The adoption of artificial intelligence (AI) across industries is growing fast. It is helped by the democratization of many AI tools such as ChatGPT which reached more than 100 million monthly active users in January, just two months after launch, making it the fastest-growing consumer application in history, according to a UBS study. In healthcare, AI is being used for medical diagnosis and drug discovery. In finance, AI can improve credit scoring, fraud detection and risk assessment. In manufacturing, AI is being used to optimize production processes and improve quality control.
AI gets nourished from enormous amounts of data. One of the main sources of data is definitely the edge. By 2025, Gartner predicts 75% of enterprise-generated data will be created and processed outside a traditional data center or cloud. So. it is natural that AI may expand its footprint at the edge. In autonomous vehicles, it makes a safer solution thanks to the rapid data processing enabled by edge AI that allows the system to respond quickly to the world around it. In security cameras, Edge AI’s use of computer vision, object detection and facial recognition makes some security cameras particularly effective. In Smarthome, from video doorbells to voice controlled light bulbs and refrigerators that monitor things like food consumption and expiration dates, smart homes contain a web of IoT devices that are meant to work together to make the residents’ lives easier without the need to send back all the data centrally for processing.
There are indeed mainly advantages to run AI at the edge in terms of privacy and data security, low latency, offline functionality, bandwidth usage reduction, user experience enhancement and customized edge deployments. It expands the range of applications where visual recognition, business logic, intelligent and interactive conversational capabilities,… are required, even in resource-constrained or disconnected environments.
However, deploying edge AI poses some challenges in particular around security. Like other AI models, edge AI has to be trained on a regular and ongoing basis — just using data from edge devices. This often means creating an important data flow from the edge to the cloud or the datacenter, which can be rather complex (bandwidth, connectivity) and requires security features in place. Additionally, the environment on which the edge AI is operating has to be trusted to protect business operation processes and intellectual properties.