Among numerous products claiming the use of artificial intelligence, certifying the safety and effectiveness of AI functions offers manufacturers a competitive advantage.
Acquiring and using data in consumer products, such as smart home thermostats and refrigerators, to provide end users with an additional level of interactivity, information and intuitive features has become commonplace in today’s appliance industry. The convergence of several technologies — from low-cost sensors to high-speed networks — is enabling a great deal of innovation in the smart home device space. New products that take advantage of technology to deliver value to consumers are now introduced regularly. Examples include smart HVAC thermostats, ovens that can turn on and preheat when synced with a mobile cooking app and refrigerators with smart screens.
Data quality, storage, analysis and protection is critical because data can help appliances acquire a new level of intelligence to interact with other systems, provide predictive analytics and assist humans in better managing their daily tasks from anywhere in the world.
The evolution of artificial intelligence in smart homes
In the next 10 to 15 years, smart home technology will cross platforms to provide complex decision-making solutions and increase device capabilities. One technology in particular — artificial intelligence (AI) — has captured the imaginations of both consumer and product manufacturers as a unique capability that delivers amazing value. AI will become an area of strategic importance and a key driver of economic development.
Artificial intelligence opportunities and machine learning
Artificial intelligence (AI) is a broad domain. What is usually meant by “AI” is the use of an approach known as machine learning (ML). With ML, data for a particular problem of interest is gathered by a product manufacturer, and an algorithm is mathematically trained to find patterns in that data. For example, a refrigerator manufacturer may want to alert you that an item inside your refrigerator is approaching expiration. An algorithm would need to be created that can take photos of your food packages, locate the expiration date and read that information and generate the expiration date. ML can also help consumers save energy and money with technologies such as smart thermostats that can automatically set, increase or decrease the temperature in your home according to weather conditions.
ML will likely drive digital business as a highly disruptive class of technology and will enable organizations to harness data to adapt to new situations and solve problems not previously encountered. Many of today’s connected products are equipped with sensors that help product manufacturers learn about how their products are used, so they can, in turn, deliver better products. Also, consumer products often automatically download software updates to continually improve the user experience.
However, it is important to note that ML-based products can also fail. The market is full of claims about AI-enabled features — without clarity on what qualifies a product as using AI. As the number of products claiming benefits driven by AI increases, a mechanism must be put in place to help consumers have confidence in such claims. Technology has penetrated everyday products at a swift pace, but product safety standards are still in the process of being revised. Even when an AI-enabled feature is not a safety feature, the claims of an AI-based element must still be verified to justify the premium paid by the consumer and help ensure continued trust in the marketplace. Industry standards should be implemented to minimize risk and help protect the consumer.
How UL can help
The UL AI Algorithm Reproducibility Process Claim Verification program is the first independent assessment of AI-enabled products. The general principle is similar to that of peer-reviewed scientific claims: Product manufacturers share with UL all critical information about the algorithm for independent analysis while ensuring intellectual property remains proprietary. UL’s Marketing Claim Verification for Artificial Intelligence Algorithm Reproducibility program has the flexibility to deliver proportionate risk assessment, allowing non-safety critical AI systems to be a model for future AI regulations. Once tested and certified, the qualified product(s) can be listed on the MCV Products list directory and are issued a Mark to differentiate the product in the market [https://verify.ul.com/search?&dir=&order=&q].
“At UL, we support manufacturers that want to differentiate their products in the marketplace with solutions that go beyond a single feature,” said Yuwen Huang, product manager of UL’s Appliances team. “We want to provide our partners with our experience and knowledge to help them address their value proposition in a competitive landscape. The real challenge is helping to ensure user data is fully protected but can still be utilized by AI-enabled products to learn how to improve.”
As the appliance industry continues to prominently advertise AI technology, setting a benchmark is crucial so consumers can trust that the product claims enabled by AI are verified. AI verification is only one of many services UL offers manufacturers. We provide a complete suite of services to reduce the risks related to connected products through a security- and privacy-by-design strategy so that companies can maximize the effects of their investments. Together UL, developers and manufacturers can effectively address growing concerns related to smart home systems and devices — including security, privacy, electromagnetic capability and product interoperability — through a global network of laboratories, a team of experts and a single point of contact, helping protect consumer safety everywhere.