An eye tracker that doesn't need humans
Millions of years of human eye evolution and 39 people aged 18-50 years old, who thought they were doing one experiment, when they really were doing another. All in one small Mac app.
If you keep optimizing a website or an ad, at one point you most likely end up wanting to do an eye-tracking test. We're exposed to tens of thousands commercial messages every day, so every millisecond counts for the ads businesses. An eye-tracking test easily sets you back thousands of dollars, and if you want to own your own eye-tracker, it's not uncommon to shell out $20,000 or more.
But our eyes are not that unpredictable. What changes from study to study is what is being studied. So the big question for AI researchers is this: can we teach a computer where a human eye is most likely to look?
Such a model, if possible, would be useful for a wide variety of things. You could crop images without cutting out the main subject of the photos, be they faces, signs, or a flower. Cameras could auto-focus on what's important. And as you may know, this is in fact the world we live in today. So how did they make that happen?
The mother study
A group of researchers at MIT invited 39 people, carefully selected to span 18 to 50 years old. They gave them a coffee, then showed them 300 images, 3 seconds at a time. They were told they were doing a memory test, but they really were doing an eye tracking test.
It took 15 minutes per participant, that's it. Next step: make sense of the data. Come up with an algorithm, an AI model, that could take a new and unseen image and predict eye movements. This is now a field of study, and the researchers all over the world are chipping in, comparing their results on a global leaderboard.
An accuracy in the high eighties might not sound perfect, but consider the alternative: Buy a 20,000 machine, get it shipped, learn how to use it, recruit 39 people, book a room, find someone who can take their calls, cancelations, brew coffee, offer coffee, brief them on the study. Repeat most steps 39 times. And when it comes to cropping millions of photos, each in less than a second, there's just no alternative at all.
Millions of years of data
Most people consider 39 participants to be the sweet spot in eye tracking. We're simply not that different when it comes to how our eyes are wired. We like contrasts, faces, and things that look weird. In other words, hand-me-down experience from thousands of generations of humans before us. The stuff that helped us stay alive when we were facing dangers like elephants and thunder.
The data this kind of study collects is what researches call bottom-up. It's almost purely biological, and almost completely free of cultural influences. A regular eye-tracking test would produce a mix of the two, and this is an important part of the explanation of the 80% accuracy.
But, you might agree that 80% is more than enough to capture the big picture. And that's what this app does.