Capturing Human Data in the Era of AI: Why Eye Tracking Still Matters

James Breeze, founder of Objective Eye Tracking and Tobii's reseller in Australia and New Zealand since 2007, has written a reflective piece on the Tobii blog about what eye tracking is actually for — and why he thinks it becomes more important, not less, as AI-generated insight becomes cheap and abundant.

The Facebook study

The anecdote at the centre of the article is a good one. Sixty participants each spent ten minutes scrolling Facebook on their phones exactly as they normally would. Afterwards they were asked whether they had looked at any ads. Every single one said no.

Then the eye tracking replays were shown. Every participant had looked at the ads. Faced with their own recordings, people explained precisely why they had paused — the colours, the images, the headlines. They had not been lying. They simply did not remember.

Breeze's point for anyone commissioning research: without eye tracking, every participant in that study would have sworn they ignored the ads. Self-reported data on attention is not reliable, and the gap between what people say and what they do is not a rounding error.

A branch in Singapore

A second example comes from a DBS branch at Marina Bay Financial Centre, where the team studied how customers experienced the space — merchandising, queuing, the lot. Most participants were domestic workers and construction staff, initially hesitant about being interviewed. Once the eye tracker went on, their attention moved away from the interviewer and onto their own experience.

Almost everyone looked at the TV screen first. It played on a four-minute loop. After four minutes, nobody looked again.

That single finding reshaped the bank's investment strategy. Seeing that static video loops lost impact quickly gave DBS the evidence to justify producing higher-quality content and building systems to stream fresh video across branches.

What COVID changed, and what it did not

Remote testing — including webcam-based eye tracking — grew rapidly when face-to-face work stopped. Breeze is positive about it: webcam studies work well for static images and digital experiences, and they make it straightforward to scale research across markets.

But he argues the period also clarified what remote methods cannot do. Standing in a store, waiting in a bank queue, sitting in a car, playing a game in real time — these are moments that cannot be simulated. He reports a resurgence of in-person work post-COVID, concentrated in real-world contexts: retail wayfinding, service design, wearable eye trackers in live environments.

AI and eye tracking, not AI or eye tracking

The core argument is not anti-AI. Breeze uses large language models to scan large volumes of data, conversations and cultural signals, which gives breadth — patterns and themes that would otherwise stay hidden. What AI cannot do, he argues, is explain why people see the world the way they do.

His framing: eye tracking on its own can be slow, AI on its own can be shallow. Used together they give both breadth and depth — showing how people use things in real contexts and what captures attention before they are consciously aware of it. He reports the combination has already helped organisations redesign products, prioritise investment and surface mismatches between what customers say and what they do.

Method over data

The last point is the one worth carrying into your own study design. The way a study is built shapes the results, influences how participants behave, and frames the story that comes out at the end. Breeze cites the famous Nielsen F-pattern as an example of both the power and the risk — a single heat map that shaped how websites were designed and where advertising money went. Its strength was never the image itself, but the frame and the story around it. Taken out of context, the same image misleads.

NBT represents Tobii in Israel, including Tobii Pro Spectrum, Tobii Pro Fusion, Tobii Pro Glasses 3 and Tobii Pro Lab. Contact us to discuss an eye tracking setup for your research.

Source: Tobii blog.