At its core, artificial intelligence (AI) is a technology that lets computers think like a human. This is accomplished by creating artificial neuron pathways for data processing and memory. While most of today’s AI systems try to replicate the human brain, none are anywhere close to doing so.
James Marshall, a professor of computer science at the University of Sheffield is baffled by this trend. He believes that the future of AI needs to start with smaller ideas—literally.
His startup, Opteran Technologies, has just raised $2.8 million to build AI systems modeled on honeybee brains. The team argues that this is a more logical starting point for the technology and could yield more immediate results.
Marshall told Digital Trends, “If you’re going to start building a model of any brain on the planet, why on Earth would you start with the most complicated one?”
In the tech world, big ideas are usually what earn the biggest payday. However, it’s clear that thinking on a large scale is slowing down the process of AI development. Marshall and his team believe that focusing on smaller, more attainable goals is the key to eventual large-scale success.
That’s an interesting theory considering the fact that many of today’s most sophisticated AI systems already have more artificial neurons than an average honeybee. Even so, the general intelligence of those programs is nowhere close to that of a bee.
Despite the fact that many people look down on insects, honeybees can be incredibly smart. Marshall notes that bees are, “consummate visual navigators, [adept at] long-distance navigation, with very sophisticated learning abilities.”
He adds, “They’re much more than the simple kind of reactive automata that people often think insects are. Individually, they’re very clever.”
That isn’t just a hypothesis. Data has already proven that honeybees are good problem solvers. Researchers from the University of London found that bees solved the traveling salesman problem faster than the world’s top supercomputers. The test requires the bee or computer to find the shortest route between flowers that are discovered in a random order.
Ultimately, this means that honeybee-based artificial intelligence systems aren’t a token achievement. David Rajan, CEO of Opteran, says, “Building a honeybee brain in silicon could therefore help develop sophisticated navigation tools that could be lightweight, ultra-low-powered, and orders of magnitude more efficient than the deep learning approaches.”
Most of today’s deep-learning systems focus on an approach pioneered by the brain’s visual cortex. In other words, the system uses cameras to identify something and then processes that information into a useful database that it can refer back to later.
The brain—as a whole—doesn’t think like that. Marshall notes, “When you look at a complete brain, it is highly structured. You have different brain regions that do different things, that are internally structured in different ways, with well-defined connections between them.”
In that sense, Opteran’s honeybee-inspired AI is different from other types on the market. Rather than trying to create a new type of thinking, it is designed to think naturally—just like a biological brain.
Marshall says, “Having a million neurons and however many synapses isn’t the end of the story; it’s about how you connect them together.”
He adds, “It’s also about the kind of information processing that’s done at the neuron level, because there’s more than one kind of neuron in the real brain, although there’s often only one neuron type in a deep net.”
Ultimately, Opteran’s approach uses a variety of neuron types to create more efficient processing. That gives it an advantage in the power arena as it requires fewer computing resources than typical deep-learning tools.
This makes it easier for researchers to access the technology and start using it “out of the box.”
The latter is something that Opteran is focused on. It has promised that no additional training will be required for those who want to use the system. As an added bonus, the honeybee-inspired AI has transparent rules.
This addresses a major issue with AI in general, which is that researchers often don’t know how the system makes decisions.
Ready to Roll
Thanks to its “start small” mindset, Opteran is already getting close to rolling out its first commercial tool. The startup hopes to have a product on the market within the next 18 months. Its first applications will focus on things like obstacle avoidance and autonomous decision making.
Paired with Opteran See, a 360-degree camera, the algorithms could become a key component for applications like autonomous driving. With a huge push by a number of companies in that space right now, Opteran is poised to get in on the success.
That being said, there are still plenty of challenges for the startup to face. For one, there are a lot of things about the brain that researchers don’t know. The knowledge gap makes it difficult to reverse engineer certain traits.
Fortunately, Opteran doesn’t think this will be a problem. Marshall says, “What we really care about commercially is behavior, the competency of the system. As a business, we’re not fixated on saying we’re confident we’ve reproduced the way the honeybee works. [Instead, we want to say] we are confident that we have reproduced a system which is behaviorally robust, and which seems to us to behave as if it was a honeybee acting like a honeybee.”
Fans of Alan Turing might find that statement eerily familiar. It is a reference back to Turing’s definition of AI being something that can fool a human into thinking they are talking to another human.
Although there may not be a Turing test for bee-based AI systems, it will be interesting to see where this technology goes in the future. Indeed, Opteran’s approach could become a mainstream part of the AI sector. If more companies start thinking at a smaller level, perhaps it will one day be possible to create an AI that thinks like the human brain.