. That requires a real human presence.

AI can’t offer genuine empathy. It can simulate warmth, but participants can sense the difference, and when they do, they disengage. The quality of insight suffers.

The unexpected answer is often the most valuable

Some of the most powerful moments in qualitative research are the ones that weren’t planned for. A participant takes the conversation somewhere entirely different. An off-script comment reframes the whole research question.

These moments require a human moderator who can recognise their significance and follow the thread. AI works from a guide. It doesn’t really get surprised, and it doesn’t follow curiosity in the same way a human does.

Nuance, irony, and subtext

Language is complex. People say one thing and mean another. They use irony. They make jokes that reveal anxieties. They hedge. They contradict themselves in ways that are meaningful.

People also pick up on these nuances through instinctive, emotional signals – gut feelings, tone, and moments that spark recognition or laughter. AI natural language processing has improved dramatically, but it still struggles with the kind of layered, contextual meaning that real conversations are built on.

Participant trust and engagement

The quality of qualitative data depends on participants being genuinely engaged and willing to share honestly. Many participants, especially those discussing personal or emotionally sensitive topics, respond very differently to an AI interviewer than to a human one. This can affect not just the depth of response, but its authenticity.

In qualitative research, the human connection between moderator and participant is where the insight really comes from.

Related reading: AI-Moderated Interviews: Where They Add Value – and Where Humans Matter 

AI vs human researchers: a false debate?

Framing AI versus human researchers is often the wrong question. A more useful way to think about it is where each adds the most value.

AI is a tool. Like all tools, it works well when applied to the right job, and poorly when applied to the wrong one. The research professionals who are getting the most out of AI are not those who have replaced human judgement, but those who have worked out where to deploy each effectively.

Where a hybrid approach works well

Where human researchers remain essential

The researchers and agencies who will get the most out of AI are the ones who can hold both in mind – using it where it genuinely helps, while protecting the human side of research that makes insight meaningful.

Related reading: The Potential of AI in Market Research 

Related reading: AI in 2025: What We Learned 

AI in qualitative research methodologies

AI is being used in different ways across qualitative research methods. Here’s a quick look at where it tends to work well, and where more caution is needed.

Screener design

AI tools are well-suited to generating draft screeners, particularly for more standard studies. The risk lies in using AI-generated screeners without expert review. Poorly constructed screeners can let the wrong participants through and undermine the study. Read our guide: AI-Written Market Research Screeners. 

In-depth interviews (IDIs)

AI-moderated IDIs are one of the fastest-growing applications, offering more scale and speed, often at a lower cost. Platforms can run large numbers of conversations simultaneously across different markets.  

However, for research that requires depth, sensitivity, or strategic insight, human moderation usually delivers richer, more reliable findings. See our detailed exploration in: AI-Moderated Interviews: Where They Add Value – and Where Humans Matter. 

Focus groups

AI moderation is less commonly used in group settings, where human dynamics like conversation, agreement, challenge and humour are central to the research value.  

AI-supported analysis of focus group transcripts is becoming more common and can be useful for identifying themes across multiple groups. 

Online communities (MROCs)

AI tools are being used to support moderation in large online communities, helping to summarise discussions, flag emerging themes, and prompt follow-up questions. Human oversight is still important to make sure the community keeps its authentic character. 

UX research

AI is well-established in supporting usability testing, helping to process session recordings, identify friction points and group feedback. In UX contexts, human interpretation is still essential to understand what those findings actually mean. 

AI and data quality: a growing risk

One of the biggest challenges AI has introduced to market research is a growing data quality challenge. And it’s something the industry is still working through. 

Synthetic respondents: the hidden threat

AI tools can now generate convincing open-ended responses, mimic natural language patterns, and produce survey answers that look, on the surface, entirely plausible.  

These synthetic respondents, such as bots and AI-generated personas, are increasingly capable of passing standard quality checks, entering research studies, and adding responses that may appear convincing, but aren’t based on real people or real experiences. 

This isn’t a theoretical concern. It’s already happening. And its implications for qualitative research, where every participant’s voice matters, are serious. Explore this further in our blog: Synthetic Respondents in Market Research. 

The scale of the problem

According to Wave 1 of the Global Data Quality Initiative’s benchmarking study, a meaningful proportion of survey data is routinely removed during fieldwork due to fraud, duplication, and poor-quality responses — typically in the region of 10–15%, with additional post-survey cleaning increasing that figure further.  

Why Data Quality Risks Are Growing 

Earlier GDQ findings suggest the true scale of problematic data may be significantly higher when inattentive and disengaged responses are taken into account. Together, these findings highlight a critical point: even with active quality controls in place, a substantial proportion of research data cannot be taken at face value. 

UK benchmarking has also shown that risk is not evenly distributed. Agency-led projects and supplier-sourced data can carry different levels of exposure, with some supplier routes seeing significantly higher rates of fraud or quality issues. The message is clear: data quality cannot be assumed. 

As Debrah Harding, Managing Director of MRS, notes, while the industry should embrace the opportunities AI brings, maintaining ethics, quality and integrity must remain non-negotiable if research is to retain its impact. 

It’s not just fraud

Alongside deliberate fraud, AI adoption has amplified other data quality risks. Participant fatigue, inattentive responding, and poorly designed AI-generated screeners that fail to properly filter audiences all contribute to datasets that may appear complete but lack the depth and authenticity research depends on. 

Why qualitative research is especially vulnerable

In quantitative research, poor responses can sometimes be statistically absorbed. In qualitative work, where projects may involve 8, 12, or 20 participants, a single fraudulent or disengaged respondent can noticeably distort findings.  

Increasingly, this also includes participants using AI tools to help shape or generate responses – not always fraudulently, but in ways that can still impact authenticity and depth. The stakes are higher, and the human safeguards matter more. 

How Angelfish approaches data quality

At Angelfish Fieldwork, we have signed the Global Data Quality Excellence Pledge. Every participant we recruit is validated in-house by a trained, RAS-accredited team member.  

We speak to people directly, confirm eligibility, and assess articulation and engagement before any fieldwork begins. Read more about our approach on our market research data quality page. 

The ethics of AI in market research

As AI tools become more embedded in research, a range of ethical questions are starting to come into focus. These aren’t hypothetical. They have real implications for participants, clients, and the credibility of research findings.

Transparency and consent

Do participants know they are being interviewed by an AI? Do they know their responses may be analysed by AI tools? Research ethics have always required informed consent, but the rapid adoption of AI has, in some cases, outpaced the frameworks designed to govern it.  

The Market Research Society continues to provide guidance on ethical standards, while initiatives like the GDQ are helping to define best practice around data quality. But researchers and clients still need to be asking these questions themselves. 

Bias in AI systems

AI tools are trained on data, and data reflects the world as it is, not as it should be. This introduces the risk of embedded bias. An AI moderation system trained primarily on responses from one demographic group may handle responses from others less effectively.  

An AI analysis tool may systematically underweight certain types of language or experience. These biases can often be hard to spot, which makes them particularly risky in research. 

Participant wellbeing

Research that involves sensitive topics; mental health, bereavement, identity, financial difficulty; requires human care and judgement.  

Using AI moderation in these contexts raises important questions about participant experience and the duty of care researchers have. 

Data ownership and intellectual property

When AI tools are used to generate, analyse, or synthesise research content, questions arise about who owns the output, where data is stored, and how it is used by the AI platforms themselves.  

Clients and agencies need clear answers to these questions before using AI tools in commercial research projects. 

The regulatory landscape

The UK and EU are both developing regulatory frameworks that will affect how AI is used in research and data collection.  

The MRS Campaign for Better Data and the GDQ’s industry standards represent a coordinated effort to ensure AI adoption in market research happens responsibly. Staying informed about these developments is becoming essential for research professionals. 

How Angelfish Fieldwork approaches AI

We’ll be straightforward about our position: we’re not anti-AI. That wouldn’t be honest or helpful.

AI tools offer real value in parts of the research process. We use them, and we expect their role to grow. But we have a clear view about what AI can’t do, and just as clear about protecting the human practices that make qualitative research valuable.

Our founding belief, still

At Angelfish Fieldwork, we’ve been connecting real people with qualitative research since 2011. We’ve spent over a decade building relationships with participants, investing in rigorous validation, and focusing on the human side of research.

Our Angelfish Opinions community is made up of real, opt-in participants who take part in research because they genuinely want to share their opinions. Every one of them is validated by our in-house team before they take part in any study.

Where we embrace AI

Where we draw the line

We don’t use automated systems as a substitute for human validation

We don’t recommend AI moderation for sensitive topics, vulnerable audiences or strategic research where human judgement is critical

We actively screen for and reject synthetic respondents

If you’re navigating the role of AI in your market research programme, whether you’re exploring new tools or trying to protect data quality, we’d be happy to talk.

Talk to Angelfish Fieldwork about AI in market research

As AI in market research continues to evolve, balancing innovation with genuine human insight has never been more important. Whether you’re exploring AI-powered research tools, reviewing your qualitative recruitment process, or looking to strengthen data quality safeguards, Angelfish Fieldwork is here to help. 

Our team combines human-first qualitative recruitment with a practical understanding of how AI can support modern research workflows without compromising authenticity, ethics, or participant quality. 

Speak with our team today

If you’d like to discuss your next qualitative research project, explore participant recruitment options, or talk about the role of AI in your research strategy, we’d love to hear from you. 

Start a conversation with our team today. 

Explore more from Angelfish Fieldwork

Our accreditations and commitments

  • Cyber Essentials certified: participant data handled securely and in full GDPR compliance 

Frequently asked questions: AI in market research

What does AI do in market research?

AI is used across many stages of the research process, from writing screeners and conducting automated interviews to transcribing, translating and analysing responses. It can help teams work faster and at greater scale, but how it’s used depends on the study and the level of insight required.

Can AI replace human researchers?

Not in qualitative research, at least not meaningfully. AI can support speed and scale, but it lacks the empathy, contextual awareness and judgement that human researchers bring. In practice, the strongest approach combines AI support with human insight.

What are synthetic respondents?

Synthetic respondents are AI-generated personas or bots designed to mimic real participants. They can produce convincing responses that pass basic checks, but they don’t reflect real experiences. That makes them a growing risk to data quality.

How does AI affect data quality in market research?

AI has introduced new risks to data quality, including synthetic respondents and AI-generated answers that can bypass traditional screening. It can also amplify existing issues like disengaged or inattentive participants. Strong human validation at the recruitment stage remains one of the most effective safeguards. Read more on our market research data quality page.

What does ‘human-first’ mean for qualitative research?

A human-first approach means keeping people at the centre of the research process. That includes human-led recruitment and validation, thoughtful research design, and analysis that focuses on what responses actually mean, not just what patterns appear.