"Edgy Business: What Is Edge AI, And How Will It Impact Business?
Peter van der Made
Forbes Councils Member
Forbes Technology Council
COUNCIL POST| Membership (Fee-Based)
Jun 22, 2023, 08:15am EDT
Peter van der Made is the founder and CTO of BrainChip Ltd. BrainChip produces advanced AI processors in digital neuromorphic technologies.
The “edge” refers to computational AI devices that do not rely on the cloud for their results. This makes AI more accessible to users. It runs computations directly on local devices like laptop PCs, internet of things devices, sensor data processors and medical devices using many autonomous parallel neural processors (NPUs) that function independently of the central processor. In the past, edge devices used the cloud to perform their AI computation.
The cloud is physically realized as an internet connection to a large data center that houses thousands of powerful rack-mounted computers. Before 1981, nearly all computing was done this way. The point of contact was what was called a “dumb” terminal that was connected to a large mainframe computer through telephone lines. The terminal did very little processing beyond receiving input data, uploading it to the mainframe computer and receiving the computed results.
Today’s PCs are far more powerful but still need more processing power to perform the many millions of calculations needed in artificial intelligence applications. Hence, the input data, whether an image, video stream or other data, is uploaded through the internet to a data center, where it is processed, and the results are sent back to the “edge” PC or other device. Examples of this include Amazon’s Alexa and OpenAI’s ChatGPT.
Recently, small devices have been developed that significantly increase the processing power of edge PCs and AIoT devices by offloading the processing task entirely from the main processor, enabling them to perform many computations locally on the device. Edge AI computing does not need to rely on an internet connection any longer.
This has several advantages. Think, for instance, about the risks of uploading video data from a security camera to the internet, where criminals can hack into the video stream to look inside your home or business whenever they want. This is one of many problems that can be solved with edge AI by processing the video stream locally to recognize critical events and send only alerts over the internet.
Medical data, sensory data of critical systems, manufacturing processes, and networked driver assistance systems all face the same security issues that are alleviated by processing the data at the edge. Another solved problem is that cloud-based devices will work only where an active internet connection is available.
The move to edge computing presents a significant shift in the way businesses deal with data and how decisions are made. Cloud services can be expensive over a long period, and those charges continue over the product’s lifetime. On the other hand, cloud services require only software development with little or no investment in hardware development.
The cost savings of edge computing are just one factor. Reduced latency, eliminating annoying wait times, higher power efficiency, interacting directly with locally generated data and increased security are others, as well as availability in areas with no or limited internet availability. Instantaneous learning at the edge enables adding product features or reconfigurability at the edge. Combining these factors introduces new opportunities and flexibility to the business.
A wide range of industries can benefit from edge AI. Edge AI can be used in smart thermostats, security systems and other home automation devices to optimize energy usage. The home can learn the preferences of its occupants and adjust the lighting and temperature accordingly.
In autonomous vehicles, edge AI can monitor moving parts to warn of a future failure, for instance, by vibration or engine noise analysis and processing the input from cameras and other sensors to determine if the driver is alert and not incapacitated.
In agriculture, edge AI can monitor crop health, determine if nutrients are missing from the soil and identify weeds and pests for selective spraying. Edge AI can monitor environmental health hazards and traffic control in smart cities.
In healthcare, edge AI can monitor patient indicators, be used in wearables and implantable devices, predict vital signs and alert medical staff of an impending crisis in real time. It also provides sophisticated hardware to execute algorithms for AI IoT devices, machine learning solutions and autonomous deep-learning model applications independent of cloud services to provide better process and quality control.
Over the next few years, many new products will emerge equipped with edge AI to enable advanced functionality in portable and battery-operated handhelds and wearables. The time is now for businesses across all industries to evaluate what functionality their clients will expect and what product enhancements are needed, with the introduction of new products that can compete in an edge-AI-equipped world. Product enhancements could include a way to control a device with speech commands, facial recognition, emotion recognition, food quality monitoring, hazardous gas identification, sound identification, matching of retail barcodes to products' images, and impact and vibration classification.
Whatever the application, your development project should always start with a feasibility study that includes an overview of the current product, an indication of what needs to be added, a technical evaluation and a project scope with deliverables and milestones. Is training data available, and what data engineering tasks must be undertaken to make the data suitable for neural network training and verification?
The project plan needs to identify talent already available in the business, what additional training may be required, the additional talent that needs to be hired and when—and a detailed timeline with resource allocations and dependencies.
The cost and effort of adding AI to an existing or new product must not be underestimated. Still, the effort is well worth it when it strengthens the company's competitive position in the market and leads to increased sales.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives.
Do I qualify?
Follow me on
LinkedIn. Check out my
website.
Peter van der Made
Peter van der Made is the founder and CTO of
BrainChip Ltd. BrainChip produces advanced AI processors in digital neuromorphic technologies. Read Peter van der Made's full executive profile
here."
https://www.forbes.com/sites/forbes...t-is-edge-ai-and-how-will-it-impact-business/
seinfeld.fandom.com
