AI Devices in Social Care

Signed in as:

filler@godaddy.com

Account

The PRoblem

Commerical AI for Bespoke Care Needs

The UK's switch to digital phone lines could make many council telecare devices obsolete. Councils and informal carers are already struggling with costs: the UK spends £57 billion a year on unpaid care and £23 billion on social care. Cheap, widely available voice assistants and AI monitoring systems look like an easy replacement.

These systems are sold on providing independence, safety and peace of mind. But little was known about how they fit into real care, or how the way they're introduced shapes who ends up in control.

The question: can commercial AI devices reduce the burden on social care and informal carers, and what gets in the way?

My Role

User researcher and service designer

co-running a longitudinal study on a home health monitoring system with households recruited through Bristol City Council
coding and analysing interviews from a parallel longitudinal study on the Echo Show with households recruited through Bristol City Council
interviewing households before, during and after set-up, and observing onboarding sessions with technicians
analysing marketing, onboarding materials and interviews with critical discourse analysis, a method rarely applied to interview data in HCI
leading the writing of two academic papers and a policy brief

I was partnered with a PhD student who conducted the Echo Show interviews alone and then we split the interview work for the other study. We worked with one line manager.

Approach

Map the promises

I analysed the website, pamphlet and how-to guide for an AI home health monitoring system, and compared them with commercial products like Lilli and Alexa Together. Three promises dominated: independence, safety and peace of mind.

Follow onboarding as it happened

With 5 households (8 people), we interviewed people before onboarding, observed the technician session, interviewed them straight afterwards, and again after at least a week of use.

Test a cheap alternative over three months

10 households receiving social care used an Amazon Echo Show 8 alongside their existing council telecare. We interviewed them before, shortly after set-up, and at three months. We gave no instructions beyond exploring it as a care device, to see how people really adopted it.

Make sense of it in terms of care

I mapped findings against everyday care tasks, sorting what the device could help with from what it can't, and shouldn't, replace.

Translate for decision-makers

I shared findings in a policy brief for practitioners, local government and policymakers on the digital switchover, and in papers at CHI 2023 and CUI 2025.

Insights

Onboarding decided who held power, before anyone used the system

The materials were addressed to carers, and some features needed a separate phone or tablet. People receiving care were pushed into a passive role before the kit even arrived. A new hierarchy formed: the person being monitored had the least control, the provider the most.

"Peace of mind" could cause harm

Pets set off motion sensors, which could lead a carer to think a relative was wandering and to consider locking doors. Following the guidance in good faith could still hurt the person being cared for.

A cheap device helped in unexpected ways

The Echo Show helped with 5 of 6 everyday care tasks, such as managing medication and communication. Reminders let carers prompt without "nagging". One participant used it to call his carer directly after a fall.

Impact

Evidence for councils facing the digital switchover

A policy brief for practitioners, local government and policymakers.

Clear design recommendations

Write onboarding and marketing for everyone in the care network, not just carers.
Show who accessed data or changed settings, and let the person being monitored override changes.
Offer more than one way to see data and controls, such as a screen, voice or print.
Keep a backup for emergencies rather than replacing telecare outright.

Two peer-reviewed papers

At CHI 2023 and CUI 2025.