Nara Logics’ new AI engine uses recent discoveries in neuroscience to replicate brain structure and function. The research is part of a decades-long quest to make computers that can “think” as well as or better than humans. Simulating brain function is one promising approach. “There are obvious benefits to copying what seems to work in biology and implementing them in machines to aid automated decision making in a broad spectrum of daily activities,” Stephen T.C. Wong, a computer science professor at Houston Methodist Research Institute, said in an email interview. The uses for humanlike AI could range “from playing chess, recognizing faces, and trading stocks to making a medical diagnosis, driving autonomous vehicles, and engaging business negotiations or even legal litigation,” he added.
Nature Beats Software
Nara Logics claims its new AI platform beats traditional neural network-based systems. While other systems use fixed algorithms, users can interact with Nara Logics’ platform, changing variables and goals to explore their data further. Unlike other AI models, the Nara software also can provide the reasons behind every recommendation it makes. “A lot of our health care customers say they’ve had AI systems that give the likelihood of somebody being readmitted to the hospital, for example, but they’ve never had those ‘but why?’ reasons to be able to know what they can do about it,” Nara Logics CEO Jana Eggers said in a news release. AI modeled on the brain could offer processing efficiency and reduction in energy costs compared to traditional AI, Steve Levine, the chief marketing officer of AI company Cortical.io, said in an email interview. “The human brain only needs about 20 watts to reason, analyze, deduct, and predict—less than a light bulb,” he said. “There have been a number of recent articles about the massive energy requirements and carbon footprint of the current data-centric AI approach. For example, approaches such as IBM Watson need 1,000 times more power to process information.” Another benefit to AI that works like the brain is the reduced requirement for training materials, Levine said. Most forms of AI now require thousands or millions of examples to be accurate. “Compare that to a human who only needs a few examples to learn a new concept, and it becomes obvious that an approach that mimics the way the brain learns will require much less material to be trained,” Levine added. Human-like AI could bring more flexible thinking, experts say. Most AI can’t handle new scenarios that they are not trained on, Manish Kothari, the president of the nonprofit technology research institute SRI International, said in an email interview. “AI systems today can repeatedly make the same mistakes,” Kothari said. “Even with retraining, today’s systems are prone to ‘catastrophic forgetting’ when a new item disrupts previously learned knowledge.”
Human-like AI Won’t Be Here Soon
But AI that truly mimics brain function is a long way off, some experts say. “The main challenge is that we don’t actually know how the brain processes information,” Levine said. Researchers are working to understand how the brain works and apply these insights to AI. The Machine Intelligence from Cortical Networks program, for example, aims to reverse-engineer one cubic millimeter of a rodent’s brain. “But, to put this into perspective, this represents only one-millionth of the size of the human brain,” Levine said. It’s possible that to build super-smart AI, we don’t need to mimic the brain at all, Wong said. After all, planes fly, but bear little resemblance to birds, he pointed out. Meanwhile, the world’s brightest scientists are working hard against the “non-intelligent” COVID-19 virus. “The bottom-up approach in mimicking the brain may not contribute to fundamental insights in the study of intelligence,” Wong said. “Even if neuroscientists can re-create intelligence by faithfully simulating every molecule in the brain, they won’t have found the underlying principles of cognition.”
