Voice Assistants and Conversational AI in Aged Care: Five Applications and Their Risks

Voice is one of the most natural ways people communicate. It allows us to ask questions, express needs and interact with our surroundings without navigating menus, keyboards or touchscreens.

For older people and those living with disabilities, voice interfaces can reduce barriers to accessing information and digital services. This is particularly valuable where vision, dexterity or mobility limitations make conventional technology difficult to use.

When eHomeCare first explored voice assistants in 2017, the possibilities centred on smart speakers, reminders and connecting everyday services. Today, conversational artificial intelligence is extending those capabilities considerably.

However, the fundamental question remains unchanged: how can voice technology improve independence, care coordination and wellbeing without introducing unnecessary risks?

From Voice Commands to Conversational AI

Traditional voice assistants were primarily designed to recognise commands and perform predefined actions. A person might ask for the weather, set a reminder, make a telephone call or control a connected device.

These functions remain useful, particularly when they remove the need to operate a screen or remember a sequence of instructions.

Conversational AI introduces a different level of interaction. Instead of relying exclusively on predefined commands, newer systems can interpret more complex questions, maintain conversational context and generate responses in natural language.

This creates opportunities for more personalised support, but it also changes the risks. A traditional assistant might fail to recognise a command. A generative AI assistant may produce a convincing response that is factually incorrect.

For care providers, this distinction is critical. Technology that helps someone access information should not automatically be trusted to interpret clinical needs or make consequential care decisions.

For aged care providers, evaluating conversational AI requires more than assessing functionality and ease of use. Each proposed application should have a clearly defined purpose, an accountable business owner and appropriate safeguards reflecting the consequences of incorrect or misleading responses.

Providers must also understand how personal information is processed, whether conversations are retained or used to improve AI models, how the technology integrates with existing care processes and when human intervention is required. These considerations should form part of technology selection and ongoing governance, rather than being addressed only after implementation.

1. Accessible Information and Education

Voice interfaces can make information more accessible to people who have difficulty reading screens or navigating digital applications.

A person might ask about an upcoming appointment, receive instructions for preparing for a visit or request an explanation of a care service. Voice can also support education about health conditions and everyday wellbeing.

However, providers should distinguish between delivering approved information and generating personalised medical advice.

A voice assistant reading an established care instruction is performing a different function from an AI system independently recommending what someone should do about new symptoms. The second requires considerably greater clinical assurance.

Providers should therefore establish which information sources are approved, how content is maintained and when the system must direct the individual to a qualified professional.

2. Medication Reminders and Daily Routines

One of the earliest applications of voice assistants was supporting medication adherence and everyday routines.

A system can remind someone that medication is due, prompt them to complete an activity or provide information about their daily schedule. The individual may also be able to acknowledge a reminder verbally.

This can be useful for people who find written schedules difficult to follow or prefer spoken prompts.

However, acknowledging a reminder does not establish that medication was taken correctly. Speech recognition errors, misunderstandings and incomplete responses can also affect the reliability of recorded information.

For providers, voice-based reminders should therefore be considered part of a broader medication-management process rather than a substitute for appropriate clinical oversight.

Similar principles apply to hydration reminders, exercises, appointments and other daily activities. The technology may support consistency and independence, but its effectiveness depends on individual suitability and appropriate follow-up arrangements.

3. Care Coordination and Communication

Voice interfaces can simplify communication between people receiving care, their families and care providers.

For example, someone might ask when their next care visit is scheduled, request assistance with transport or initiate contact with a family member.

When appropriately integrated with care management systems, voice technology could also support service enquiries and routine administrative interactions.

These capabilities may be particularly valuable for people who live alone or have difficulty using conventional communication tools.

However, the usefulness of voice-enabled coordination depends on the systems and processes behind it.

A request for transport is only valuable if it reaches the appropriate service, is recorded correctly and results in a confirmed arrangement. Likewise, a request to change a care visit must be handled according to the provider's scheduling and service-delivery responsibilities.

This reinforces a broader principle explored in Integrating Home Care Technology: Connecting Systems, People and Care: integration is not simply exchanging information between systems. It is ensuring that information results in appropriate action.

4. Monitoring, Surveys and Personalised Engagement

Voice assistants can also provide a convenient way to collect information about wellbeing, symptoms and everyday experiences.

Instead of completing a written questionnaire, a person might respond verbally to questions about their condition, mood or progress towards agreed goals.

An early feasibility study discussed in eHomeCare's original articles explored this approach with 16 people living with diabetes.

Participants, aged between 12 and 72, used voice assistants over two weeks to complete daily interactions, reflect on personal goals and share experiences.

The reported findings suggested that voice interaction could encourage engagement and make some participants more comfortable discussing their experiences.

One interesting observation was that some participants appeared more willing to acknowledge when they had not achieved their goals. Others described personal experiences they might have found more difficult to discuss in conventional interactions.

This highlights a potentially valuable characteristic of voice technology: an automated interface may offer a consistent, non-judgemental opportunity for people to reflect on their circumstances.

However, the study was small and short. It did not establish improved clinical outcomes or demonstrate effectiveness across broader care populations. Participants also experienced speech-recognition and technical difficulties.

These limitations matter.

Conversational technology may help providers collect information more conveniently, but the information still needs to be interpreted appropriately.

AI could potentially identify recurring concerns or changes in responses over time. Nevertheless, such interpretations should be treated as indicators for further assessment rather than definitive conclusions.

For a broader discussion of how monitoring information becomes actionable, see AI and Sensor Monitoring: Turning Data into Better Care Decisions.

5. Independent Living and Everyday Assistance

Voice technology can support independence through its connection with smart home devices and everyday services.

Someone may use spoken commands to control lighting, adjust heating, contact another person or receive reminders about daily activities.

For individuals with mobility limitations, these functions can reduce dependence on physical controls. For people experiencing cognitive changes, carefully designed prompts may help maintain familiar routines.

This was one of the opportunities eHomeCare recognised early in the development of voice-enabled care applications.

Rather than treating a voice assistant as an isolated device, the concept was to connect everyday interactions with care activities, reminders and support services. The aim was to make assistance accessible through natural conversation rather than requiring someone to navigate multiple technologies.

That principle remains relevant, even though the underlying technology has evolved.

However, voice technology is not universally accessible.

Speech difficulties, hearing impairment, background noise, language differences and cognitive changes can affect how reliably a system understands and responds.

The individual's preferences are equally important. Some people may find voice interaction intuitive, while others consider it intrusive or frustrating.

Technology should therefore be selected and configured around individual needs rather than assumptions about age or disability.

Our article Independent Living with Dementia explores this opportunity in greater detail.

Where AI Should Assist, Not Replace, Human Judgement

The growing sophistication of conversational AI creates opportunities to automate routine interactions and reduce administrative effort.

For example, AI may help answer general enquiries, summarise conversations, identify requests requiring follow-up and support contact-centre operators.

However, automating interactions is different from allowing AI to make decisions that affect someone's safety.

Consider an emergency monitoring service receiving a call from a distressed older person.

An AI agent may be capable of recognising speech, asking questions and collecting information. Yet it may misunderstand an unusual response, fail to recognise distress or incorrectly conclude that an alert does not require escalation.

The consequences of such errors could be severe.

For critical monitoring and emergency response, AI-assisted workflows with trained human oversight remain the more prudent approach unless an autonomous system has demonstrated sufficient safety and reliability for the intended use.

This distinction should form part of every provider's assessment of conversational AI.

Privacy, Trust and Accountability

Voice-enabled care services may process sensitive information about health, daily routines and personal circumstances.

Providers therefore need to understand what information is recorded, where it is processed, how long it is retained and whether it is used to improve or train AI systems.

Consumer voice assistants should not automatically be assumed suitable for handling identifiable health information.

Australian privacy obligations, consent arrangements, cybersecurity requirements and supplier contracts must be considered before integrating these technologies into care delivery.

Providers should also establish clear responsibilities for reviewing information, correcting errors and responding when technology fails.

These considerations become more important as voice systems move beyond simple commands towards personalised conversations and automated decision support.

Evaluating Voice Technology for Care Delivery

Before investing, providers should establish which care or operational problem they want to address.

Is the objective to improve accessibility, support independence, reduce administrative work or provide additional information to care teams?

The answer determines what capabilities, safeguards and integrations are required.

Providers should evaluate usability, reliability, accessibility, privacy, interoperability and the evidence supporting supplier claims. They should also consider what happens when the technology misunderstands someone or cannot complete a requested action.

A successful implementation requires more than a convincing demonstration. It requires an operating model in which technology supports people and established responsibilities remain clear.

Through Technology Advisory, eHomeCare helps aged care and disability providers evaluate emerging technologies against their care objectives, operational requirements and long-term value.

Voice Is an Interface, Not a Care Model

Voice assistants have evolved considerably since their introduction into everyday homes.

Conversational AI is making interactions more flexible and creating new opportunities for accessibility, communication and personalised support.

However, the value of voice technology does not come from how naturally a system speaks or how convincingly it responds.

It comes from whether the technology helps people achieve meaningful outcomes while preserving safety, independence, dignity and trust.

eHomeCare helps aged care providers assess emerging technologies, evaluate supplier risks and establish the governance needed for responsible implementation. Explore our Technology Advisory services to support informed digital investment decisions.

Hubert van Dalen is the founder of eHomeCare, providing independent digital strategy, governance, technology and transformation advisory services to Australian aged care and disability providers.