Your brain lies to you, and AI falls for it, too

Our eyes don’t always show us reality as it is, and some AI systems are fooled by the same visual tricks.

That overlap is helping scientists learn more about how human perception works.

Optical illusions, like the Moon appearing larger near the horizon, show that the brain doesn’t process every detail it sees.

Instead, it relies on shortcuts to quickly make sense of the world.

These shortcuts usually work well, but they can also lead us astray.

AI vision systems are built to detect fine details humans miss, which is why they perform so well in areas like medical imaging.

Yet researchers have found that some deep neural networks also misinterpret certain optical illusions, particularly those involving motion or perspective.

In experiments, scientists tested an AI model designed to predict what it expects to see next, similar to how the brain uses past experience to anticipate visual input.

When shown motion-based illusions, the AI was tricked in the same way humans are, suggesting that both rely on predictive assumptions.

That said, AI still sees differently.

Humans can reduce some illusions by focusing attention, while AI processes the entire image at once.

This highlights how close, and how far, artificial vision remains from human perception.

Some researchers are now using unconventional approaches, including ideas from quantum physics, to model how humans switch between different interpretations of ambiguous images.

This work could help explain how perception changes in unusual environments, such as space, where astronauts experience visual illusions differently due to the absence of gravity.

So is the dress black and blue or white and gold goddamit - MV