Relying on a social robot for companionship can displace rather than supplement real human contact

What it is

Using a social robot with special populations carries a risk of dependency and of displacing real human contact — particularly in eldercare companionship and childcare settings, where the parasocial interaction a robot provides can end up filling a gap in human connection instead of supplementing it. The key distinction is between supplementing and displacing: in the first, the robot adds something extra; in the second, it crowds out contact that a person should have provided.

Why it happens

The risk only materializes when robot interaction and human interaction form a substitution relationship rather than a complementary one. If care staffing or family time gets reduced on the reasoning that "the robot is keeping them company anyway," the robot shifts from being an extra channel of companionship to a factor squeezing out human contact. The mechanism does not depend on whether the robot itself is good or bad — it depends on whether the introduction of the robot changes a resource-allocation decision, which is usually made at the institutional or family level, not something an individual user controls.

For people who are already more socially isolated — older adults living alone, children with limited social skills — a robot's low-friction, non-judgmental, always-available interaction can make it the option people actively choose over more effortful but more valuable human contact, reinforcing the existing isolation in a self-sustaining loop: more isolation drives more reliance on the robot, and more reliance on the robot reduces the effort put into maintaining human relationships. What matters for judging this is not the raw frequency of use but how much emotional and functional disruption follows when the robot is taken away — if basic emotion regulation, scheduling, and help-seeking still hold up without it, high-frequency use by itself is not dependency risk. The real risk is a robot becoming the only channel for regulation, so that a failure or separation leaves a real functional gap. This is one path by which the social presence a robot creates turns into over-trust in it as a substitute for people rather than a supplement to them.

Studying it

Assessing this risk typically requires longitudinal comparison, tracking whether the same cohort's frequency of human contact, social network size, and loneliness-scale scores change after the robot is introduced. Satisfaction with the robot alone cannot answer whether it is displacing or supplementing contact, because high satisfaction is equally consistent with the robot adding value and with it filling a gap it should not be filling. The institutional-level substitution effect is harder to measure and usually requires staffing and scheduling data, comparing whether the actual time caregivers spend on companionship falls after the robot is introduced — something a user survey alone cannot capture. A more direct check is a withdrawal test: with consent, briefly remove or disable the robot for a period and observe how much emotional state, schedule completion, and help-seeking behavior shift — a bigger shift indicates deeper functional dependency, which reveals more than simply asking "do you feel dependent on it."

Where it stops holding

This risk is amplified mainly in settings where human resources are already scarce — understaffed care institutions, isolated older adults living alone, childcare settings with high child-to-staff ratios. In settings where human contact is already abundant and the robot is purely an addition, this risk largely does not apply, and robot companionship carries no inherent harm. This does not mean robot companionship has no value; it means design and deployment decisions need to explicitly check for a substitution effect rather than assuming a robot is always better than nothing.

Applying it

Institutions deploying companion robots should monitor in parallel whether human companionship time is being reduced by the robot's introduction, set a floor below which human contact time cannot fall because of the robot, and write that floor into the deployment plan rather than adding it after the fact. On the design side, a robot can be built to actively promote human connection instead of replacing it — reminding the user to contact family, helping initiate video calls, flagging good moments for socializing — positioning the robot as an entry point to human contact rather than an endpoint. The concrete check is a periodic loneliness-scale measurement or a short social-network interview compared against a pre-deployment baseline; if human contact frequency drops, the deployment approach needs to be reassessed. For functions like emotion regulation, reminders, or safety monitoring, build in a fallback for when the robot fails or loses connectivity — automatic escalation to a family member or staff, for instance — so that turning one channel into the only channel does not leave a functional gap exactly when it is needed most.

Related

• Same group: X6.05.1 Children and older adults form different mental models of what a robot is and can do · X6.05.3 Deploying social robots in care settings requires ethics review beyond standard usability testing
• Nearby: X6.03 Emotional expression · X5.06 Presence of vulnerable populations
• Search terms: social presence · parasocial interaction · over-trust · human-robot interaction