A groundbreaking study by the University of Sheffield has revealed that bees can distinguish complex visual patterns through their flight movements — an ability that could inspire the next generation of artificial intelligence (AI) and robotics.
Scientists believe this discovery could revolutionise AI development, enabling future robots to become smarter, more adaptive, and vastly more energy-efficient.
A Tiny Brain Solving Complex Tasks
Lead researcher Dr Hadi Maboudi explained in an interview with Al Jazeera Net that complex visual patterns are images made up of multiple elements — shapes, colours, textures, and arrangements — that are difficult to identify at a glance.
“Recognising a simple flower with a few petals is relatively easy,” he said. “But recognising a flower with many petals, coloured stigma, detailed sepals and leaves, or spotting a predator hidden among harmless objects is far more challenging.”
Bees overcome the physical limits of their small compound eyes by actively moving. Through micro-adjustments in flight and head position, bees transform static scenes into dynamic, three-dimensional sequences. These movements generate rich streams of visual data that their small brains process with astonishing accuracy.
“This combination of motion and perception shows that even animals with very small brains can solve complex recognition tasks,” Maboudi noted, calling it a principle that could inspire more flexible and efficient AI systems.
Bees See the World Differently
Unlike cameras or current AI systems that passively record images, bees actively move their eyes, head, and body to rapidly alter their visual field.
“These movements allow bees to capture far more detail than the physical resolution of their eyes would normally allow,” Maboudi said. Remarkably, they achieve this with a brain no larger than a sesame seed — containing fewer than one million neurons — compared to the human brain’s billions.
Equally striking is their efficiency: bees perform such complex tasks using only fractions of a watt of power, whereas today’s AI systems consume massive amounts of energy for far simpler recognition tasks.
Bees can navigate more than 10 kilometres, remember countless flowers and landmarks, and return to their hive with precision. They even communicate this knowledge to other bees through the famous “waggle dance.”
Lessons for the Next Generation of AI
If bees were robots, they would be highly agile and adaptive — avoiding moving obstacles and flying smoothly through unpredictable environments, all with a brain smaller than a pinhead.
Current AI systems fall far short of this. But by mimicking nature’s strategies, Maboudi argues, AI and robotics could shift from passive data processors to active explorers, gathering information more efficiently and adapting to dynamic real-world environments.
“This could lead to smarter, lighter, and more energy-efficient machines,” he explained. “They would be capable of navigation and pattern recognition even in noisy or constantly changing conditions.”
Intelligence Emerges From Interaction
One of the key conclusions of the Sheffield study is that intelligence is not just in the brain — it emerges from the continuous interaction between brain, body, and environment.
“The brain relies on body movements — whether fast or slow — to convert raw sensory signals into useful information,” Maboudi said. “The environment then responds to those movements, providing new inputs that the brain uses to guide the next action. It creates a closed loop where perception and action cannot be separated.”
For example, when a bee recognises a flower, it doesn’t rely on a still snapshot. It flies around it, tilts its head, and changes angles of view. By doing so, it identifies a few key visual features that make the flower unique. On later visits, the bee quickly finds these features without needing to memorise every detail. This makes recognition and memory far more efficient, allowing a tiny brain to solve complex problems with limited resources.
From Bees to Smarter Machines
“The lesson for AI and robotics is clear,” Maboudi concluded. “Machines must not remain passive processors. They should be designed as active systems operating in closed loops with their environments. Such systems would adapt to new tasks and settings without endless retraining, applying flexible rules across many different situations.”
Instead of building ever-larger AI models that demand massive datasets and energy consumption, future AI could become simpler, more dynamic, and closer to the flexible intelligence seen in living creatures.
One Ummah. One platform. One mission.
Your support keeps it alive.
Click here to Donate & Fund your Islamic Independent Platform
