Bolt Insight Virtual Session: Unfiltered Truths: How AI-Moderation is Redefining Honesty in Consumer Research

Unfiltered Truths: How AI-Moderation is Redefining Honesty in Consumer Research 

Editor's note: This article is an automated speech-to-text transcription, edited lightly for clarity. 

This session was presented during the Quirk’s Virtual Sessions – DIY Research series on June 11, 2025. The session focused on AI moderated interviews in qualitative research.

Ester Marchetti, co-founder and head of innovation at Bolt Insight and Caroline Brethenoux, managing director at Human Dot+ discussed the benefits of AI moderation and illustrated them through a case study with U.S. Hispanic and Black cancer patients.

Marchetti also discussed where she sees AI technology going from here in different aspects of life.

Session transcript 

Joe Rydholm

Hi everybody and welcome to our session, “Unfiltered Truths: How AI-Moderation is Redefining Honesty in Consumer Research.”

I’m Quirk’s Editor, Joe Rydholm and before we get started let’s quickly go over the ways you can participate in today’s discussion. You can use the chat tab to interact with other attendees during the session. You can use the Q&A tab to submit questions for the presenters during the session, and we will answer as many questions as we have time for during the Q&A portion.

Our session is presented by Bolt Insight. Enjoy the presentation!

Ester Marchetti

Hi everyone, I'm Ester Marchetti. I'm a co-founder and head of innovation at Bolt Insight. 

Today we'll explore how AI moderation is not just a trend but a pivotal shift that's redefining honesty, empathy and depth in consumer research. I'm joined today by Carol, managing director of Human Dot+.  

Carol, over to you.

Caroline Brethenoux 

Thank you very much, Ester. I'm very happy to be here. I'm very excited to share as well one of the case studies that we have had in our strong and lasting partnership with BoltChatAI. 

We are Human Dot+., we are next gen insight and innovation company, tapping into different methodology to make sure we understand emotions, drivers, desires and triggers that helps us understand the ‘why’ behind the ‘what’ and what shapes people's behavior. 

Ester Marchetti

Wonderful. Carol, thank you so much for being here.  

So, at the heart of consumer understanding lies conversation, I hope you'll agree with me, not just what people say but how they say it and what they truly mean.  

Why are honest responses are the holy grail of qualitative research? It's because they really make us understand why they are doing what they're doing. But there is an honesty paradox. People want to be truthful, but they fear judgment.  

As a former client, consumer connection was everything. I loved speaking to people, visiting their homes, championing real consumer closeness because it was not just about insight, it was about empathy, nuance and the everyday realities behind behavior. There was that real world perspective when I would go and meet some of the consumers. I was serving as a client going into their lives, understanding their reality, their setup. How their kitchen was laid out was so eye opening in understanding how well they would relate with the product that I was working on. 

Also, I was trusting interpreters and translators to bridge the gap between data and emotion to help us as brand stay human.  

So, great quality research was so vital to really understand and feel with consumers, but there were some problems sometimes because it was not always possible for every project.  

There are some problems that market research faces because despite this intention, there are some challenges when it comes to interviews. I'm going to go through some of this with you.  

Social desirability bias.  

Often people tailor their response based on what they think their interlocutor would like to hear. They try to filter what they say through what is socially acceptable or expected rather than what they really believe or do. That leads to less authentic insights. Especially in an interview where they feel watched, evaluated.  

There is a fear of judgment when topics are personal, taboo or emotionally charged. They might hold back out of fear of being judged and this might be conscious, or they might not. This discomfort really limits their willingness to share openly, especially in face-to-face setting.  

There is some limited consistency at scale. Human led interviews can vary in tone, in approach and interpretation from one moderator to another. They're also resource heavy, which makes it difficult to reach large diverse samples consistently, especially across markets or sensitive topics.  

Given that, why and when can AI help?  

Imagine an interviewer that doesn't judge, that doesn't get tired and that asks just the right follow-up questions every time. That's what the AI can do. It's not about replacing humans. It's about removing barriers and the areas AI can help support what we do as humans.  

For example, AI provides real anonymity. We know that when the AI is on the other side of the conversation, people feel less scared. They don't feel judged as much. So, they're able to open up sometimes more when the topic is taboo.  

AI helps with scalability. By offering thousands of in-depth interviews, not only a handful.  

It can provide asynchronous flexibility when it gives people some space to think and reflect before responding.  

And AI doesn't get fatigued. So, it can provide some conveniency when analyzing thousands of hours of transcripts.  

When we launched BoltChatAI, we were thinking about an end-to-end qualitative engine for qualitative at scale. From dynamic discussion guides to global recruitment to AI led interviews in local languages, it's quality scale reimagined. The way it works, it can really help you every step of the way.  

From the start when based on your prompt, so your target audience definition and your research objective, it can create a discussion guide. A conversational one tailored to your objectives and available also 50 plus local languages. To the recruitment of engaged participants. They will be connected for a one-to-one in-depth 30-minute interviews that are moderated by the AI. 

Our AI moderator conducts these one-to-one interviews with hundreds of participants in multiple countries and local languages. So, at scale with quite a sophisticated method in which the AI has been trained based on real human conversation. So, the AI knows how to interact with respondents and how to probe deeper.  

Then finally to the reports. When analyzing all of those transcripts, BoltChatAI generates bespoke reports in less than 24 hours.

Based on what I was mentioning before, an AI moderator can really help and support human researchers in quite a different way.

On one side, this is not a generic chatbot. It's not just a technology that anybody can replicate, but it was trained on hundreds of real interviews by experienced moderators. It combines AI precision with qualitative intuition. So it's methodological, rigorous and emotionally intelligent.

It also has dual agent moderation. So, an AI moderator will follow the discussion guide. An AI referee will check every single answer that respondents give and then tell the AI moderator to probe deeper when the response is too short or not specific enough or has not been understood. It reacts to what the respondent shares, in terms of text, voice, videos, pictures and more. Then has a two-way discussion to really uncover deeper insights. It probes deeper at several levels, including when it detects strong emotion from the respondent that we need to dive deeper into.

And that is fundamentally centered on a human in the loop. So, not only has been trained on a human chat, but we keep training it every day, every week with new conversation with rigorous human quality checks made by experienced researchers.

I also have a video here to show a bit more of the respondent experience.

[At 8:13 to 9:47 a video shows the respondent's experience using BoltChatAI. The following is a transcription of the video audio.] 

Since taking these vitamins, I definitely say my body feels more healthy. I used to feel quite fatigued in the morning and like I said earlier, I used to get headaches quite often.  

So, my last purchase, I actually went in store, to Boots. I'm very lucky that I actually live quite close to my local store. So it's sort of now in my routine. Whenever I notice that I've only got two or three tablets left, I will actively make the choice to go down to my local shop and I will purchase a new bottle of the daily vitamins that I use. 

Ester Marchetti 

In terms of use cases from early-stage idea testing to quote collection, market exploration, trade storytelling and understanding of creative, our clients are using BoltChatAI to make better faster decisions with higher quality across the innovation and marketing life cycle.

So, just something to think about for a moment. We have seen hype cycle before, but this time AI is not a gimmick, it's utility driven. It's constantly improving. It's already embedded in our daily work life and in everything that we do. 

We are in a surging real world adoption moment, but soon there will be AI agents and autonomous workflow coming. Actually, generative AI moderation is already a thing, and it was embedded into what BoltChatAI does from day one with our AI moderator and AI referee.  

Companies are working towards AGI. It is plausible, still speculative, but we can really think about how they are improving more and more and being able to handle more tasks independently. So, really this is only the beginning of this journey.  

If some of you have seen the movie “Her,” we are not that far from those moments. Something that really stopped me in my track and when I saw this information, that you can see on the slide here [slide referenced is at 11:02-11:56], were the key use cases for using generative AI are changing from more utilitarian tasks of generating ideas to really providing emotional support and connection. Therapy and companionship coming up at place one followed by organizing my life and finding purpose. 

The way respondents and people are engaging with AI has already changed. We can expect them to be more and more comfortable with engaging with AI. At BoltChatAI we disclose from the start that there is going to be an AI moderator interviewing them. That really helps people to open up without that fear of judgment.  

AI alone gives you speed, but AI with humans gives you imagination, moral judgment and intuitions. That's why it's so important when tools like ours or what Human Dot+ does when tools are built with human and AI at the core.  

When we built our AI moderator, the point was not to replace humans, but it was to extend human empathy at scale. This moderation doesn't just ask a question, it listens, empathizes and plays back what respondents say creating a connection and helping them to open up more. So they don't feel that on the other side they will be a judgmental person but they feel there will be a neutral entity that will help them open up and connect.It can really help unlock honesty nuances even with some targets and topics that might be taboo. 

I wanted now to give the stage to Carol so you can explain a bit more about the journey of Human Dot+ with BoltChatAI. Over to you.

Caroline Brethenoux 

Thank you so much, Esther.  

We have had, since the first day we met with BoltChatAI, an alignment on purpose because we are all about really understanding people and it's all about, yes, AI is our superpower, but at the end it's about really leading with that human intelligence, that empathy, to really understand people where they are. 

This is just an example of one of our latest partnerships. We, at Human Dot+, were working  with a cancer center in New York with a very specific task to understand Black and Hispanic populations in New York, New Jersey and Connecticut. Especially when it comes to cancer, their cancer journey, their drivers and barriers to consider cancer centers and even to really understand what access to cancer excellence means for them.  

So, we did a lot of work through our own AI powered property data, really tapping into what was the spontaneous peer-to-peer chatter already online about cancer, their emotions mapping their triggers in terms of seeking care, etc. But we felt that we were missing a little piece of our insight journey that was really understanding and probing people. 

In our methodology we usually don't recruit anybody, we don't ask any questions, we really go into it about discovery. But sometimes there are some hard questions we need to actually ask people.  

Now health care is a very difficult topic because even when we recruit people, there are the biases that Ester was talking about. Especially when it comes to healthcare in the U.S. and the Hispanic and the Black communities, especially, being so distrustful of the system. Sometimes they don't really want to tell you exactly how they feel because they don't want even their opinion to shape a system that works against them. 

What was really interesting was to be able to work with BoltChatAI to remove all of these different barriers because the interaction that our respondents had through the platform was really consistent.  

The fact that it was through those AI moderators removed some kind of, I would say biases in answering and we got some amazing data, some amazing insights. That really helped us to use all those different quotes to complement all of our insight journey and to really shape the strategy that we were leading with.  

So, this was an amazing case study for us because that was removing all the different barriers that we usually see. On top of that, those cultural barriers the difficulty to engage in health care topics through the lens of the cultural background of Hispanic and Black communities.  

What we really appreciated, for example, was the ability to be very iterative in the process. Many times, in our methodology that we use, we like to drill into our data set to really understand the ‘why’ behind the one and keep on iterating those ‘whys.’  

This is where really we saw the strength of the BoltChatAI platform. With all of those moderators that have all the smart probing type of capabilities helped us to go even deeper so that people could really open up in a way that they felt really safe.  

Especially when it comes to cancer. It's a very emotionally charged topic. So, to feel safe, to be emotionally vulnerable in this conversation was really the key for us to unlock. And I have to say that's with chat AI in the platform, we really managed to get to that space and get those really insightful conversations in terms of synergy and scalability.  

What was really interesting for us is the fact that it's very difficult to recruit people that are at a very specific stage in their journey.  

So, here we're looking at people who had a cancer diagnosis or who had somebody in the family that had a cancer diagnosis, who were from the Black and Hispanic community, were living in New York fundamentally, but also New Jersey and Connecticut. It was very important for us to find the right solution that could help us to get to those conversations that matter.  

At the same time to be able to do that in absolutely no time, just a couple of days at maximum because we were running against the clock. This is where I think for my entire team, the platform blew our minds in its ability to get as granular as we usually have our standards. Really looking for that emotional and very deep rich conversation but in just a couple of days.

Here you have some of the key quotes and there were too many to actually put on this slide [slide appears from 18:05-18:47], but I would just read one of them from a male African American that was age 56.  

“My cousin was diagnosed with breast cancer and it got me pretty hard. She owned her own business in hair styling and used to let me work there when I was younger... She recently passed away because of it so now I get breast screening when I'm supposed to, so I don't put my family through the heartbreak. We have already experienced losing my cousin.”  

So, this ability to get really close and personal with people and to really open up about cancer was really a differentiated factor for us. 

Of course what was really interesting in the platform is this ability to really guide the conversation and get the answers to some of the big questions that we had in the project brief. 

One of them was really understanding that journey, but as well what motivates that journey towards treatment and seeking care. Especially within that overall mistrust and the difficulty of access to the health care system in the U.S., we could understand how trust and the human connection where actually the way to unlock that kind of connection and that access to care for Hispanic and Black communities.  

Another important insight as well for us was really to understand what do people view as accessible care and when we talk about personalized care. When you talk about access, usually people think about health insurance, but for us to be able to understand that access is actually about being able to speak my language, being able to sit down with me, being able to understand me as a whole person and to treat me not as a diagnosis, not as a case, but as an entire person with her own story was very important. 

That's why I can't rave more about our partnership because it gave us a lot of beautiful insight that were really important for us in our discovery and in shaping the strategies to really create impact with our client.

Ester Marchetti 

Thank you so much, Carol. It was such an incredible project to work on and thank you so much for your partnership. 

So, just a couple of things before we close. I just wanted to mention our new features, our dynamic persona. When we created this tool, we created it basically because as a former client it's a tool that we wish we had.  

I know when I was a client, qual was not just about one project. Insight was about cumulative learning. So, without some of our features like dynamic persona and meta analysis, we can reactivate all research, track shift, build a living strategy grounded in real conversation. 

With persona, for example, we can really drive continuous understanding. 

You can ask your persona why they behave the way they do, compare across demographics and test live messages. So, BoltChatAI can really turn your research into an always on, living conversation with your audience.  

Each new project adds to the ecosystem. With every aeration your ‘Einstein stack’ becomes sharper, faster and more powerful. We compound intelligence as we go still maintaining that incredible depth.  

To close, can AI truly connect? It's a fair question I think, but empathy doesn't mean feeling emotion. It means really creating a space where others feel safe to share. And its exactly what AI does best. It never judges, it never interrupts and always listens.  

The future, to us, is that this is only the beginning. AI is really evolving into a trusted listener. Scale, consistency and sensitivity are really helping us get to truth that were previously out of reach so that a quadrangle of speed, cost, effectiveness, quality and scale at the same time. The future is not just tech enabled but is more human than ever.  

I'm going to close by saying that we are proud to partner with over a hundred global brands already redefining how they do qualitative. So, if you're curious to explore what this looks like for your team, let's chat.  

Thank you so much.