This optical illusion is so confusing it even blows AI's mind
Published on Jan 30, 2026 at 12:12 AM (UTC+4)
by Henry Kelsall
Last updated on Jan 29, 2026 at 5:10 PM (UTC+4)
Edited by
Kate Bain
There is seemingly very little that surprises artificial intelligence, but even AI had its mind blown by this optical illusion.
Experts and researchers have been training artificial intelligence to create optical illusions.
The idea is to see how it interprets them and whether it experiences them in the same way that we humans do.
And as it turns out, one optical illusion in particular left the AI completely baffled.
DISCOVER OUR SUPERCAR AUCTION SITE – View live auctions on SBX Cars
The optical illusion that baffled AI
Some of the research has been conducted by Eiji Watanabe, an associate professor of neurophysiology at the National Institute for Basic Biology in Japan.
His team has been looking at the deep neural networks (DNNs) behind AI algorithms and how they respond.
Watanabe and his team presented the DNN of an AI with static images that look like they’re moving.
One way in which they did this was with Akiyoshi Kitaoka’s famous rotating snakes illusions.
These are formed of concentric rings in various colors and segments.
They’re still images, but our brains view them as moving.
Click the star icon next to supercarblondie.com in Google Search to stay ahead of the curve on the latest and greatest supercars, hypercars, and ground-breaking technology
How did AI’s algorithms respond to the circles?
Watanabe’s team used PredNet, a DNN used to predict future frames in a video.
It does this via knowledge acquired from previous ones.
Videos of natural landscapes that we tend to see were used to teach the DNN.
However, illusions were not a part of it.
Various versions of the rotating snakes illusion were shown to PredNet.
A version that doesn’t trick human brains was also shown to the AI.
As it transpired, the model was actually fooled by the same images that tick humans.
This supported Watanabe’s theory that humans first predict what we expect to see via our visual system.
Any discrepancies are then processed with the actual visual input.
The DNN was unable to freeze a particular circle from the illusion as we can.
Every circle was seen as moving when PredNet viewed it.
As remarkable as artificial intelligence is, it is incredible that a simple illusion blew its mind and tricked it like this.
DISCOVER SBX CARS: The global premium car auction platform powered by Supercar Blondie
Henry joined the Supercar Blondie team in February 2025, and since then has covered a wide array of topics ranging from EVs, American barn finds, and the odd Cold War jet. He’s combined his passion for cars with his keen interest in motorsport and his side hustle as a volunteer steam locomotive fireman at a leading heritage steam railway in England.
