Pascal de Boer, VP Consumer Sales and Customer Experience
The rapid adoption of Artificial Intelligence (AI) continues, with 9 in 10 leading businesses[1] now revealing they utilise the technology in their operations. It’s perhaps not surprising when you consider the many, varied use cases AI already has brought to market. Benefits include better productivity, accurate real-time analytics, and the automation of arduous manual tasks. AI can be generated at the edge, in the cloud or in data centres. However, new technology advancements and innovations will likely embed or put AI directly onto more end devices – like phones, tables, laptops, and sensors. This will impact the architecture at the edge, demand for specialised storage and pose new challenges that businesses will need to address accordingly.
Understanding the Technology
While embedded AI technology on devices at the edge is still developing, it is quickly becoming common and will likely be widely available soon. For the technology industry, this is an exciting concept, as by transitioning away from the network core, embedded AI applications will become even faster and more useful, further enhancing AI’s use cases and capabilities.
Rather than being a tool individuals can leverage on their devices, embedded AI systems will be equipped to effectively interact with the environment, analyse data in real-time, make intelligent decisions and perform complex tasks[2] independent of connectivity or cloud-based computing.
By switching to AI-embedded devices, response times will be reduced, while safety and operational will improve. This will be transformational in the fields of healthcare, manufacturing, transport and entertainment. The functions of an AI-powered device can be performed almost instantaneously rather than having a delay by computing in the cloud or data centre.
Embedded AI in Action
In smart vehicles, embedded AI is already detecting congestion and obstacles on the road. For autonomous cars and other vehicles, these insights can keep passengers safe without the input of a human driver. Through the improvement of the reaction speeds of the AI-enabled vehicle, the automatic braking system and other functions are more likely to be approved for road use as the vehicles will be responsive to potential hazards on the road.
When used in healthcare devices, embedded AI massively improves the functionality of a sensors, both inside and outside the body. Because of the increased speed associated with in-device AI computing, a device can immediately act to provide required assistance for the patient as rather than simply triggering an alert. In cases where patients have extreme allergies, for example, embedded AI insights can automatically administer necessary medication to the patient. In critical situations, this could even save a patients’ life.
The Storage and Power for Embedded AI
By changing to device-led AI, computing power in the device itself will significantly increase. Rather than having to rely on connectivity to the core of the network where AI processing takes place, devices themselves can facilitate the AI model.
In addition, these devices will require high-capacity, high-performance storage to allow for the AI applications and to facilitate machine learning. As embedded storage is a necessity, it will have to be physically small enough to fit into devices such as wearables or other lightweight portable technology. Therefore, the hardware must be durable, compact but massively powerful, with features to handle AI workloads.
As with any innovation bringing something new to the market, early adopters will likely see the first iterations of these devices in premium settings such as private healthcare or high-end vehicles. However, as compact solutions for embedded AI become more commonplace, the use cases for embedded AI will expand to consumers and businesses of all sizes and industries.
The Future of Embedded AI
Soon, embedded AI devices will undoubtedly bring new and improved functionalities and experiences to end-users through better speed, productivity, safety and entertainment. Even though the technology is largely still in the development phase, development is likely to be rapid due to its many potential benefits. With the device-led AI gaining traction, we will likely see vendors integrating storage and AI computing function into smaller, sleeker devices. Therefore, these ground-breaking devices will help enable the next stage of AI development, as computing continues to move from the core to the edge.
[1]Ransbotham, S., Khodabendeh, S., Kiron, D., Candelon, F., Chu, M. & LaFountain, B. (2020). Expanding AI’s Impact With Organizational Learning. MITSloan Management Review. https://web-assets.bcg.com/1e/4f/925e66794465ad89953ff604b656/mit-bcg-expanding-ai-impact-with-organizational-learning-oct-2020-n.pdf
[2]Culjak, M. (2023). Embedded AI Demystified: A Deep Dive into Applications and Benefits. Byte-Lab. https://www.byte-lab.com/embedded-ai-demystified-a-deep-dive-into-applications-and-benefits/
