As AI permeates more areas of business, there is legitimate concern over whether it will eventually take jobs from people. It’s been a big area of debate in the research space, where I’ve worked for the past 15 years.
Recently, I was invited to speak at the Board of Innovations Autonomous Innovation Summit, talking about the future of AI in market research. What I shared with attendees was probably not what they expected: I believe that the use of AI will make the future of research more human-led than ever. And that’s because AI will help researchers increase the volume of research they can do, by an order of magnitude, bringing more human inputs to business decision making.
A CPG Case Study
Earlier this year, I had the opportunity to work with a global CPG company—one of the world’s largest—on a research project. All CPG companies leverage consumer insights throughout their R&D pipeline to decide which projects or products to fund. Business and brand leaders need increasing levels of confidence as they pour resources into new products while designers and engineers need consumer input to drive design decisions. Each stage-gate has its own processes, methods and tools, and as a result, it’s infamously slow.
In this case though, they wanted to test 15 different product concepts across three markets, which varied widely in terms of culture and languages, all at once. The leaders wanted to move faster and more confidently in this category to keep up with competition. In other words, they wanted to speedrun their stage-gate process. They were open to leveraging AI to get this done.
Here’s what happened:
- An AI system was able to conduct 360 consumer interviews over four days—essentially a long weekend—and spent a total of 20 consulting hours on the research.
- They estimate that if they had taken the traditional approach, it would’ve taken the team over two weeks to interview just 24 consumers with 140 consulting hours spent on the research.
As a result, they not only saved time, but achieved the thing researchers have always wanted with qualitative insights: scale. They tested the concepts in multiple countries, with a real revenue opportunity in each! And they developed real confidence in the insights.
So why does this mean that AI will make research more human than ever? AI will fundamentally reshape the sheer scale with which we can learn and internalize real human experience. That’s what you see a glimpse of in this project. But this is only the beginning. As you play out the future of research with the advent of large language models, it’s easy to see how all research will be reshaped to become more human-driven.
At a basic level, AI is decreasing the ‘cost per insight.’ That’s not a future prediction—it’s happening right now, from using AI to synthesize unstructured data or to collect data at a speed that was never possible.
But driving costs down does not mean we simply pay less for the same research we do today. Just because compute got cheaper in the last 50 years doesn’t mean we’re all using mainframes for less cost. Instead, lower cost for compute enabled an exponential boom in compute everywhere pushing new innovations, like the cloud, smartphones and, yes, AI.
What’s the big takeaway here? AI will unlock new and different ways of gathering insights, and it will happen in four significant ways:
- Surveys will evolve to become more qualitative. Surveys are a product of the limitations of deterministic systems—participants must choose between predefined options. But why should massive surveys be limited in that way? AI-moderated research tools can add conversational, qualitative exercises to surveys that are infinitely dynamic. Imagine collecting rich, contextual data at the scale of all quant research.
- Today’s qualitative research will massively expand. Why stick to a sample size of 10? Using AI to moderate means that there’s no structural limit, so researchers and stakeholders will begin to expect larger, more diverse sample sizes to make smart decisions. AI will be running massive focus groups, in-depth interviews, co-creation sessions, diary studies and probably new methods that don’t even exist yet.
- Analysis will never be a blocker again. LLMs (large language models) are built to understand and organize unstructured qualitative data. Over the next five years, these systems will be incredible tools for all kinds of quant/qual analysis. We’re seeing pieces of that start coming together already, but we are still limited by compute (i.e. you can’t pass 1,000 in-depth interview transcripts into ChatGPT in one prompt…yet)—a resource that is becoming more abundant every day.
- And lastly, imagine you had an infinite number of researchers on your team… Where would you deploy them? Maybe you’d have them observe shoppers down a grocery aisle, or join thousands of support calls, or even accompany people in their real-world experiences. This is where we will emerge with new methods that gather more human insights than ever before.
Isn’t this the future we want? More humanity injected into the products, services and experiences we encounter every day. More understanding of the challenges we all face. More empathy in everything we build.
I strongly believe AI will make research faster, bigger and more ubiquitous. It will unlock types of research that were never possible before (or that we can’t even imagine yet!), all orchestrated by human researchers at the center leveraging an increasingly powerful set of AI tools. Deeper and more human input will lead us to build more human-centered innovation, and ultimately, this will make everything we build MORE human.
Aaron Cannon is Cofounder and CEO of AI platform Outset.
Network
