An Alternative Approach to AI in Market Research to Hedge against a Potential AI Bubble Burst

While the world has been very vocal in generative AI, some in our industry have been diligently working away behind the scenes to understand how AI can be used within a market research context. But not just used (as that is anyone’s game), but used in such a way that respects the very tenets of the market research profession. 

At its core, market research is a shared understanding of people, markets and organizations with the highest level of confidence. For those of us who have spent decades developing technology to help deliver insights with that level of confidence, AI is definitely disrupting the process. So, our shared industry challenge is how we can maintain this level of confidence while utilizing what is both currently and potentially possible with AI.

When Dependence Becomes a Risk

One way market research and data analytics technology companies are addressing this is to infuse their platforms with AI. As you would expect and have seen, this has led to both exceptional and questionable results. 

After testing various methods, the direction we have landed on can be described as a “double-prompt” approach. And it’s possibly not quite what you think. In this model, AI does not prompt itself, but instead prompts a dependable, trusted, time-tested analysis and reporting engine. It gives users a set of initial prompts, which in turn automatically prompts the platform, providing both relevant commentary and visualizations. In the technology world this is referred to as orchestration. One of the most significant benefits of using AI to orchestrate, rather than to control, is that users can expect the same level of certainty as they get when working manually.

One of the results of the intensive investment we have seen in AI around the world is that organizations are building market research platforms that are reliant on technology that is dependent on other systems. One big danger of this is the risk associated with the reliance of an external technology partner, like AI, in a platform’s processes. If AI is baked into a platform and there is an AI bubble burst, the analysis platform becomes inoperable.

Preparing for an Uncertain Future

A more resilient approach is to ensure that platforms can still operate in a traditional manner, such as allowing users to drag and drop variables, maintain auditability and achieve repeatable results. This gives market researchers confidence to freely share their work with stakeholders. Should an external technology partner fail, the platform maintains its operability.

Another upside of taking a parallel approach to AI is that organizations can choose to operate with or without artificial intelligence in their environment. With IT departments becoming increasingly careful in their security posture, it is essential that research technology can still be used and can generate the same level of excellence without the presence of AI.  

More advanced users can, of course, choose to take full advantage of artificial intelligence functionality, using conversational or guided features that help them dive into projects quickly and explore what types of analysis might be most meaningful for the data set they are investigating.

When early innovators in our field (like Infotools back in 1990) set out to disrupt the market decades ago and give client-side researchers access to their own data, and make their own analyses and not be so reliant on third parties to cut and supply data, we couldn’t have imagined this level of functionality. It’s an exciting step for us as an industry, both on the agency and client side. 

I don’t wish an AI bubble burst on anyone, but if one does happen, those in the insights industry can prepare themselves if they balance innovation with independence, and by ensuring that tools and methods remain rooted in the principles that have long defined sound research.