Samsung's making a big play on AI with its home appliances, and it appears to be paying off. Its Bespoke AI Combo washer and dryer has become an instant hit in many markets, its Bespoke 4-Door Flex with AI Family Hub is the smartest fridge Samsung has ever made, and AI has even found its way to Samsung's Jet Bot vacuum cleaner.
AI isn't just making these devices smarter and more efficient, it's also opening up a whole new avenue of collaboration between appliances and mobile devices in Samsung's ecosystem. This enables the company to provide highly personalized user experiences the likes of which few are able to match. It's certainly an exciting time for Samsung's appliances business as it seeks to maximize the opportunities that exist in this segment.
To learn more about how the integration of AI has helped this segment of the business, what the challengers are, and how Samsung views the future prospects, we had the opportunity to interview Samsung's, Miyoung Yoo, EVP, SW Development Team, Digital Appliances Business, ahead of the Samsung Developers Conference 2024 in San Jose. As was to be expected, Samsung has introduced new AI functionality for its appliances, further iterating its focus on achieving the full potential.
Answers have been paraphrased for brevity and clarity.
Q1. How has the market responded to Samsung’s AI-enabled Bespoke appliances so far? Are there specific features powered by Generative AI that have resonated particularly well with users?
A: Since home appliances are often shared by multiple users, ease of use for every user is a must. With this in mind, Samsung focuses on designing its AI-enabled appliances to simplify tasks and ensure they seamlessly integrate into everyday life while making them easy for everyone to use.
We’re already seeing positive reception to this strategy in consumer feedback and sales metrics. For instance, in this year July, Samsung received more J.D. Power Awards for Kitchen and Laundry Appliances than any other brand in 2024 in the U.S. We can assume that this result reflects the market response towards Samsung’s Bespoke AI appliances, underlining the usefulness of their AI features and the impact they’re having on consumer experiences.
Q2. Has the adoption of AI-powered Bespoke appliances met Samsung’s expectations in terms of both consumer engagement and sales performance? What were some of the key metrics or indicators that the company focused on to gauge success?
A: We’re excited to see strong interest in the market so far. For example, the Bespoke AI Laundry Combo has achieved remarkable sales results since its release earlier this year. Within just three days after launching in South Korea in February, 1,000 units were sold, and by April, total sales had surpassed 10,000 units.
Having captured over 40% of drum type washing machines in Korea, this product has quickly established itself as a trendsetter. Building on this success, Samsung is expanding its availability beyond South Korea to the U.S., with plans to extend sales to Latin America, Southeast Asia, and Europe in the fourth quarter.
This success can be attributed to two key strategies. First, enhancing the consumer experience is the foundation of our AI strategy. This focus allows Samsung to uncover AI functionalities that resonate with diverse consumer needs, to make household tasks easier. Second, Samsung has prioritized security, addressing one of the top concerns consumers have when using AI
functionalities.
Q3. How has the customer adaptation been with respect to the learning curve for AI-enabled appliances, and has Samsung encountered any unexpected challenges in how users are interacting with AI features in their appliances?
A: While users are generally receptive to AI features, as we explore this new technology, it can be challenging for some users to integrate AI into their real lives. To help users get the most out of AI-enabled products, we developed the AI Home, a 7-inch LCD display, and rolled out the upgraded Bixby.
Users are now able to intuitively understand and control product features, including AI features, through the larger screen. And of course, we will continue to develop our products in ways to create synergy between software and hardware to improve usability for consumers.
Q4: Why hasn't Samsung made any Matter compatible home appliance? Even though SmartThings supports Matter, Samsung's own appliances don't support Matter. Why is that the case?
A: Samsung is committed to applying and expanding industry standards to enhance the connectivity of user experiences. Through the SmartThings Station, the refrigerator with AI Family Hub, and smart TVs, Samsung continues to support Matter integration. These devices serve as hubs and controllers for both Matter and non-Matter devices. It allows small Matter devices like light bulbs to easily and more extensively connect through our hub devices.
We will continue to expand the application of these Matter-supporting hubs to our devices. For large appliances, they usually have brand-specific features and integration scenarios, so they work well when directly connected to the cloud than integrated through a Matter controller.
Q5: How do the AI algorithms on Bespoke Appliances optimize functions like energy consumption, cleaning cycles, or temperature control in real-time based on user behavior and environmental factors?
A: We’ve equipped several products with a variety of on-device AI features that track user behavior and environmental factors in real time to enhance consumer experiences. More specifically, the AI Object Recognition of the Bespoke Jet Bot Combo AI utilizes an AI deep neural network (DNN) model trained on 1.7 million images.
It identifies the shape of objects through a camera on the front of the robot, and then the DNN model detects the recognized objects. It will become smarter, able to understand changes in the environment as users interact with it, through continuous updates to the data and AI models.
Q6: How exactly does generative AI work in the context of Bespoke appliances? Are there specific use cases where generative models are helping appliances to learn and adapt to individual household patterns, and how does this differ from traditional AI methods?
A: The Bespoke AI appliances we launched this year take a more complex and diverse approach in the implementation of AI to better address consumer pain points. We upgraded Bixby based on natural language to enhance user convenience when operating appliances. While the previous generation of Bixby supported hundreds of essential commands for device usage, the upgraded Bixby has evolved to better get the complex, multiple commands and understand the context of user commands.
For example, as people race to get their children ready for school in the morning, they can ask
Bixby to start the vacuum and the washing at once, and to also let them know if bananas are in the refrigerator and then the related recipe recommendations using them. We are also applying various on-device AI features during development and through updates.
Q7: How does Samsung manage the demanding computational load of running these models on home appliances, which have limited processing capabilities?
A: We have designed to process data in the cloud for operations that require large AI models. These cloud-based AI algorithms are continuously optimized for efficiency and speed for users. In cases where privacy is critical, such as with cameras, and fast processing is required, on-device AI models are used for processing. We continuously improve and upgrade these AI models by training them further.
Q8: Could you explain how Generative AI contributes to the personalization features of Bespoke appliances, such as creating tailored usage patterns or automating tasks based on inferred user preferences?
A: Samsung appliances’ AI functions are designed to adapt to individual lifestyles for a completely personalized experience. For example, the Bespoke AI Laundry Combo and Bespoke A Washer have AI Wash, which uses AI algorithms to adjust the washing cycle based on weight, fabric type and turbidity.
It also continuously monitors the level of soiling in real time, adding water, detergent or rinse cycles when needed to provide a personalized wash that helps your items are properly cleaned. Additionally, frequently used cycles are automatically prioritized on the screen, and users can even get recommendation on wash cycles based on their preferences
or factors like weather.
Q9: What kind of data is used to train the generative models in Bespoke appliances, and how is this data sourced and processed to ensure accuracy in tasks like recipe suggestions, cleaning patterns, or energy management?
A: Samsung is using an internally developed AI model (Further details cannot be disclosed due to internal policy).
Q10: What do you see as the next frontier for AI integration in home appliances? Are there any upcoming innovations, perhaps through more advanced contextual understanding or autonomous decision-making?
A: Samsung plans to expand our recently upgraded Bixby to a wider range of products. Currently Bixby has been updated for five product lines. While I can’t share specifics on the timeline, I can say we are preparing to bring upgraded Bixby to previously launched models too and are so excited to bring Bixby functionality to even more Samsung users this way.
Additionally, to broaden the range of applicable products, we are developing Tizen Lite, a lighter version of the Tizen OS that is currently featured in premium models. The UI and Bixby are also being streamlined to operate efficiently in a lighter Tizen environment. This will allow us to improve usability by expanding Bixby to more affordable models in the future.
Q11: How far can hyper-personalization go in the context of Bespoke appliances? Do you foresee a point where appliances could proactively adjust their operation not only based on user habits but also on predictive analytics about future needs or preferences?
A: Samsung has already implemented AI capabilities in some products to predict situations and adjust operation accordingly. One example is the AI Hybrid Refrigerator introduced at IFA. Currently available in South Korea, it combines a compressor with a new cooling method called Peltier technology.
It generally uses only the compressor, but when necessary ― such as when a user keeps the door open for a long time to store many items, or it’s a hot summer day ― it activates the Peltier component as an extra cooling source to use energy efficiently. This product utilizes a machine learning model and based on the analysis, it predicts future temperature changes inside the refrigerator and adjusts its operation accordingly to provide optimally cool temperatures.
Another example is the smart oven, which currently recognizes 88 types of food in the U.S. market. When consumers cook dishes beyond the recognized list, the oven autonomously learns and expands the range of items it can identify. The more it is used, the more it evolves into a device that not only recognizes a broader range of foods but also provides personalized recipes, becoming increasingly hyper-personalized.
Lastly, we are applying AI models to improve usage quality through predictive analysis. The AI models are advanced that they can predict and diagnose problems SmartThings-connected devices and guide consumers to receive quality services in a timely manner.
Q12: Could we see a future where generative AI in Bespoke appliances extends to more creative tasks, like designing optimal meal plans based on nutritional needs or generating custom energy management strategies for the household? What are the limits of generative AI in this space, and how do you envision pushing those boundaries?
A: Samsung leverages AI across various services to support the creative tasks our consumers might carry out. For instance, our platform Samsung Food assists with meal planning by allowing users to modify existing recipes based on their health conditions and preferences. AI Vision Inside that features in the AI Family Hub recognizes certain food items taken in or out of the refrigerator and links them to the Food List.
A great deal of research is being conducted on AI models, and ongoing studies within our internal teams further support this effort, which is why we believe that the potential for AI technology to evolve is limitless. Additionally, we aim to continuously strengthen related services through our proprietary upgrade service, Smart Forward.
We are planning to optimize and enhance high-performance APs, NPUs, and other components capable of processing vision, voice, and data to bring even more of these innovative technologies to home appliances in the future.
