New research shows that honey bees could impact the future of AI (Artificial Intelligence).
Scientists at the University of Sheffield have discovered how honeybees make swift and precise decisions. This finding that could influence the design of advanced robots and autonomous machines. The study published in the eLife journal was led by Dr HaDi MaBouDi from the Department of Computer Science at the University of Sheffield and Professor Andrew Barron from the Macquarie University in Sydney. It revealed the intricate strategies bees employ when deciding which flowers to investigate for nectar.
Better Insight into Honey Bees´Brain Function and Evolution
The research enhances our knowledge of honeybee brain function and evolution. It also inspires the development of AI robots capable of bee-like decision-making: quick, accurate, and efficient.
Honey bees need to quickly and accurately identify nectar-rich flowers for survival. The researchers studied these decision-making processes, focusing on the speed and precision of flower selection or rejection in a controlled flight arena with varied reward and punishment stimuli.
The study involved training 20 bees to identify five different colored artificial flowers. Researchers then observed their performance in a custom garden where flowers only contained distilled water. The researchers tracked the bees’ paths and recorded the time they required to make a decision.
In the experiment, honey bees quickly decided to land on those flowers they were confident had food. If the bees knew a flower didn’t have food, their decision was equally swift. A computer model replicating the bees’ decision-making process was then created. Interestingly, the model’s structure closely resembled a honeybee brain’s physical layout.
Honey bees’ decision-making Comparable to Primates
It revealed that the sophistication of honey bees’ decision-making is comparable to primates. Their choices depend on both the quality and reliability of available information. They tend to make more accurate decisions when accepting flowers than rejecting them, and these decisions are influenced by changes in evidence and reward probability.
Interestingly, quicker acceptance decisions are more likely to be correct, which is a pattern also observed in primates. This implies that the decision threshold varies dynamically with the sampling time.
Dr MaBouDi highlighted that despite their tiny brains, bees excel at making quick and accurate decisions based on subtle variations in color or odor. These findings can now be used to design better robots that think like bees – some of the most efficient navigators in nature.
Professor Barron added that even though a honeybee’s brain is smaller than a sesame seed, it can make faster and more accurate decisions than humans. A robot mimicking a bee would require the support of a supercomputer.
To further understand the minimum circuitry needed for such decision-making abilities, the researchers created a unique model of reverse engineering insect brains, such as honeybees, to design next-gen autonomous technology. This model aligns with known insect brain pathways and is biologically plausible. It suggests a AI robust system for autonomous decision-making, potentially applicable in robotics.
Opteran, a spinoff company, is developing low-cost silicon brains enabling robots and autonomous vehicles to see, sense, navigate, and decide like insects. This approach, known as Natural Intelligence, could significantly expand the market for autonomous machines and robotics.
Professor Marshall, a participant of the study, commented,
“Our research has demonstrated how bees are capable of complex autonomous decision making with minimal neural circuitry. Millions of years of evolution has led bees to have incredibly efficient brains with very low power requirements. This biology can inspire the future of AI.”