What does AI have to do with qualitative research?
Absolutely nothing. According to many of you.
You lot say that true qualitative research is all about deep insight, personal connection, close observation, strong intuition.
And then you machine learning types say, 'but computers can write poetry - why can't they do qual research?'.
For you, qualitative is just about smarter ways to distill meaning from large scale human conversation.
You're also right.
For what it's worth, I'm a big fan of GOFUQ (Good Old Fashioned Unadulterated Qual) AND these new AI-based approaches.
I can see both sides now; and I think everyone in this debate needs to give a little bit.
Most of the platforms listed here actually use AI to enhance the moderator's work - not dispense with them - and everyone should welcome a bit of that.
At least it's better that this.
2020 Research Qualboard 4.0
2020 Research has been around a long by todays tech standards.
But it turns out that 10 years' worth of qual discussions represents over a million moderator-respondent interactions, and provides a uniquely rich training set to build machine learning models.
One application in the new Qualboard 4.0 release is Smart Replies - a feature to help lighten the load for moderators. Just like messaging apps, it interprets the content of a post and suggests two or three likely follow-up responses or probes to save the moderator time when replying.
This is an example of Natural Language Generation - you can find more in the Guide to AI for Research & Analytics.
Qualboard also makes use of AI to help with analytics - with automatic keyword extraction, concept identification and image tagging.
Also featured the AI ebook.
Not to keep banging on about it, but nearly 500 people have downloaded it so far. You don't want to miss out.
At its heart, Remesh uses a range of AI techniques to manage, analyse and respond to large volumes of unstructured text data.
Compared to a human moderator, 'Big Qual' tools like this are much faster at summarising content and extracting keywords.
The Remesh platform can manage feedback from up to 1000 participants over a sixty-minute session.
Responses are analysed on-the-fly, clusters of similar opinions are visualised, and moderators can then focus on the most relevant topics or promising ideas and probe deeper in real-time.
Insights.us is an advisory board platform and managed service.
Customers take part in online discussions, answering questions with free text. Machine learning algorithms identify and group similar statements; participants then verify these groupings and propose further insights; and finally the Insights.us research team reviews and finalises the output for clients.
It's a sort of AI-enabled insight community / public consultation platform.
CRISbot (Conversational Research Insight System, no less) is a virtual moderator that uses AI to conduct text-based interviews through a chat interface.
It collects both qualitative and quantitative data through a web-based messaging platform. Results are reported automatically in an online dashboard that includes a full summary report, survey statistics, sentiment analysis of free text questions.
You can learn more about chatbots in this article.
QualSights is a platform for remote video observation, interviews and focus groups.
It can also be used to enhance face-to-face projects and to enable customers or field staff to broadcast live.
The software can capture video from any camera, and uses machine learning to transcribe audio; generate keywords and topics; apply sentiment and emotion analysis; and recognise objects and scenes in videos.
The software also features a drag-and-drop video editor to create clips and showreels.
Another AI-enhanced video tool you should check out is Qualie, which uses consumers' video to help generate consensus around a topic.
And there are loads more video research platforms in the directory, some of which also feature smart video analytics using AI.
All these tools deploy various AI approaches - mostly NLP and a bit of computer vision - to analyse content, interpret meaning and suggest responses. Some of them do this with huge volumes of data - and compress timelines from weeks into minutes.
It don't matter to me whether you call that qualitative or unstructured data analysis or you make up a new label.
I just care that these things work, and that they're useful.