by Christina Silver, PhD. SFHEA. FAcSS. Director of the CAQDAS Networking Project
At the CAQDAS Networking Project (CNP), we’ve been raising awareness and building capacity in the use of computers for qualitative data analysis (QDA) since 1994. Over the years, there’s been much debate about the role and implications of the tools researchers use to facilitate QDA.
Debate is healthy
We’ve fostered much of that debate through seminars, conferences, the qual-software jiscmail list, webinars, and our podcast – because debate is a good thing. But unfortunately over the years much misleading information has also circulated – see for example this article by Kristi Jackson, Trena Paulus and Nick Woolf (2018) on the perpetuation of citation error in the CAQDAS field.
The beginning of a new era?
We’re now at the cusp of a potentially seismic change in the actual role that computers can and are playing in QDA practice. When OpenAI released ChatGPT in November 2022, there was suddenly a great deal of interest in “AI” and it’s place in QDA. A level of interest we hadn’t seen with previous implementations of AI such as topic modelling, sentiment analysis and other forms of supervised and unsupervised machine learning. See a post I wrote in 5th May 2023 emphasizing that “It’s not new, folks”
Since late 2022, we’ve spent a huge amount of time fielding queries about Generative-AI in QDA, and we’ve run training sessions in its appropriate and responsible use, and spoken at events all around the world on the developments and their implications. There’s a lot of interest, from advocates and early adopters suggesting Generative-AI can do quicker and better analysis, to sceptics who are fearful of the loss of human interpretation. And those in between these two extremes who are bewildered or overwhelmed, and just need to find out what’s going on and what they implications are for their practice.
Why this new blog?
Fostering debate around the implications of technological developments is one of the core purposes of the CNP. This blog, along with the resources we’re curating in the new area on our website are designed to encourage and open-up discussions about these important issues.
We’re not in the business of creating an echo-chamber; not only do we recognise diverse views, we value them, because there isn’t a one-size-fits-all answer to the question of what constitutes appropriate use of any form of AI – or any other tool – in QDA. Why not? Because there are many types of qualitative data, many qualitative approaches that span the methodological spectrum, many methods of analysis and many tools.
The need for authoritative voices
But we’ve also observed something a lot more disturbing; lots and lots of hype, lots and lots of fabricated information all over social media about what Generative-AI tools can do and how qualitative researchers “should” be thinking about and using them. Many of these opinions are not grounded in the fundamentals of research methodology. This is a real problem for the field.
And so it has become increasingly clear that there’s a need for a place where researchers can find authoritative information about what’s actually possible and how to use these new tools appropriately – if indeed you decide it’s methodologically appropriate and ethical in your context.
We’re very thankful to Dr Anuja Cabraal for suggesting the idea, for encouraging us to push this forward and reminding us that there is no better place to collate authoritative voices and resources on this topic than here at the CAQDAS Networking Project.
And so our Qual-AI pages and this blog are born.
Continuing the tradition
The founders of the CNP – Professors Nigel Fielding and Ray Lee – had the foresight back in the early 1990s, to realise there was a need for awareness raising, capacity building, debate and training around the role and use of the use of computers in QDA. At that time qualitative software was in its infancy and a large part of the CNP’s role – led on a daily-basis by Ann Lewins – was fielding questions about what these programs could do, and supporting researchers to learn to use them. For more information on the genesis of the CNP check out these episodes of our #CAQDASchat with Christina podcast:
- Episode 1 to hear Nigel and Ray talk about those early days, and what they were aiming to achieve in setting up the CNP.
- Episode 11 to hear Ann recount her role in that process.
Part of what we’re doing with our new Qual-AI pages and this blog, is continuing that tradition. Because more than 30 years on, although the use of computers is much more – if not fully – entrenched, debate about the use of technology for QDA remains.
Appropriate uses of tools
That involves really emphasising that Generative-AI is another tool. We have to think very carefully, as researchers and as human beings engaged in social research, what the appropriate role of these tools is in our process. How can we harness them ethically and when is it appropriate to do so – and when is it not?
And there is no one answer to that question. The answer depends on a number of factors, not least:
- Where you sit on the methodological spectrum.
- What kind of data you’re working with.
- How much data you’ve got.
- What your analytic approach to those materials are.
- How you conceive of your relationship to AI tools
The answers are contextual; different for each QDA. The work of Trena Paulus and Jessica Lester on developing a reflexivity framework for technological consequences when integrating digital tools into qualitative research workflows is important here.
What we’re NOT doing
We’re not telling anyone whether or how they “should” be using Generative-AI for QDA. That’s for you to decide, because your project, your data, your context, is what drives whether the use of such tools is methodologically appropriate and ethical.
What we ARE doing
We’re collating resources on the role of AI for qualitative analysis. Publications, webinar recordings, discussions, software developments, and so on. Any resources that are useful to the community of practice. And we’ll also be compiling and linking to key voices in the space, those sharing useful, quality information about this really important space.
Together, with you, we want to discuss what the use of AI looks like in different contexts as it develops, and to share high-quality resources about these issues. Because there really isn’t a one-size-fits-all response or approach.
Work-in-progress : join the endeavour
This is very much Work-In-Progress. What you’ll find on our Qual-AI webpages so far is absolutely not the end result of anything. It’s the beginning.
And we need your help in deciding what to include as we continue building the resources. We don’t profess to know everything there is to know about Generative-AI and QDA. We do know quite a lot – but not everything 😉
We want to build with you, a community around this, a place for sharing ideas, an authoritative space where researchers working in different sectors, disciplines, countries, using different types of qualitative data and methods of analysis, can gather. Not only to find quality information about what’s going on in the space, and advice on how to use tools appropriately and ethically, but also to contribute to the collective building of resources, and the discussion that leads us forwards.
So if you know of high-quality publications, webinars, key voices, or any other resources, around these issues that we haven’t got on this site yet, then get in touch. And let us know why you think they’re good quality resources. We won’t put everything that’s been ever written about QDA and AI on these pages. We’re pulling out the really useful, authoritative and accurate resources. So when making a suggestion, tell us why it should be included, so we can incorporate that into our evaluation.
Guest blogposts
As well as suggestions of others’ resources, we’re also interested in your inputs and musings. As well as posting ourselves on this blog, we’ll be inviting posts from key people in this space to share their musings. And we’re interested in hearing from researchers and students reflecting on the use or implications of, and experimenting with or applying AI to QDA in different contexts.
So get in touch if you’d like to contribute to this blog. We’re really interested to hear your ideas for a post.
Our focus: qualitative analysis
The final thing to say is that our focus at the moment is specifically on the analysis of qualitative data. We’re very much aware that AI can be used to generate qualitative data, to write up the findings of a qualitative analysis and all other kinds of parts of an analytic workflow. But seeing as we are the Computer Assisted Qualitative Data AnalysiS (CADAS) Networking Project, we’re starting off focusing on analysis phases.
That said, we’re quite flexible about what those analysis phases look like. So, for example, we’re keen to think about transcription and the influence AI-generated transcription is having, because transcription is very much an analytic act. Considering the influence of the massive jump in technological capabilities around transcription and what that means for analytic process is therefore important.
To share is to educate
So feel free to be creative and flexible about what qualitative analysis entails when suggesting resources, but be aware that we may decide your suggestions don’t fit with our current focus for inclusion on the webpages. But let’s discuss it. Let’s debate these issues. Don’t be shy to educate us – we love that.
Both sides of the coin
We’re also very interested in hearing from anyone making conscious decisions not to use these tools and to share reflections on that decision. We’re not here to push anything or to reject anything. We just want to create a space of authoritative voices talking about these issues to cut through the hype, cut through false information and fabrications, and really provide something that’s useful to researchers working in all kinds of different contexts.
AI use: This post was drafted by recording a series of voice notes on my phone via the Evernote Ap, which were then automatically transcribed in the Ap. I exported the transcript and edited the draft into the version you read here today. No AI was used to correct or improve the written text. Because I enjoy writing and I want my readers to hear my authentic voice.
