5 FEB 2025 | 5 min read

People tend to gravitate toward one of two camps when discussing the impact and implications of artificial intelligence. There are those who are eager to embrace the technology and the benefits it poses to deliver, and then there are those who feel acutely allergic to its perceived downsides.

I’ve spent years exploring and measuring public perception, sifting through these complex and divisive emotions to iron out and contextualize a more fulsome picture of what’s true. As with many topics, the truth around artificial intelligence sits somewhere in the middle of society’s polarization — and I say this having first-hand experience with the very best and worst of this emerging technology.

Measuring genuine public perception is a complicated blend of art and science, an endeavor requiring careful navigation through bias, interpretation, distortion, and unpredictable noise. This has always been the case, even before our increasingly digitized society changed the speed and means by which we collect and discern opinion. That said, our work has become harder to carry out — even as it has arguably become more important than ever to produce.

For business leaders looking to set strategic priorities and investments, having an accurate idea of public opinion and trends is critical. Any cloudiness around these areas can introduce risk and costly mistakes. In today's modern digital landscape, where bots and misinformation pose new challenges and cast long shadows, it has become increasingly difficult to collect, analyze, and understand genuine public perception. Traditional methods struggle against a static-filled backdrop.

In 2023, a Stanford University communications scholar, Jeff Hancock, posed a question that gets right at the heart of the matter. In a world in which we see a proliferation of AI-powered tools that mimic human natural language abilities, what does it mean to be truthful and authentic? And how do you strike a balance when discerning trust from deception?

To understand and illustrate how we understand public perception and what challenges we now face when gathering it, I sometimes turn to the ancient allegory about shadows cast on a cave wall.

In that story, a group of prisoners being held in a cave are only positioned to see shadows cast on its walls by objects passing before a campfire that’s blazing behind them. For the prisoners, the shadows represent reality — they can only make guesses about the actual objects — with all their abstractions and shape-shifting mystery — or the world beyond the cave. When one prisoner manages to break free, only then does that individual realize the shadows are mere illusions compared to the fuller reality of the world. When they return to the cave, they struggle to convince the others of the true reality because they cling to their familiar but limited perceptions.

For those of us tasked with discerning genuine public perception in an era of AI, the shadows on the wall represent the distorted data generated by bad actors, bots, and misinformation campaigns, all of which emit illusory signals that masquerade as public opinion. Our role is to free ourselves of that context, to understand the broader picture and to see beyond the shadows.

It's hard work, and it typically involves triangulating our way through three broad challenges every single day.

The first is the imperative to fight for eyeballs, which is increasingly difficult. It is an activity that gained traction when most people’s primary access to the internet happened by sitting down at a desktop computer, back when asking for 30 minutes of someone’s time to fill out a survey wasn’t as fraught. The shift to mobile and online surveys shifted our work dramatically. Now you have five minutes or less of someone’s time and it needs to deliver an engaging and rewarding experience for participants to deserve a share of peoples’ time in today’s always-on world that’s saturated with over 8.9m mobile apps vying for attention.

The second challenge is all about combatting fraudsters and bad actors. Many firms working in the market intelligence industry have been loath to admit that AI is making it very difficult to carry out our work. Fraudulent activity is on the rise, particularly finding its way into survey samples, the percentage of which that’s being rejected for low quality skyrocketing by about 300% in recent years. Not only is this becoming more common, it’s also becoming increasingly sophisticated thanks to automation and AI. In some respect, one could argue that online fraudsters looking to distort public perception have become industrialized.

Thirdly, there is data privacy legislation. As a global business, you must be compliant in all markets. Still with no one-size-fits-all approach and the landscape of policy shifting so rapidly, it can be increasingly difficult to keep pace with the hard work required to assess public opinion.

The good news however is that, for as much as AI poses unique challenges to conducting the business of figuring out public perception, it can also light the path forward. You just have to be bold enough to step outside the proverbial cave.

In an era of AI, emerging technologies are being used to sift through the onslaught of noise, identifying patterns and authentic voices. When well-executed, AI and emerging technology tools are helping to provide a net-good to society and will prove critical as we increasingly seek to pierce through the static and discern with confidence public perception.

It will take time for many firms to realize the full potential of AI in assisting in this capacity. I’ve learned that in my field you typically need to train AI and algorithms on critical datasets for about 1.5 years before you can deploy it to find truth, and then an ongoing, everyday commitment to feeding these algorithms with real-time data to ensure they are continually learning and adapting to be representative of real people in a real world context. Secondly, you must ensure your talent can work comfortably with these tools. You can have the best AI in the world, but you need to strike a balance between it and your human talent to get the best results. Without that effective balance between machine and human collaboration, you can develop great models, but at some point it won’t be able to depict the real world. A strong AI governance model is key to achieving that.

As part of my work, we are in a process of deep experimentation on investigating the concept of ‘synthetic panels’ and more specifically ‘digital twin technology’. The concept of compliantly building ‘look-alike’ avatars of real people who are who they say they are is complex, but the potential opportunities these perpetually engaged personas can offer in terms of more robustly representing niche target groups, mitigating fraudsters, and driving innovation faster is truly exciting - as long as we can regularly check-in with the real humans they represent to ensure their avatar evolves with them.

Imagine creating virtual models of populations and behaviors that reflect real-world dynamics, gathering data from a variety of sources — social media, surveys, transaction records, etc. — to simulate how people think, feel, and act. When we run scenarios and test hypotheses in this virtual environment, we’ve found that we can use AI to uncover patterns, predict trends, and identify genuine public sentiment without having to re-survey people or rely on distorted data. This can help cut through noise and work around bots and fake accounts to offer more accurate, data-driven insights into how people are thinking. Yes, we must be ever vigilante and always prepared to react to bad actors as they attempt to thwart our work. But, learning from how we are already leveraging AI to block more than 2k fraudulent panelists joining our proprietary panels every day and to prevent 6.6 times more fraud than the overall industry average, we do already know that using AI to beat fraudsters at their own game is working.

The job of collecting and sharing public opinion will always be an evolving pursuit, a response to technology and trends that are constantly reshaping the world in which we live. Keeping up with that change will require careful attention to data, with particular care as to how it’s organized and used by emerging technologies, including AI. Being tech ready for what comes next is just as much about the tools you have in hand as it is about the people who’re behind them as we all have the choice to either be rivalled by the AI bulldozer or drive it . So, being ‘ready’ in this space is a continuous process of learning and adapting.

But if you’re willing to step outside the cave, beyond your comfort zone, we’ll together find ways to cut through the noise and light the path forward through the many interesting and thorny challenges that lie ahead.