From competing on model parameters to racing to be the first to enter real home scenarios: Are AI terminals starting to compete for the "elderly and children" user base?
The next round of competition for AI terminals may not take place on the parameter sheets of launch events, but in real scenarios: whether an elderly person can use the toilet independently at night, whether a child can safely talk to toys, and whether a family is willing to entrust care data to machines.
Since the start of this year, national standards for artificial intelligence terminals have begun to cover multiple types of products including mobile phones, computers, glasses, speakers and earphones; the new round of consumption-boosting policies has put AI phones, smart wearables, robots and smart home appliances on the same consumption list as elderly-care products, rehabilitation assistive devices and smart care equipment.
The signal is crystal clear: AI hardware is moving from the stage of "whether a model is available" to the stage of "whether it can solve real problems". The reason why the "elderly and children" group has become a contested entry point for enterprises is not only due to the huge market space, but also because the demands here are specific enough, and the usability of products can almost be tested immediately.
From "Being Able to Chat" to "Being Able to Get Things Done", AI Hardware Faces New Assessment Criteria
In the past two years, the most common selling points of AI hardware have been large models, computing power and interactive capabilities. Mobile phones are equipped with
AI assistants, glasses can identify objects, earphones can perform translation, and speakers seem to better understand human speech.
The problem is that after the novelty wears off, consumers will still ask: Are these capabilities worth paying for a new device?
If an AI function only moves things that can already be done on mobile phones to another screen or another piece of hardware, it will be difficult to form stable demand. Users may be willing to experience it, but they are not necessarily willing to pay for it in the long run.
The difference of the "elderly and children" track lies in that the demands are not created by technology.
The transfer, toilet use, bathing assistance, fall warning and medication reminder for disabled elderly people originally require a large amount of time input from family members or caregivers; children's companionship, enlightenment, care and safety management have always been high-frequency demands for families. AI does not create a consumption reason out of thin air, but tries to reduce the cost of solving existing problems.
The implementation opinions issued by 7 departments including the Ministry of Commerce propose to expand the application of artificial intelligence and other technologies in elderly-friendly products and rehabilitation assistive devices, promote exoskeleton robots, electric wheelchairs, smart mattresses and equipment for defecation care, transfer, bathing assistance and other scenarios to enter families; for children's products, it proposes to develop green, smart and safe products such as educational toys and smart care equipment.
This means that the competition dimension of AI terminals is changing: in the past, manufacturers competed on how many questions their models could answer, but now they compete on whether their devices can safely and stably complete a real task.
"Being able to chat" is a demonstration of capability, while "being able to provide practical help" is the real reason for purchase.
The Rigid Demands of the Elderly Are More Unforgiving Than Parameter Rankings
At the 2026 Shanghai Senior Care Expo, a smart toilet robot designed for disabled, semi-disabled and mobility-impaired groups made its debut. It supports one-click call, autonomous movement, sewage cleaning, washing and drying, with a retail price of 28,999 yuan. The enterprise disclosed that the product has been operated in laboratory, real family and institutional scenarios for about a year and a half before the launch.
This product may not quickly enter thousands of households, but it reveals an important path for the commercialization of AI hardware: starting from the most basic, most repetitive and most labor-consuming care actions.
For elderly people with mobility impairments, using the toilet is not an "advanced demand" that occurs occasionally, but a daily life problem they have to face. Traditional assistive devices often still require family members to carry, clean and take care of the users. If a robot can reduce part of the manual links, the value it creates can be measured by time, nursing cost and life dignity.
This is easier to calculate than "adding an extra AI function".
But precisely because of the rigid demand, the elderly scenario puts more stringent requirements on products. Young people can tolerate AI assistants giving irrelevant answers occasionally, but care robots cannot make frequent mistakes during transfer, obstacle avoidance and emergency calls. A chat toy crash will only make people disappointed at most, while a malfunction of nursing equipment may directly bring safety risks.
Whether the elderly can understand the prompts, whether the buttons are simple enough, whether the device can identify dialects, whether it can still work after disconnection, whether it can pass through narrow spaces at home, and who will come for on-site maintenance when it breaks down — these details that seem less "attractive" than model parameters determine whether the product can truly enter families.
The elderly care technology project guideline released in Shanghai has incorporated this complexity into assessment requirements: relevant solutions need to support access to multiple devices, cover scenarios such as medical guidance, transfer and accompanying care, and adapt to different interaction modes including voice, text and images; for the emotional companionship direction, it also puts forward requirements such as dialect recognition, medication reminder, remote family linkage and safety monitoring.
This shows that elderly care AI terminals are not as simple as selling a machine, but a long-running "hardware + software + service" system.
Children's Companionship Is Both Opportunity and High-voltage Line
The entry point of the "children" market is lighter and easier to spread rapidly.
Since the start of this year, multiple toy and maternal and child enterprises have launched AI companion dolls. Shifeng Culture disclosed in its financial report that its AI toys cover modules such as emotional companionship and educational enlightenment, and have launched a number of AI plush products; Kidswant also launched AI smart companion dolls, and clearly stated that relevant businesses are still in the initial development stage and account for a small proportion of the company's total business.
The business logic of AI toys is not difficult to understand.
Traditional plush toys complete a one-time sale relying on appearance and IP, while AI toys can extend the life cycle through voice interaction, content update, growth memory and subscription services. Enterprises no longer only sell a doll, but may also include subsequent content, cloud services and membership relationships.
If children are willing to keep using the product, enterprises will have the opportunity to transform from "one-time hardware revenue" to "hardware + content service". For companies with IP, toy manufacturing capabilities and maternal and child channels, this path is particularly attractive.
But the children's scenario is also a high-voltage line.
Children may tell toys their home addresses, school information, emotional states and family member relationships. Issues including whether the device keeps recording continuously, where the data is uploaded, whether parents can delete records, and whether the model will generate age-inappropriate content cannot be dismissed with a simple sentence of "AI will become smarter and smarter".
Children's companionship cannot be replaced by "substituting for parents". Products can tell stories, answer questions and provide interactions, but they should not be packaged as all-powerful roles that can take responsibility for education, psychological judgment and even guardianship.
A more realistic problem is how long children's novelty can last. If an AI toy is only "a general large model wrapped in a plush shell" and starts to give repeated answers after several interactions, it will be difficult for parents to continue paying for subsequent subscriptions.
For children's AI terminals, real repurchase not only comes from children finding the product fun, but also from parents considering it safe, beneficial and controllable.
National Standards Begin to Define "How Smart It Actually Is"
Another long-standing problem of AI hardware is that there is no unified standard for "intelligence".
Different manufacturers can all claim that their products are equipped with large models, have intelligent agents or support multi-modality, but it is difficult for consumers to figure out what these terms actually mean. A device that can only respond to fixed instructions and a device that can understand tasks, call tools and collaborate to complete work may both be called "AI terminals" in the past.
The series of national standards "Classification of Intelligent Level of Artificial Intelligence Terminals" released in 2026 adopts the "2+N" architecture, divides the intelligent level of terminals into L1 response level, L2 tool level, L3 assistance level and L4 collaboration level, and formulates specific standards for categories such as mobile terminals, computers, TVs, glasses, vehicle cabins, speakers and earphones.
The value of the standard is not to select products for consumers, but to reduce the ambiguous space where "everything can be called AI".
For manufacturers, the standard can form R&D and testing benchmarks, and reduce the communication cost between different devices; for channels, it can display product capabilities more clearly; for consumers, they get the chance to distinguish between "advertised to be very smart" and "tested to reach a certain level".
However, this set of classification does not cover all nursing robots and children's toys at present, and the guiding technical documents are not equal to mandatory safety certification. The capability standards, product safety standards, data safety rules and service responsibilities of AI terminals still need to be further connected.
Especially in the "elderly and children" field, a higher intelligent level does not necessarily mean better performance naturally.
The elderly may need simpler, more stable devices that will not cause misoperations; children's products need stricter content boundaries and parental control. If enterprises only pursue higher levels and more complex functions, they may instead raise the usage threshold and fault probability.
The truly useful intelligence is not how many things the machine can do, but whether it can do the right thing properly in specific scenarios.
Offline Experience Is Becoming a New Channel for AI Terminals
Consumers can check the parameters of AI mobile phones online, but it is difficult to complete the sale of nursing robots and AI glasses only relying on a product detail page.
Consumers need to judge on site whether the device is comfortable, whether the operation is simple, whether the voice can be heard clearly, whether the elderly resist using it, and whether children like it. Especially for products with high prices or involving safety, experience is often more convincing than advertisements.
But this is also a cost barrier for the commercialization of AI terminals.
Establishing offline stores, training sales personnel, preparing experience devices, and providing installation and after-sales services all require continuous investment. Even a startup with excellent products may not have the ability to lay out national channels on its own.
The latest consumption-boosting policies propose to build "artificial intelligence + consumption" clusters, collection stores and artificial intelligence application experience centers, which is exactly to make up for this short board. Some shopping malls and elderly care consumption scenarios also begin to display smart assistive devices, companion robots and health monitoring products in a centralized manner.
In the future, AI terminals may develop a channel structure similar to that of the home appliance industry: manufacturers are responsible for products, shopping malls and professional experience centers are responsible for trial use, communities, hospitals or elderly care institutions undertake the role of recommendation and service entry, and third parties are responsible for installation and maintenance.
Whoever can enter these high-trust channels will be more likely to get family orders first.
However, offline experience will also expose real problems. Devices that run smoothly at launch events may encounter dialects, noises, weak networks, narrow spaces and complex usage habits in ordinary families. Experience centers should not only be product exhibition halls, but also become test sites for collecting faults, verifying demands and iterating services.
Before Mass Production, Pass Five Family-related Barriers First
The "elderly and children" group provides the most real demands for AI terminals, and also sets the highest commercial threshold.
To judge whether a product can achieve mass sales, we can first focus on five things.
The first is the rigid demand scenario. Does it solve problems that happen every day, or just a novel function that is used occasionally?
The second is the interaction threshold. Can the elderly use it without learning complex menus, and can children use it independently within the controllable range of parents?
The third is offline experience. Can consumers try the product before purchase, and does the channel have the ability to explain and demonstrate?
The fourth is after-sales service. Who will repair the device when it breaks down, whether it can still work after the software stops updating, and whether the expiration of subscription will affect basic functions?
The fifth is data security. Where are the recordings, images and health information stored, who can access them, and can users turn off and delete the data at any time?
The beneficiaries may be comprehensive enterprises with capabilities in hardware manufacturing, channels, service networks and vertical data; those under pressure will be followers who only install general models into new shells and lack after-sales support and scenario understanding.
And the current judgment also has clear boundaries: policy support, standard release and new product launch do not mean that commercialization is mature. Real sales volume, usage frequency, return rate, subscription renewal rate, maintenance cost and safety records are the final report cards that determine whether AI terminals can enter families.
The real popularization of AI hardware in families may not rely on a more beautiful parameter list, but on a very simple standard: the elderly are willing to use it, children really need it, and family members can buy it with full confidence.
References:
1. Ministry of Industry and Information Technology of the People's Republic of China: Series of National Standards "Classification of Intelligent Level of Artificial Intelligence Terminals" Released, May 8, 2026
2. 7 departments including the Ministry of Commerce: Implementation Opinions on Promoting the Expansion and Upgrading of Commodity Consumption, August 13, 2026
3. Shanghai Science and Technology Commission: Application Guideline for the "Elderly Care Technology Support" Project of Shanghai 2026 High-quality Technology Industry Development Plan, June 26, 2026
4. Xinhuanet: A Toilet That Can "Walk"? Focusing on Accessible Technology, the Smart Toilet Robot Officially Debuts, June 6, 2026
5. Shifeng Culture: 2026 First Quarter Report, April 25, 2026
6. Shanghai Securities News: Improve Standards and Break Through Channel Barriers, Policies Support AI Terminal Products to Focus on the "Elderly and Children" Track, September 2, 2026