The meteoric rise of AI in our industry is unsettling for many. The recent announcement from Anthropic that its Claude platform was used to conduct ‘qualitative’ interviews with 81,000 people has created controversy.

From Anthropic’s perspective, it has achieved something special: “We heard from people across 159 countries in 70 languages. We believe this is the largest and most multi-lingual qualitative study ever conducted.”

On the surface, a simply breath-taking achievement; huge in scope and ambition. The ultimate fusion of deep meaning at a size unmatched by any methodology to date? Well, it depends, really.

What are you talking about?

We need to take a step back, dust ourselves down and re-focus. From my perspective, the most important question that needs to be asked is “what do you mean by ‘qualitative research’?” 

‘Qual’ can mean very different things – some of which are contested, others widely agreed upon. Ask a seasoned anthropologist what it means and you will, almost certainly, get a very different answer to someone working in the commercial sector.

It’s fine for there to be different interpretations – that’s healthy, but we must have points of view in the first place. Much less healthy is a situation where qualitative research is poorly defined and taken for granted; allowed to become hollow, minimal and simplistic. 

I argued in a previous article for Research Live that much commercial qualitative research has been reduced to simply talking to people, presenting huge existential risks for the future. If ‘qual’ is just talk, then all AI has done is to put a huge boot through an already open door.

The thinning of qualitative research

Anthropologist Clifford Geertz talked about ‘thick description’ as the major purpose of the qualitative arts – specifically, but not exclusively, ethnographic methods. Thick description involves unravelling and presenting layers of meaning, supported by detailed context and analytical rigour. It looks at the symbolic and cultural as well as the patterns contained in everyday discourse and interactions.

In contrast, ‘thin description’ is superficial and ‘light touch’. In thin description, you have both simplicity and the very real threat of misrepresentation and misunderstanding. Thin descriptions tell us very little about ‘the world’ and rely heavily on assumptions. They are usually underpinned by everyone in the room benignly suggesting that they know exactly what the thin descriptions are saying.

Commercial qualitative research has, for far too long, allowed thin description and practice to reign – this has made it super easy for AI to walk right in and steal the show, because AI can do ‘thin’ research brilliantly.

Blurred genres and bothersome bulls

It’s fascinating how much the world changes and how little our sector frameworks have evolved. We still have a narrative of ‘qual’ and ‘quant’ as being, pretty much, the universe of what we do. Yes, we have semiotics, cultural analysis and even linguistic analysis as part of the mix – but they’re pretty marginal and often actively marginalised.

AI has been the proverbial bull in a china-shop, making a mess of what we use as definition. In commercial qualitative research, the blurring of boundaries (read: lack of definition) has been like a gift from the gods to a growing number of AI-centred organisations.

The question is, do we just dust off the chintz, or do we find the gold in what we do?

We should live in different houses

Now is the time for a fundamental re-positioning of commercial qualitative research. There is simply no alternative. Without change, AI will overtake the sector – it’s as simple as that.

This does not need to be a loser/winner scenario. Commercial human-centred qualitative research and AI-enabled research can co-exist, but they need to live in different houses – respect without reproduction. They should know where they live and why each house is different, each beautiful in their own way.

Why this separation, you may ask? Quite simply, because AI-enabled interviewing and human-centred qualitative research are different; they do things differently and have (or should have) different uses. AI interviewing will provide unimaginable scale, but its value will be on gaining simple answers to relatively simple and pre-defined questions. There is, if we are being truthful, not much room for the kinds of nuance and cultural framing that human-centred qualitative research should give us (when done properly).

What’s next?

We need to stop grumbling about AI. It’s here. It’s staying. It will continue to be disruptive. What’s needed is a practical manifesto for change; one in which human-centred commercial qualitative research is re-positioned and its principles and values clarified. This will represent a foundation for both growth and pride and, critically, help clients make the right choices. 

In brief, here’s my suggested starter manifesto for change:

1. MRS, Esomar, the AQR, and any other relevant body, join forces to craft the principles behind qualitative change
2. There is clear identification of the differences between human-centred commercial qualitative research and computer-mediated interviewing
3. A series of qualitative ‘commitments’ are crafted and companies that sign up are accredited
4. A random selection of organisations will be audited on their qualitative performance each year, impacting accreditation
5. Of those employed with a degree, a minimum of 50% of qualitative researchers should have majored in a social science subject (including AI generated research)
6. Qualitative training should be hard-wired in every business, with training visible on the websites of all accredited companies
7. New qualitative certification is available for those who pass industry qualifications. Importantly, this process should not be the province of organisational bodies alone; clients and agencies should be integral to ideation, planning, delivery and maintenance
8. Instead of just talking about the value-add in ‘proper’ qualitative research, we get better at demonstrating it. We could outline this value-add in every proposal and debrief. What did you do? How and why did you do it? Where is the evidence of expertise?

This is a starting point, and it represents a basis to build upon. Think about how you could take it on, improve it and contribute to the change that’s sorely needed.

Mark Thorpe is head of thought leadership at Truth Consulting