Edge AI isn’t a type of algorithm in and of itself, but it's about where AI happens, regardless of the algorithm in question.
www.forbes.com
Edge AI: The AI No One Is Talking About
Aurangzeb Khan
Forbes Councils Member
Forbes Technology Council
COUNCIL POST| Membership (Fee-Based)
Oct 11, 2023,06:15am EDT
Aurangzeb Khan, Senior Vice President, Intelligent Vision Systems at
GN Jabra.
GETTY
If you ask ChatGPT, “What types of AI are best suited for businesses?” it will tell you that the answer depends on the use case. Natural language processing (NLP) models help computers process language. Computer vision algorithms allow machines to evaluate images. Generative AI, which has been all the rage lately thanks to ChatGPT and its ilk, seeks to create new data samples based on existing patterns.
There’s a chance that, in all the chatbot furor, you’ve heard about various types of models, like NLP and generative, that enable AI. One piece of the AI puzzle that has remained absent from headlines so far, however, is edge AI.
That’s partly because edge AI isn’t a type of algorithm in and of itself. Instead, edge AI is about where AI happens, regardless of the algorithm in question. Edge AI—or AI at the network’s edge—may be the most important development for the future of business and AI symbiosis.
The network’s edge is a goldmine for business.
The edge is becoming vital in IT infrastructures for making real-time decisions, particularly when a business needs to analyze large amounts of data with ultra-low latency. Moving data processing near to where data is generated enables faster processing and reduces energy consumption, network congestion and latency.
Faster, more efficient processing boosts reliability and responsiveness. Consider self-driving cars, equipped with a host of sensors. Each sensor must swiftly capture and analyze information about its surroundings (hello, computer vision models and AI), and then help the car’s autonomous driving system make good split-second decisions. If the sensor must send data back to the cloud or a regional data center for processing, it can undermine timely decisions.
Integrated collaboration tools, including cameras, microphones, speakers, edge AI and compute system-on-chip integrated circuits—
combined with cloud-delivered unified communications (UC) services, like Microsoft Teams and Zoom—
empower businesses to support dispersed teams using current networks. The powerful duo optimizes bandwidth for video conferencing and upholds quality. Security safeguards implemented at the edge and in the cloud help companies identify and thwart cyber threats, even when team members connect from disparate locations.
And edge computing eases strain on networks and the cloud, enabling seamless scalability for businesses. Need a new branch office? Welcoming team members from a different state? Edge computing ensures networking and security concerns will not slow company growth.
AI at the data’s source is a powerful tool.
Communications tools that keep distributed workforces moving in sync are an edge AI case with which many are familiar.
Edge AI already plays an integral role in enabling seamless communication by improving the performance, security and overall efficiency of collaboration tools. Microsoft, for example, has an
automated assistant for its Teams app, which manages calendars and helps users find information. Zoom launched an AI-powered tool for sales teams that analyzes
customer conversations to deliver useful insights.
The most advanced cutting-edge collaboration systems employ edge AI processors for real-time processing of large data volumes. This ensures a responsive and natural collaboration experience, delivering curated audio-video streams to cloud services from the source of information.
In industrial settings, firms outfit mechanical assets with sensors for continuous monitoring. Algorithms that detect usage anomalies enable predictive maintenance and keep systems safe from cyberattacks. Smart city grids with equipment at the edge can enlist AI to optimize traffic flows or monitor environmental factors. In retail settings, models that analyze video can review shopper movement patterns to optimize store layout and product placement.
Back in the (home) office, edge AI can continue to further enhance operational efficiencies. Voice assistants like chatbots, transcription or translation services use NLP algorithms to better understand what humans are saying and to produce natural-sounding output. And advanced, personal audio and video devices—which integrate edge AI processors within them—similarly optimize the experience, autonomously and efficiently, enabling users to focus on their business purpose and communications.
What are the potential pitfalls?
Of course, implementing AI-enabled solutions isn’t without challenges. From an external standpoint, one major hurdle is the limited memory and power available. The constraint on power can hinder the deployment of AI models, which could result in trade-offs between model accuracy and resource consumption.
Beyond getting models to run successfully,
other potential issues are scalability and specialization. Often, businesses must customize models to fit the location, hardware and use case, which can result in friction and limit the ability to run several models on edge devices. It’s important for users to choose the right AI framework and models that fit their edge AI scenarios and uses.
Additionally, it’s important to ensure proper security measures are in place. Yes, edge AI reduces threats as it processes data on local devices—however, decentralized environments come with their own challenges, as sensitive data processed on local devices could be vulnerable to breaches. Data at the edge can be secured using a container-based approach. It can also be anonymized for transport if needed.
Overall, the successful deployment of edge AI hinges on careful orchestration and involvement from many stakeholders. It shouldn’t boil down to one team—engineering, product, IT and data teams should all be involved. Because of the complexity, involving all of the necessary stakeholders can show potential challenges beforehand and allow them to be solved quicker when they do arise.
Edge AI can help steer companies toward growth.
AI might be a relative newcomer to the world of communications, but it’s here to stay. For companies to grow, they must scale operations. Scaling tends to complicate everything from management hierarchies to IT infrastructures. With thoughtful implementation, edge AI can act as a sort of rudder—steering everyone and everything in the same direction even amid growing complexity.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives.
Do I qualify?
PVDM may want to have a chat with his fellow Forbes Council Member and tell him about his brilliant solution to swerve around those pitfalls…
...
