Expanding applications of AI and ML are rocking the world of qualitative research. Once a low-tech bastion of high labor intensity, a large team of professionals was standardly needed to execute a project from recruiters to moderators, notetakers and “taggers,” to translators, transcription typists, video editing and report writers. Now with new AI applications, many of these functions are easily and inexpensively automated to promise a much-improved customer experience.
And yet, this does not mean that the need for team members is disappearing completely. The world of qualitative research has always been known for the human touch. It is at the center: humans talking to humans, using words and interpreting meaning. It requires deep listening, sensitive probing and insightful interpretation to add value in recommendations to clients.
Having the right combination of “High Tech and High Touch” is now winning the day. By embracing the applications for labor-saving AI-based processing functions, the qual experts can focus on the part which is essential and critical to them which is probing and analyzing.
The optimum human qual assistant now has their own equivalent of an AI Assistant to free them up to organize, listen, think, find the “Big Aha”’s and start the report. Meanwhile, the following is being done for them by their AI assistant, referred to as AVA for “automated virtual assistant.”
What AVA can do
- Create real-time transcripts immediately after each session. No need to wait anymore. Analysis can begin immediately. This is because AI automatically tags the transcript based upon keywords entered at the beginning of the session. Then the assistant or moderator or analyst can immediately find what the respondents said verbatim on the designated topics. Doing this by eye can take days.
- Automatically tag the sessions with AI tagging. The keywords and phrases are input into the system and appear in the annotated sidebar of the transcripts. A simple word search by the report writer or assistant calls up all the relevant quotes for the report saving hours if not days of the analyst’s time.
- Simultaneously translate the sessions up to 5 languages at a time for each viewer to see subtitles in their own language(s) of a video.
- Immediately extract video clips: AI makes video editing faster and easier because it enables quick clip extraction based upon AI tagging. So, for instance, if every clip that mentions TikTok is desired, the clips can all be extracted at once and placed into a folder. Use back-room chat to find quotes and clips. Colour coded chat by viewer enables search in the transcripts and videos. So, if a key client says, “This is important!” it can be found and extracted in a snap.
Why you need to keep your human qual assistant
Despite all the wonderful things that AI can now provide to the moderator or qual project manager, there is still the need for human support with respect to communication and troubleshooting. It is recommended to have someone play the role of conduit between the back room and the moderator, especially for online focus groups or individual interviews (IDI’s). This real-life assistant can provide a myriad of key functions including:
- Overall project management
- Write screeners and discussion guides
- Supervise recruiting and screening
- Assure respondent attendance
- Manage incentives
- Tech cheques for audio and camera with respondents
- Support for organising and presenting stimuli such as concepts, products or ads
- Bridge communication between the back room and the moderator
- Oversee quote and clip extraction
- Analysis and report writing
What will the future of qual look like?
Short term, so much is coming to qual research that will make it an advanced customer experience beyond expectation. Given what is in the pipeline by developers at HARK Connect, this year promises to bring real-time sentiment analysis (the ability to capture positive and negative emotion from the words being used via a live dashboard indicator) to live sessions, plus biometric overlays for measuring specific emotions of respondents.
Longer-term, we can expect much-expanded automated functionality. This can include something relatively simple such as AI-led stimuli exposure for the moderator with a “playlist” type approach. But imagine if AI takes us much further such as writing draft reports.
Or ultimately, consider bot type interviewers who are programmed to follow a discussion guide, and to also probe thoroughly with prescribed delicacy and “charm.” Think about the ramifications of such a system, particularly for high-security clients and projects where the fewer people involved the better. Information is extracted and wholly protected as private. Both for the respondent and client. This would have high appeal presumably for sensitive subjects such as medical conditions or personal finances.
Conclusion
The engine of technology is an unstoppable force, even as it applies to traditionally low-tech sectors such as qualitative research. AI and ML in particular can eliminate many of the repetitive and time-consuming user experience parts of the qual research process such as looking for quotes or sections of video to extract.
What AI cannot be expected to do, however, is to fully replace the critical sensitivity, empathy and ethics of human judgement. Risk management requires retaining humans for their critical capacity for judgment. Not every factor or scenario can be anticipated in AI programming and humans provide instant insight into the relative importance (or weights) of variables, as well as the moral center of what’s right and wrong. We cannot responsibly leave that entirely up to technology.
Hence the value of high touch human interaction and support on a project will continue to be irreplaceable. We will continue to need human assistants, but with the promise that they will be freed of the mundane in order to be open to more rewarding and highly valued roles such as applying judgement from both AI and humans, insight and analysis for higher quality outcomes in shorter periods of time.