Don’t delegate to AI the stuff that makes us special

An anecdote at WPP Converge caught my attention. It was a comment in a conversation about all the incredible things that we can now do with AI: “It’s great that AI can now write poetry, but can it do the dishes instead so that I can write the poetry?” WPP Open’s Roy Armale, chief product and growth officer, was talking about the partnership between human and AI in our industry and beyond. 

AI is changing and improving how we do our jobs, but among the smart delegation of tasks, we need to be careful not to delegate the poetry. 

In their excellent book 98% Pure Potato, John Griffiths and Tracey Follows take us back to the origin of account planning. Talking to those who were there when Stephen King and Stanley Pollitt were at JWT and BMP respectively, the authors present a history of the job and people’s motivations for doing it – and understanding people is paramount.

There is a huge emphasis from all the planners interviewed on the necessity to interrogate what motivates and drives people. Reflecting on what has changed in planning since the early days, the authors pick out a pulling back on this insatiable curiosity about consumers.

It is true that much less of a planner’s time is now spent doing research. It is rare for a planner to commission qualitative research, let alone conduct it themselves (we are very fortunate at Ogilvy UK to have in-house research practitioners).

Planners have 10 times as many demands on them now; 10 times as many things they need to know or manage. But the desire to be, as Leslie Butterfield put it (and quoted in the book), “a detective. Somebody who has got a curiosity, enthusiasm, thirst for knowledge, almost a childlike inquisitiveness”, that is as characteristic now for planning as it has ever been. 

Real insight – the goose-bump-producing, senses-tingling feeling of finding a truth – comes from connection. On the shop floor, around a table, in someone’s sitting room or even on-screen. Every planner knows how it feels to have touched on something that creates a chemical response.

It is often those very moments that have inspired the creative work we remember and built the brands we love. As AI tools develop, we need to think carefully about not just how well they are replicating human-like responses, but how we elegantly marry that with the real thing. 

Synthetic audiences have been a hot topic this summer. In some like-for-like trials, there are reports of 95% correlation of responses between real and synthetic audiences. But that 5% could be where the magic lies. Isn’t it the outliers that are often the most interesting? 

A blend of real and synthetic research may be the smart way to harness nuance and scale. Ogilvy Consulting’s 2024 Behavioural Science Annual outlines a tool, The Cognitive Profiler, which uses AI to add further scale to an initial sample of 30,000 surveyed (real) humans. The benefit of having such a large authentic dataset as a start point is the accuracy with which AI can build on it. 

Armale presented a similarly balanced view of empathy and augmentation coming together to ensure we don’t sacrifice human connection in marketing. While it sounds like common sense, some commentators are beginning to predict the end of human-centred research entirely.

Cost, time pressures and access have always been challenges in doing good quality, well-recruited research. AI can be a hugely important addition in removing some of those barriers – and it is an easy step to go from fewer humans to no humans at all. 

As with many evolutionary questions, we need to ask: just because we can, does that mean we should? Is the real win for efficiency here to cut real people out of the process? I think we might find more impact with less consequence elsewhere.

Once people become surplus to requirements in an industry that exists only to engage them, we should question the usefulness of ourselves in the mix too. 

Jo Arden is chief strategy officer at Ogilvy.