It would be fair to say that the emergence and rise of AI in the research and insights sector has generated strong reactions. Some of these reactions have been highly positive, some more circumspect, and others more negative, driven by concern, anxiety and/or trepidation about the repercussions for an industry that is so fundamentally founded on people.

Change can be problematic, especially when rooted in technology. The reality is that AI is here to stay; our challenge is to ensure that it is incorporated in such a way as to genuinely boost the sector, as well as helping to provide enhanced outcomes for both agencies and clients.

There is, however, an existential risk in implementing and using AI in our sector without due regard to outcomes and consequences. There is jeopardy in the fact that AI carries the risk of disconnecting ‘insight’ from the realities of everyday life. There is a fundamental break in the ‘reality loop’ that connects us to the human world we purport to represent.

Data should represent people, not replace them

Some of the narratives around AI and the ‘production’ of synthetic data provide important clues as to potential jeopardies we need to be conscious of. There are three key areas that are particularly important:

1. ‘Data’ should help us understand real people. This may sound an obvious statement to make, but it does seem as though there is a danger of losing sight of this ‘realness’ of everyday life in our haste for data at speed and with extended reach/scale. The capturing of data becomes the be-all and-end-all, a virtue in and of itself, rather than a reflection of fundamental human truths.

2. Data is meaningless without life. Something to avoid in our incorporation if AI is the situation where data is collected that is lifeless. By this we mean that there is an absence of real human meaning and a decoupling of the captured data from the lived experience that it was either generated from or created to represent.

We should reflect on some of the learnings from the big data boom of a decade or so ago. Too often the endpoint was a mass of de-contextualised information that impressed with its scale but was highly elusive in terms of real context, meaning and application.

The lesson from the big data boom was not that scale lacks value, but that scale without context can create the illusion of understanding. AI risks repeating that mistake, only faster and with more fluent outputs.

3. ‘Reality loops’ are central, not an option. No one method or approach should be used as either fully representative or validating. We need to remind ourselves that ‘triangulation’ – the use of multiple approaches, perspectives and/or methodologies – is core to best practice. We seem to be drifting away from this, and it is, ultimately, counterproductive.

When we have triangulation in the research process, we are effectively introducing what can be described as ‘reality loops’. When using AI in research and insight, we need to recognise the importance of these necessary reality loops.

When everyone is a loser

People matter, not just for moral reasons; they matter also because they/we are the focus of enquiry, whether as ‘shoppers’, ‘consumers’, ‘targets’, or simply ‘our audience’. People, real human beings, buy products and services, shop in stores, they are loyal or fickle in relation to the brands we represent and build, and are our constant opportunity as well as potential jeopardy. Data should serve and support this reality, not stand aloof from it.

When data becomes the end goal, everyone is a loser. This is the point where ‘reality drift’ really kicks in. We can see this happening everywhere. Data is presented, compared, annualised. It gets bigger, broader, faster, louder. But we lose human centricity. The landscape of meaning becomes thinner, flattened by the absence of real human meaning. We make judgements and comparisons with data that is bereft of life.

It is also important to remember that there is another jeopardy in this scenario. Without ‘reality loops’, which act as both context and a sense-check for data, we run the risk of transforming data lakes into swamps; a steady flow of error and inaccuracies polluting the sources we increasingly draw upon.

Clients lose in this scenario; big decisions are made based on inaccuracies and without meaningful context. Agencies lose because their role becomes one of process management, rather than interpretation and context creation. Participants, consumers, shoppers and the universe we research all loses – because representation is limited, partial and decontextualised.

When the medium is the message

It seems that much of the narrative around AI currently focuses intensely on what the technology is and can do. Outside the implied or stated critiques of ‘traditional research’, not much is said about understanding human beings, as complex and fluid entities. What we too often get is a valorisation of data, with speed and at scale, but little of the ‘fuzzy brilliance’ that makes human beings, and the lives they lead, so fascinating.

This is highly problematic. Life is messy. Data is as irrelevant as it is relevant. Context matters. Human understanding matters. Culture and context matters. Strip these away and what’s left?

In contrast, and for example, qualitative research (broadly defined) aims to create a depth of understanding around foundational human emotions, passions, needs, ideals and beliefs. Qualitative research should never be the answer, it should always be the route through which we access the answers required.

It does seem as though AI is often positioned as both approach and answer. AI becomes the story, the star of the show; a celebrity presence that dwarfs the original strategic intent. AI becomes both the medium and the message. Reality loops are broken.

Step back and breathe

We all need to take a step back and think carefully about where we are now and our direction of travel in relation to AI. The reality is that AI could be harnessed to achieve wonderful things; but in a sector where understanding human beings is critical, it could also represent a damaging detour. We can become consumed by method and the process of capturing ‘stuff’.

In the rush for newness, we risk losing sight of what really matters. As we fine-tune our ability to capture ‘data’ with greater speed and at scale, we must be cognisant of the ‘what’ and ‘why’ behind the things we do. Are we seeking scale or understanding? Can the two exist together? Will AI-enabled data bring us to the heart of meaning, or does it serve simply to raise more questions because it isn’t twinned with a deeper, human understanding?

Ultimately, we need to accept that people matter and that they are central to the research and insight, even in the age of AI. If we lose sight of this reality, we risk our data becoming worthless; billions of points in a world without meaning.

Mark Thorpe is director at Truth Consulting

• The full Delphi report, ‘Who owns understanding? How AI is reshaping value, expertise and accountability in research’ is accessible via the MRS website.