The use of open AI in marketing research is greeted with suspicion or skepticism by some, but also arouses enthusiasm in others. This is clearly the case forActFuture founder Isabelle Fabry, for whom we are witnessing the beginnings of a real revolution in the business of insight hunting. And one that proposes new approaches based on a marriage – a fusion – between the most traditional qualitative research techniques and AI. She shares her convictions with us.
MRNews: In every professional sphere, artificial intelligence arouses contrasting attitudes and often fears. What’s your vision? Should we be afraid of AI?
Isabelle Fabry(ActFuture): For my part, I’m really into the “Go!” of AI usage, which I’ve integrated into my daily life. Including finding new recipes or optimizing my own (laughs), but above all as a marketing research professional. Quite simply because these tools enable us to go further, faster, and open up a whole new world of possibilities. Of course there are fears, and I understand them. AI updates to the power of 2 or 3 the long-standing fear of the machine replacing man. Another major concern is the development of unethical behavior, whether universal or specific to a profession like ours. But it’s perfectly possible to define a framework for these issues and stick to it. It’s up to us, so in my opinion there’s no reason to deprive ourselves of such a brilliant tool. For the qualitativeist at heart that I am, I say to myself, now I can finally break out of the confines of craftsmanship, I have a thinking and stimulating matrix, now installed in my workshop! And all this while retaining and even capitalizing on our fundamental expertise, which is strengthened and boosted thanks to this super “sparring partner”.
Is AI a “contribution” like many others in the history of marketing research? Or a real revolution?
I see it as an upheaval or revolution indeed, but one that is largely still to come. I don’t remember who said that people often overestimate the short-term impact of innovations, but underestimate their medium- to long-term impact. It seems to me that this applies very well to the case of open AI, which is a revolution of which we have so far only seen the beginnings. With probably a huge gap between those who will get on the train and those who will stay on the platform.
As you said, AI is already part of your daily routine. You’re even proposing new approaches that incorporate it. What are they? To answer which questions?
The key principle is to “marry” and even “merge” the research approaches we traditionally use – based on group meetings or individual interviews – with AI. It’s certainly not a question of replacing human intuition, which remains “master on board” and is at the heart of our know-how. But to nurture and boost it, enabling it to benefit from the speed and openness of AI.
The key principle (of our proposal) is to “marry” and even “merge” the study approaches we use conventionally – based on group meetings or individual interviews – and AI. It’s certainly not a question of replacing human intuition, which remains “master on board” and is at the heart of our know-how. But to nurture and boost it, enabling it to benefit from the speed and openness of AI.
In fact, we’ve designed two distinct approaches, covering two different types of need. The first is to use ChatGPT to “increase” the number and relevance of insights generated in our group meetings. In practice, as we identify the insights generated by participants, we work on them in real time with ChatGPT, enriching them. Let’s imagine we’re working on the detergents of tomorrow, and consumers mention the concept of detergents in sheets. AI enables us to instantly grasp the points we need to explore with them, and in particular the conditions for success. And we immediately feed this material back into the rest of the meeting. What used to take several hours, or even days, can now be done in just a few minutes. In the knowledge that the client advertiser, who follows the live generation of insights, can also intervene and have the relevant signals investigated live.
You integrate into the group animation process an iteration between consumers and AI…
Absolutely! This allows us to go much faster and much further than we would have been able to do with the usual protocols. We’re not putting off until later – possibly in a second study – soliciting consumers to look into what needs to be looked into, we’re doing it right now!
The second approach is to use AI not downstream of the participants’ idea generation, i.e. during the group meeting, but upstream. On a given subject, we use AI to capture ideas that we can then rework with consumers. Let’s imagine that I’m looking for the conditions that would best meet the needs of a target in a given market. AI provides me with elements that I can investigate further, and on which our consumers will be able to bounce back. It is, of course, entirely possible to supplement this research work by drawing on sources other than AI, the ones we usually exploit.
In both cases, we orchestrate a form of iteration between the consumers and the AI, so that we can advance our thinking in hyper-accelerated mode and on a very open field. At the same time, we have the latitude, if necessary, to sit down with our contacts on the advertising side to rule out avenues that it would be pointless to pursue. Or to zoom in on others.
In both approaches, we orchestrate a form of iteration between consumers and AI so that we can advance our thinking in hyper-accelerated mode and on a very open field.
Do these approaches seem particularly relevant to specific sectors or issues?
We can spontaneously think of fields where developments are rapid, such as fashion or technology. But, in reality, I believe that today everything can move very fast everywhere, there are no more “plan-plan” sectors!
However, its use presupposes certain “mental” or cultural conditions on the part of advertisers’ teams. Some choose not to set limits too early in the search for innovations, recognizing that they will have to make decisions, but preferring to do so after having explored many avenues, even if some may be a little “disturbing”. Others prefer to limit themselves fairly quickly to what they consider possible. The approaches we’ve just outlined will undoubtedly be better suited to the former than to the latter.
ChatGPT can be “bluffing”, particularly in its ability to summarize information. But it also produces aberrations… Isn’t that a problem?
That’s right. It even became a reflex for me to confront him with them. Here again, I see a dividing line. For some, the fact that AI generates aberrations can be a stumbling block. Others don’t, and I’m on that side. I believe that AI is not intended to deliver certainties, but rather to provide a set of hypotheses that we can compare with other sources, test and try out with consumers. It’s a bit like learning to drive… For a while, you have the impression that the car is going to do what it wants, until one day you know you can control it. We can also draw a parallel with working with a design manager. This one is more or less “good”, but it’s up to you to decide what you like or don’t like about what it does. And you’re the one who makes it happen!
I believe that AI is not intended to deliver certainties, but rather to provide a set of hypotheses that we can compare with other sources, test and try out with consumers.
Does exploiting AI in this way generate additional costs?
No. This is part of a documentary research process that is fully integrated into our work processes. If these tools enable us to go faster, to provide more information, it’s to our customers’ benefit, and it doesn’t cost them any more. But this implies a form of investment on their part, and they need to have the resources to make choices. Studies don’t tell you what to do; they’re designed to provide the information you need to make decisions. The more lighting you produce, the more important it is to manage it properly. But we’re also here to help. Our relationship with our customers is based on partnership. At every stage, we discuss and refine together… And we make adjustments, because we need to be flexible, to know how to abandon certain paths and take new ones. It’s all part of achieving the best results.
Is there anything else you’d like to add?
I’m struck by the fact that many people are afraid of AI, of these aberrations we’ve mentioned, or of the ethical issues its use raises. My vision, however, is that it’s always humans who define the rules of the game, the framework within which to act with these tools. I’d like to take this opportunity to point out that Esomar is doing some interesting work on these ethical issues.
I’m convinced that the marriage between qualitative research and AI can produce something really great! But only if we remain true to the very nature of qualitative research, and to the spirit of openness it demands. We’re always on an adventure to explore new ideas. But instead of walking, you can now take a rocket!
I’m convinced that the marriage between qualitative research and AI can produce something really great! But only if we remain true to the very nature of qualitative research, and to the spirit of openness it demands. We’re always on an adventure to explore new ideas. But instead of walking, you can now take a rocket!