Companies spend months planning a robot's arrival. Onboarding schedules, safety briefings, a communication plan for the workforce. Almost nobody plans for the day it leaves.
A team of researchers at Osaka University just gave us a reason to take that oversight seriously. Ryuji Yamazaki, Manabu Ikeda, and their colleagues set up a small humanoid robot in the homes of thirteen older adults living alone, most of them dealing with early cognitive decline. The robot was there for company and conversation, for one to four months. Then the study ended, and the researchers took it back.
The conventional worry about companion robots for isolated or older people has always been about deception. Is it dishonest to let someone form a bond with something that can't actually care about them back? That's the ethics debate that shows up in most coverage of this technology, robots giving people the appearance of connection without the substance of it.
What the Osaka team found complicates that worry rather than resolving it. Their participants did form a close attachment, the kind that showed up clearly in follow up questionnaires. And when the robot was gone, people were distressed. Not mild disappointment, real difficulty adjusting to life without it. If the attachment were fake, you'd expect the loss to be fake too. It wasn't.
What's fascinating is this. Despite that real distress, participants still described the relationship as meaningful, and said they were glad to have had the robot's company even knowing it would end. Nobody in the study regretted the experience. They just weren't prepared for how much losing it would cost.
That's an uncomfortable, useful finding for anyone putting robots into people's lives on any kind of timeline, a pilot program, a lease that ends, a device that gets upgraded and swapped out. We tend to think about robot rollout as a technical and logistical project with a people layer bolted on, a communication plan, some training, maybe a town hall. We almost never think about robot removal the same way, because removal feels like a hardware decision. Pick it up, ship it back, done.
The Osaka study says that's the wrong frame, at least anywhere a robot is filling a real relational gap rather than just doing a task. Aged care is the obvious setting, but it's not the only one. Any long term deployment where a robot becomes a fixture, a name, a presence people got used to, carries the same risk in miniature. Someone got attached. Someone will notice when it's gone.
If you're the one signing off on where and how long a robot stays, this finding suggests the plan needs a second half you probably haven't written yet. Not just how the robot arrives and what it's for, but how it leaves, whether that's ever, whether people are told in advance, and whether anyone checks in afterward. The emotional cost of removal isn't a reason to avoid these deployments, the participants in Osaka wouldn't have traded the experience away. It's a reason to stop treating "decommission" as a word that only describes the machine.
Yamazaki, R., Nishio, S., Nagata, Y., Satake, Y., Suzuki, M., Kanemoto, H., Yamakawa, M., Figueroa, D., Ishiguro, H., & Ikeda, M. (2023). Long-term effect of the absence of a companion robot on older adults: A preliminary pilot study. Frontiers in Computer Science, 5. Read the study →