David Colburn

The holy grail for artificial intelligence computing would be to one day have computers that could think exactly like humans, but the complexities of the human brain exceed our current understanding of how it exactly works.
Modern day imaging processes have greatly expanded our knowledge of the inner workings of the brain. Firing neurons need more oxygen than dormant ones, and an MRI-based scan can track blood flow to active regions of the brain, pinpointing areas involved in various activities such as movement and cognition. Positron emission tomography uses radioactive tracers to study brain metabolism and neurotransmitter activity, yielding significant findings for brain functions in patients with Alzheimer’s and Parkinson’s diseases. Scanning the magnetic signatures of neuron electrical activity, using near-infrared light to measure changes in blood oxygenation, and mapping the diffusion of water molecules along neural pathways all provided fascinating and useful insight into how the brain works.
But with all of that, programming for artificial intelligence that can accurately mimic the brain’s 82 million neurons, remains elusive, likely for decades to come.
However, scientists in the world of AI are thinking smaller right now. A big human brain with all of its complexity may be too much to bite off, but what about tiny brains that do specific things really, really well?
Enter insects. Yep, that’s right, bugs. In standard computer programming, bugs are things to avoid, but when it comes to AI, the next big steps are coming by trying to analyze the tiny brains of insects and mimic their remarkable skills in ways that will benefit human-focused tasks.
Take bees. They’ve been navigating through complex environments far from home and returning with ease for millennia using brains about the size of a pinhead. Driverless cars need navigation systems that can do the same thing, so why not look at a reliable navigation system that’s been around for millions of years that does what we need our navigation systems to do?
And that’s what scientists have done. They’ve glued radar tags to bees’ backs with tags to track and analyze their flight patterns to develop algorithms that process optical data in a similar way. They’ve put bees in stationary harnesses and shown them movies simulating complex flight patterns while monitoring their tiny brain activity to understand how they process visual information and make mental maps based on environmental cues. Bees can accomplish their amazing feats of navigation without the use of GPS – can human-created robots do the same?
The answer is yes, they can. Using the information gleaned from studying bee brains, specialized processing chips that mimic the function of bee brains have been developed, and coupled with machine learning, a form of artificial intelligence, we now have guidance systems for drones that don’t need to depend on satellite location data to know where they are and to get around. This autonomous navigation ability has far more uses than simply getting your driverless Tesla safely from point A to point B.
Bees operate in swarms, and so, too, can autonomous drones operating with artificial intelligence. You’ve likely already seen drone swarms in operation as entertaining replacements for fireworks displays. But AI drone swarms can be put to agricultural use, planting seeds, identifying disease outbreaks and applying crop treatments. They can be used to locate missing people or deliver emergency supplies in natural disasters or track and control the spread of wildfires. They’re also being studied for use in security for large industrial complexes.
And because they’re a human creation, drone swarms are also being developed for military use. Imagine a swarm of hundreds or even thousands of AI-driven military drones, some specialized for scouting, some for attack, able to adapt their mission on the fly to assessed conditions. Drones are already an integral part of the Ukraine-Russia conflict, used for surveillance and to deliver small explosives cheaply and accurately, but nothing like what will soon be possible.
It may well be up to another insect, dragonflies, to develop effective counter measures, both for drone swarms and ballistic missiles. What’s the specialized capability of dragonflies? Intercepting things in flight. Whether pursuing zooming fruit flies or mosquitoes, dragonflies have a 95 percent success rate in capturing their prey, making them one of the world’s top predators. A dragonfly’s reaction time to flying prey is 50 milliseconds, six times faster than the blink of an eye. That’s only enough time for about three neurons to fire. Creating algorithms and neural networks based on this simplified architecture, work currently being done at Sandia National Laboratories, could be the key to developing effective defense systems against the weapons of the future.
Given the immense power of the human mind, it is somewhat ironic to think that the simple minds of bugs may lead to our destruction or our salvation. Mosquitoes don’t seem nearly as bad by comparison.