Insight leaders are moving from experimenting with AI to putting it into practice, revealing new opportunities, challenges and questions.
Six months ago, we hosted our first AI-in-Research Roundtable in San Francisco, bringing together insight leaders to discuss the growing role of AI in the research function.
At the time, the conversation was dominated by uncertainty. Teams were under pressure to adopt AI, questions around quality and trust were everywhere, and many researchers were still trying to understand where AI genuinely added value versus where it simply generated noise.
We recently brought together a new group of insight leaders in Palo Alto to see how perspectives had evolved.
The concerns we heard in March have not disappeared. In fact, almost every theme from the first session resurfaced. But the conversation has shifted from whether AI should be used to how it can be used most effectively.
The industry is moving from experimentation to implementation, and that transition is revealing a new set of opportunities and challenges. Here we share our seven key takeaways on what’s changed, and what hasn’t in the last six months.
1. The pressure to adopt AI hasn't gone away, but resistance is fading
Back in March, AI adoption felt largely driven by external pressure. Insight teams were being asked to work faster, deliver more, and demonstrate progress whether they felt ready or not.
Six months later, that pressure remains firmly in place. Many participants described leadership expectations for measurable productivity gains and ongoing experimentation with AI tools.
What’s different is the attitude.
Rather than viewing AI as something being imposed on them, researchers are increasingly finding ways to make it work for them. Teams described using AI to handle transcript clean-up, support coding and analysis, identify patterns they may have overlooked, and even challenge their own thinking.
One participant described AI as helping them answer an increasingly important question:
“What am I missing?”
As researchers come to rely on the value AI delivers, the conversation is shifting from adoption to optimization.
2. Quality concerns haven't disappeared, but trust is becoming more nuanced
In March, trust emerged as one of the biggest barriers to AI adoption. Researchers worried about hallucinations, shallow synthesis and outputs that appeared polished but lacked real substance.
Those concerns remain. But the recent discussion was noticeably more balanced.
Participants acknowledged that AI outputs have improved significantly over the past year. Better tools and more sophisticated prompting practices mean less time spent correcting obvious mistakes.
However, trust is still something that must be earned.
Researchers spoke about developing ‘trust but verify’ habits and validating outputs across multiple sources before acting on them. Rather than trusting AI universally, they are learning which tools perform well for specific tasks and where human oversight remains essential.
3. The biggest quality debate has moved to synthetic respondents
One area where interest and skepticism remain particularly high is synthetic personas and synthetic panels.
While vendors continue to promote increasingly sophisticated solutions, our roundtable participants questioned how effectively synthetic respondents can replicate genuine human behavior, unpredictability and context.
At the same time, several researchers noted that traditional sample quality is becoming more challenging too, with concerns about AI-assisted respondents and fraudulent participation increasing.
The challenge facing the industry is no longer simply evaluating AI-generated data. It’s determining how to maintain confidence in data quality across both AI-generated and human-generated sources.
4. AI moderation is finding its place, but it isn't replacing human moderators
One of the clearest areas of consensus was around AI moderation.
Participants recognized significant benefits in terms of scale, speed and multilingual research. AI-driven interviews can reach larger audiences and create more engaging experiences than traditional surveys.
However, researchers drew an important distinction. The benchmark for AI moderation shouldn’t be human moderation - the benchmark should be unmoderated research.
AI is proving highly effective at improving self-complete approaches, but participants were largely aligned that it still lacks the adaptability, intuition and contextual judgment that experienced human moderators bring to a conversation.
In other words, AI moderation is expanding the researcher’s toolkit rather than replacing human moderation.
5. The role of the researcher is evolving faster than many expected
Perhaps the most thought-provoking discussion centered on how AI is reshaping the research profession itself.
In March, participants expressed anxiety about AI replacing researchers. By August, those concerns felt more tangible.
Several attendees discussed organizational restructuring, reductions in specialist research roles, and growing expectations that individuals can cover broader responsibilities with AI support.
At the same time, there was concern about what this means for future talent development.
Historically, many researchers built expertise through the operational tasks that AI is now automating. If those entry-level opportunities shrink, how do future researchers develop the judgment, critical thinking and commercial understanding that organizations will continue to need?
This is becoming one of the most important questions facing the profession.
6. Human value is becoming clearer, not less relevant
Despite all the discussion about automation, one message came through consistently.
Participants were more confident than ever about where humans continue to add unique value.
Strategic interpretation. Contextual judgment. Storytelling. Stakeholder influence. Empathy.
As more organizations build AI-powered knowledge repositories and self-service insight platforms, researchers increasingly see their role not as delivering information but as making sense of it.
Having access to insights is not the same as understanding them.
Several participants described using AI to generate alternative perspectives or even create stakeholder-specific viewpoints, but they were clear that determining what matters, what is relevant, and what action should be taken still requires human judgment.
The ‘so what?’ remains a distinctly human responsibility.
7. The industry needs a clearer framework for responsible AI use
One finding that surprised us was how few organizations have formal frameworks governing AI use within the research process.
While larger companies have started classifying low-risk and high-risk applications, many researchers described relying on personal judgment rather than organizational standards.
As AI becomes increasingly embedded in research workflows, this feels like the next major challenge for the industry.
There’s growing agreement on the need for greater clarity around where AI should be used, where human oversight is required, and how organizations can maintain research quality and transparency as adoption accelerates.
Conclusion
Looking back at the conversations from March and August, our biggest takeaway is that the industry’s core concerns haven’t fundamentally changed.
We’re still talking about speed, quality, trust, ethics, stakeholder expectations and the future role of researchers.
What’s changed is our level of familiarity.
Researchers are no longer imagining what AI might mean for the profession. They’re living it every day.
The conversation has evolved from uncertainty to practical application. The challenge now is how we harness it in ways that strengthen the quality, impact and credibility of insight.
And throughout both discussions, one theme remained remarkably consistent: The more AI becomes embedded in research, the more valuable human judgment becomes.
Because while AI can surface information, people still create understanding.
If this has sparked any of your own opinions or curiosities – and you‘d like to discuss them with your peers at one of our roundtables, you can express your interest in future events by dropping us a note at events@strat7.com.
About the authors
- Shelby Howard
- Principal, STRAT7 Incite
As a Principal, I specialize in qualitative research, partnering closely with clients to find the insights that drive meaningful change. Research gives you the chance to talk with people from all walks of life, all over the world, people you might otherwise never have the opportunity to connect with. I genuinely love those conversations, and I believe businesses grow stronger when they truly hear and understand the voice of their consumer.
- Katie Peters
- Director, STRAT7 Incite
As a Director, I design and lead qualitative projects, and am happiest strategically partnering with clients to bring about real change. This comes from diving deep into the business issues and uncovering highly actionable insights on how respondents think, feel, act, choose, buy, and change. As a user advocate, I am deeply passionate about bringing consumer insights to life. And I absolutely love what I do.