Are those in market research, particularly in qualitative, at risk today from artificial intelligence?
Which “parts” of our work are more so, which parts are “safe”? These are questions we often ask ourselves, indeed … we asked the Oracle of Delphi 😅 - who is none other than a GPT (the best we could find), able to tell us if and how much we are at risk.
So yes, we therefore asked the AI if the AI will, sooner or later, take (part of?) our jobs away.
The model we used performs an analysis of the level of exposure to which each stage of our work (and any work) is subjected, according to a scale ranging from “0” (i.e., no exposure) to “10” (high exposure), providing for other specifics about what is driving that exposure (e.g., LLMs, audio/video-based models, complex reasoning-based models, operators, robots…).
A high level of exposure means having a certain percentage of time saved, thanks to the work done by AI -- but this, inevitably, could represent (eventually, or to some extent already) a reduction of our role, or even a replacement of the work that has been done to date by the researcher. But let's go in order…
While it is true that systems and tools already exist for moderating (mainly interviews, rather than focus groups), it is also true that such tools are mainly useful for short, 1-to-1 interviews, perhaps through semi-structured questionnaires. It is one thing to have an automatic re-launch carried out by a synthetic, entirely AI-managed moderator; it is another to have the key role of the qualitative researcher, who experiences the (often hour-long, or even longer) interview through his or her emotions, impressions, and experiences.
Ours is a work of relationships, of contact, of getting to know each other, of nonverbal communication, of smiles, of silences, of deep knowledge of the other, of ad hoc qualitative insights and probes that arise from listening, from experience, from knowledge of the research objectives, as well as from the client's brief.
On this, to date (and perhaps in the future), there is no AI that works.
The contribution of AI and LLM is definitely important (more than 50 percent perceived impact, according to the Oracle we surveyed) when it comes to drafting discussion guides or searching online topics that have to do with the subject of the market and field research to be carried out.
While it is true that in these cases there are various LLM templates that can produce documents and specifically discussion guides in a matter of seconds, it is equally true that such tools should be used with care: these are important insights that nevertheless need revision, to be used as “assistants,” as support, but never as “autopilot.” It is a starting point, a not insignificant aid, rather than starting from scratch.
Even in the presence of such aids, the direction must remain with the qualitative (human) researcher, his or her filters mediated by experience, his or her own “style” and approach, his or her uniqueness-this is what makes the difference with the client, which produces quality work (market research).
In recent times, those who do our work have seen a substantial change at the level of the materials that are used to review/read interview content. We have moved from notes taken by a note-taker, aka verbalizer or “recorder,” to automated speech-to-text transcription systems, which have the virtue (virtue?) of reporting EVERYTHING that was said, word for word.
Too bad that the note taker, being a human, could make such transcripts less textually dense, filtered according to priority and relevance, perhaps even more orderly as to sections/chapters of the discussion guide. AI reports everything, but this everything is often more of a chaos, an indistinct collection of words, not always by any means accurate, needing further orgenization, not always easy to understand.
Anyway - the work of “recorders” (the way Italians call note-takers 😅) is less in demand today (at least in Italy). It remains, those who want higher quality work often prefer a human verbalizer, “the old-fashioned way,” but the trend seems unstoppable. It is not just a matter of cost: the AI (at no cost) in this case is seen as an “Infallible Machine,” vs. the human unable to transcribe everything.
We then investigated other points - we leave you with our video
In several cases, online, we found a phrase, which sounds a bit like this:
“It's not AI that will steal your job, but those who are able to use AI better than you!”
“It's not AI that will steal your job, but those who are able to use AI better than you!”
So, it is necessary to experiment, to try, to use AI. Do it!
