Page 94 - AI Standards for Global Impact: From Governance to Action

AI Standards for Global Impact: From Governance to Action

13 Empowering innovative and intelligent solutions at the edge

(edge AI)

13�1 What is edge AI and why is it becoming important

As AI evolves, the transition from centralized cloud computing to edge AI – including tinyML,
AIoT, and physical AI – is changing how data is processed, analysed, and acted upon. By
bringing AI capabilities closer to the source of data generation – whether in sensors, mobile
devices, or IoT systems – edge AI significantly reduces latency, bandwidth usage, and energy
consumption. In many use cases, edge AI involves the use of embedded AI chips that can
perform complex computations, such as pattern recognition, decision-making, and machine
learning tasks, without relying on cloud connectivity. This shift not only enhances real-time
decision-making but also addresses pressing concerns around data privacy and security.

The role of edge AI in creating sustainable, energy-efficient systems that can operate
autonomously in diverse environments, from rural communities to densely populated urban
centres (that directly address global challenges), was explored at a workshop at the Summit.

Presentations can be found at the Edge AI Workshop website.

Some of the key advantages of Edge AI include:

1) Reduced latency
Edge AI minimizes response times by processing data locally rather than relying on cloud
servers. This is crucial for applications like autonomous vehicles.
2) Bandwidth conservation
Autonomous vehicles are essentially moving data centres on wheels, generating massive
amounts of information every second. A single self-driving car can produce terabytes of
data daily from its cameras, LiDAR, radar, GPS units, and IoT sensors. This raw information
needs to be processed immediately to ensure the car can navigate safely and effectively.
For example, autonomous vehicles may need to process between 3 Gbit/s to 40 Gbit/s
of sensor data, depending on their level of autonomy. Sending all of this raw data to the
cloud would not be practical and create unnecessary costs. Edge AI allows cars to filter,
and process data locally, ensuring that only the most valuable insights are transmitted
to the cloud. This makes fleet management more efficient, as cloud systems receive
processed summaries rather than endless streams of raw sensor data.
3) Offline capability
Edge AI enables robots to function autonomously in environments with limited or no
internet connectivity. This is essential for applications like disaster response or space
exploration. This enhanced reliability is particularly crucial for mission-critical applications
where continuous operation is necessary, even in remote or disconnected environments.
Edge AI ensures high availability for devices by enabling them to operate autonomously
without relying on continuous internet connectivity or cloud-based services.
4) Enhanced data security and privacy
Processing sensitive information locally on devices using edge AI enhances data security
by reducing the risk of exposure or attacks during transmission to cloud servers. This
approach keeps critical data within the device, minimizing the chances of unauthorized
access, data breaches, or interception.
5) Energy efficiency
Recent research shows that shifting neural processing from CPUs and GPUs to specialized
hardware AI processors can greatly reduce power use in edge devices. By performing
computations locally and sending less data over the network, edge AI improves power

82