“Revolutionizing the IoT: The Potential of Neuromorphic Computing in Mimicking the Brain” – Insights on Digital Transformation

To the best of my knowledge, with the progress of the Internet of Things (IoT) and trials on imitating the human brain, it is believed that neuromorphic computing can be the technology to help with this. Neuromorphic computing uses analogue signals to transmit information from one neuron to another, unlike digital computers which use binary-encoded signals. Neuromorphic computing is expected to be parallel, scalable and have collocated memory processing. Experts estimate that with the advent of distributed and edge computing, neuromorphic computing can be used in emerging fields such as IoT and drones, among others. “Neuromorphic chips can run on batteries with a very low power consumption and can learn autonomously with newer inputs making it adaptable to various conditions. Since the computing power is available at the edge it will have less reliance on IoT cloud computing but can enhance IoT use cases and enable solutions to be built with a combination of IoT and Neuromorphic computing,” Abhijit Roy, director, global head, energy and utility domain and IoT, Happiest Minds Technologies, a digital transformation provider, told FE-transformX, adding that IoT platforms still need to aggregate vast amount of information being processed by the neuromorphic edge.

Market statistics and use case

The global market of neuromorphic computing can surge from $92,450,000 in 2023 to $527,604,000 by 2027, as per insights from Fortune Business Insights.  For instance, brain-computer interfaces or computer-to-brain interfaces can be an important area where neuromorphic systems can be used. These interfaces allow humans to control computers and other devices using their brain activity. Another instance could be the use of an event camera or neuromorphic camera in the military which can be an important part of warfare. An event camera has a sensor which processes events in a human-like manner while at the same time using lenses that can see much better. “The creation of artificial neurons to do computations brings us closer to mimicking human intelligence and makes robotic intelligence less ‘artificial’. A large part of the research in this area is derived from a significantly greater understanding today of the human brain and how it works,” Avimukt Dar, founding partner, IndusLaw, a law firm,  explained.

The market prediction for neuromorphic computing can be enormous and constrained only by the ability of scientists to develop and industry to deliver on all potential applications. As per several media reports, the use of neuromorphic technology in deep learning applications, transistors, accelerators, next-generation semiconductors, and autonomous systems, such as robotics, drones, self-driving cars, and artificial intelligence, can result in market growth. For example, in August 2022, a research team developed NeuRRAM, a new neuromorphic chip to manage various AI applications at higher accuracy and lower energy than other platforms, as per insights from Grand View Research, a market research firm. “However, neuromorphic computing could encounter a significant challenge related to bias, as unintentional or intentional data biases may influence future decision-making processes. The ethical and legal considerations surrounding a system capable of learning, adapting, and independent thinking raise valid concerns,” Devroop Dhar, co-founder and managing director, Primus Partners, a business consulting firm, said.

The loophole in brain-computer interfaces

Industry experts believe that neuromorphic computing aims to mimic the capabilities of the human brain, which has the potential to develop more energy and compute-efficient AI. Industries such as manufacturing and security are exploring neuromorphic computing for optimising processes and enhancing tasks like facial recognition and behavioural analysis. However, critics argue this could lead to ethical concerns in hyper-realistic generative AI, as it opens avenues for potential misuse and cybersecurity concerns. “The capabilities of these systems to create convincing deepfakes and pose risks such as misinformation dissemination and public perception manipulation. This not only impacts legal structures but also presents privacy concerns, including the potential for blackmail or harassment through fake media creation,” Sachin Panicker, chief AI officer, Fulcrum Digital, an IT solutions provider, highlighted. 

Reportedly, North America dominated the neuromorphic computing market with a more than 37.4% revenue share in 2022. The United States (US) and Canada, are expected to be the frontiers of new applications. One of the trends in the region is AI-based voice and speech recognition technology, as per insights from Grand View Research. Interestingly,  physicists believe that the physical world is based on quantum principles, which are the least explored. “Neurocomputing, artificial intelligence (AI) and quantum computing have the potential to be the future of technology. IoT devices are designed for specific tasks so these neurocomputing capabilities related to that particular task only will be included on that chip. It can be an advantage over current IoT systems,” Nihar Ranjan Roy, associate professor, centre for cybersecurity and cryptology, Sharda University, said. 

Despite neuromorphic computing’s ability to have context-sensitive and advanced pattern recognition abilities, there is a risk that it will also mimic human biases and entrenched discriminations in its decisions and recommendations. Such opacity and bias could raise questions about the accountability and liability of organisations employing these processors, particularly when they lead to problematic outcomes. For example, split-second localised decisions made by neuromorphic processors in autonomous vehicles that result in accidents, or the context of bio-hybrid technologies, adverse autonomous decisions made by neuromorphic processors relating to pacemakers or medical diagnostic devices could lead to complex liability issues. “In response to this, regulatory measures can be considered once specific problems become more evident. If used responsibly, these processors hold the potential to usher in advancements in low-energy, rapid, localised, and robust processing, potentially revolutionising the realm of IoT devices,” Chinmay Deshmukh, assistant professor, BITS LAW, a law school, concluded. 

Follow us on Twitter, Facebook, LinkedIn

Leave a Reply

Your email address will not be published. Required fields are marked *