Twenty-eight South-East Queensland long-term care facilities
Verified evidenceCreator to judgeHinge: a responsive social robot can provide repeatable short engagement sessions as an adjunct to care, without replacing clinical assessment or human relationships.
Customer service · Queensland, Australia
Collections: Human still decides · Embodied work
Executive brief
A humane robotics case should report comparator-sensitive and null findings, not only engagement gains.
AI value · Observed visual and verbal engagement
VerifiedPARO participants were significantly more verbally and visually engaged than plush-toy participants.
Engagement effects do not establish broad clinical improvement.
Before
Provides usual care and available psychosocial activities. → Observes mood, engagement, and agitation.
After
Participates in an individual non-facilitated 15-minute responsive-robot session three times weekly. → Interprets response and adjusts care; stops use if distress or infection-control concerns arise.
Human boundary
Care staff and clinicians decide suitability, continuation, and treatment.
Why it matters
Hinge: a responsive social robot can provide repeatable short engagement sessions as an adjunct to care.
Before
How the work ran before the change.
Care staff
Provides usual care and available psychosocial activities.
Care staff
Observes mood, engagement, and agitation.
What changed
Hinge: a responsive social robot can provide repeatable short engagement sessions as an adjunct to care.
Decision rightHuman moves from creator to judge
After
How the same work runs now.
Resident with PARO
Participates in an individual non-facilitated 15-minute responsive-robot session three times weekly.
ControlTen-week protocol and facility supervision.
Care staff
Interprets response and adjusts care; stops use if distress or infection-control concerns arise.
ControlHuman care authority.
Exception path
Stop or substitute another activity for distress, non-engagement, infection risk, or clinical concern.
Work removed
Decision authority
Care staff and clinicians decide suitability, continuation, and treatment.
Before
Care staff
Provides usual care and available psychosocial activities.
Control: Care plan and staff observation.
Care staff
Observes mood, engagement, and agitation.
Control: Human assessment.
After
Resident with PARO
Participates in an individual non-facilitated 15-minute responsive-robot session three times weekly.
Control: Ten-week protocol and facility supervision.
Care staff
Interprets response and adjusts care; stops use if distress or infection-control concerns arise.
Control: Human care authority.
Work that left the path
Human role before
Staff delivered and observed psychosocial support.
Human role after
Staff select appropriate residents, monitor response, and integrate a robotic adjunct into care.
AI roleDecision mode: responsive therapeutic interaction. PARO reacts to touch and voice; it does not diagnose or prescribe.
Look-alike plush-toy sessions.→PARO participants were significantly more verbally and visually engaged than plush-toy participants.
Ten-week cluster-randomized intervention with follow-up. · 415 residents across 28 facilities.
Engagement effects do not establish broad clinical improvement.
Usual care and plush-toy comparison.→Video observation favored PARO over usual care, but CMAI-SF showed no significant group difference and PARO was not consistently superior to plush toy.
Ten-week intervention. · 415 residents.
Mixed outcome measures make this a qualified, not uniformly positive, case.
Anti-pattern
Using a social robot to reduce human contact or overstating mixed agitation results.
Questions
Portability conditions
Reputation risk
medium
Evidence and authority
Watch · updated
Watch status: verify the cited source and deployment condition before reusing this case.
1 peer reviewed; publication outcomes are verified.
Bundle 1.0.0 · reviewed 2026-08-23 · stable ID 5fd1310c888e85c2
IKEA (Ingka Group) · Creator to judge
Human agents must answer routine queries instead of focusing on high-value sales and design advisory.
Octopus Energy · Creator to judge
Human agents must draft responses to all customer emails.
Google Research Project Relate · Creator to judge
People with non-standard speech need not rely only on listener familiarity, repetition, or text entry.
Google · Creator to judge
Large migrations can combine machine-found locations, LLM edits, automated checks, and engineer review.