Your questions, answered: Generative AI will revolutionise market research

AI?might just be the most boring subject area in?tech . We wanted cyborg pets, robot manservants and sentient dishwashers. What we got was reliable data transfers, better fraud detection and more targeted email marketing. Advances in AI have done less to quicken pulses than many first imagined. Until, that is, last November.

OpenAI, ?a research lab launched a trial version of its ChatGPT tool, an interface that allows users to generate cogent prose by simply punching in a question. Ask it to write an Eminem rap, and it will churn out a reasonable attempt. Request a dairy-free Mac and Cheese recipe, and it will oblige with a set of easy-to-follow steps. The potential of generative AI (the term used to describe such inventions) is staggering. This technology can write your philosophy essay, but it is also likely to disrupt internet search. Right now, we type keywords into Google, and are served with an array of links we must sift through. In the future, we’ll simply ask a question and get an answer. For those of us in the?business ?of?insight ?and?market research , it’s big news.

Slow and expensive vs quick and cheap

The field of market research emerged at the turn of last century and there have been a number of certainties for those practising it (and doing it well) that are still true today. Firstly, it takes a long time. Gathering insights that are telling, true and rigorous is a business of questionnaire design, focus groups, qualitative and quantitative analysis. It’s scientific – and enquiries suffer when rushed. Secondly, it’s expensive. Want to commission a small army of researchers to inform your next move? Not only does this involve a lot of human labour, but the value of knowledge they deliver is enormous - and getting it wrong is disastrous. As David Ogilvy said: “Advertising people who ignore research are as dangerous as generals who ignore decodes of enemy signals.”

In the 2020s brands have had a taste of how quickly markets can shift. If your research study takes a fortnight to produce, there’s an increasing chance that the data may be obsolete by the time you serve it up. The future of insight, therefore, is knowledge delivered in real-time, accessed not through PowerPoint presentations or dossiers, but through interfaces like the one OpenAI has created. You ask a question, and the oracle responds. Slow and expensive, becomes fast and cheap. Crucially, generative AI harnessed in this way will allow companies to ask layered and specific questions. Right now, you can request that ChatGPT write a 200-word summary of An Inspector Calls using Shakespearean-style language. It makes a decent job of this.

Factual fix

In its current stage of advancement, generative AI is proving to be poor on detail. Pundits have spotted factual inaccuracies in the material that ChatGPT churns out. OpenAI CEO Sam Altman has conceded this himself, Tweeting that “ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness. It’s a mistake to be relying on it for anything important right now. It’s a preview of progress; we have lots of work to do on robustness and truthfulness.”?

But when this work is more developed, and the technology is paired with a backend capable of scraping data about sales, sentiment and supply chains, brands will have something approaching a god’s eye view of the world. After that, the hard part – as ever – will be deciding how to act on the knowledge.

Matt Hay is an Evening Standard contributor and the founder and CEO of technology platform Bulbshare