NEWS14 March 2017

How far should automation go – and will it put human researchers out of a job?

AI Data analytics Impact 2017 Innovations Mobile News Privacy UK

UK - The appeal of automating research is clear: it’s often faster and cheaper than a human researcher. But as machines become more intelligent, will there still be a role for real people, and for the old-fashioned business of asking questions?


“Can automation coexist with elicitation?” was the question posed to a panel of speakers at the Impact 2017 conference.

The value – and the limitations – of automated research were clear. Steve Phillips, introduced as the ‘arch druid of automation’, but officially the CEO of Zappistore, said automation’s ability to make research faster, cheaper and in some ways better than other methods of getting answers could be a huge opportunity for the market research to expand. 

It also enabled researchers to reach consumers at times that were most conducive to getting sensible answers. Asking people at an evening focus group about breakfast cereal, for instance, was unlikely to get the same kind of responses as a mobile survey done in the morning.

Panellists had a range of views on how best to define elicitation, from asking a question that requires the construction of an answer, to providing any kind of stimulus that causes a response.

But there was agreement that the two weren’t necessarily at odds. “I don’t think elicitation and automation are necessarily in conflict,” said Anjali Puri, global head of qualitative research at Kantar TNS.

“Elicitation can be automated. The fact we can have a mobile phone prompting people for a response is elicitation, creating responses in an automated way.”

The ability to carry out qualitative research at scale was one of the great appeals of automation, Puri said. “But we’ve not fully made the most of that yet.”

Panellists also agreed there were key elements of research that, at least for now, couldn’t be replicated by automated research tools and artificial intelligence.

Pyschologist and author Oliver James drew parallels with the intelligence services’ hunt for potential terrorists; access to vast amounts of citizens’ personal data didn’t necessarily make the task of sorting the unsavoury from the truly dangerous a great deal easier. That task still relied largely on human intervention.

And Phillips said that while machines were very good at finding patterns in data that could tell a story – and do it faster than people – they were very poor at telling those stories. Intuition and persuasion were, for the moment, still human qualities.

The decision on whether to automate a project depended not just on the client but on the specific task at hand. Which of two products or advertisements would be more popular with Russian housewives, for instance, was a question that could be answered through automated research in a matter of hours. But finding out what kind of new product would fit into the everyday lives of those same housewives would be better done in the flesh by a person who could not just elicit information but also soak up all kinds of other information from a person’s home.

The days of artificial intelligence being able to replicate that level of learning are some distance away, Puri said. “You absorb information that you can’t even begin to teach a computer because you don’t know how it happens.”

Distinctions were drawn between online research that simply takes traditional approaches online, and the new world of social media research. Dr Mariann Hardey, director of iARC’ and a lecturer at Durham University business school, said social media was a rich resource offering unprecedented access to individuals and their data. But access to that data carried with it significant responsibility.

While automation has already taken many data-gathering jobs from human researchers, higher-value jobs – those that require uniquely human attributes – are safe for the foreseeable future, Puri said.

“I wonder whether it’s not the jobs going (away) but the nature of the jobs that are changing. A lot of us are having to learn new skills … but I can’t really see those jobs going away.”