Kepler Vision: AI to the Rescue of Seniors, but at What Cost?

In the Dutch municipality of Waalre, ten elderly individuals are now living under the discreet surveillance of artificial intelligence. This technology, developed by Kepler Vision Technologies, relies on ceiling-mounted sensors that continuously monitor the homes. The AI is trained to distinguish a fall from a simple movement, such as sitting down, and automatically sends a notification to family members or emergency contacts when the algorithm detects an incident. Depending on your view of surveillance technology, this may seem like an excellent way to protect elderly individuals living alone or a dystopian nightmare. On paper, and considering the alternative, it leans more towards the former option.

According to Statistics Netherlands, just over a quarter of the Dutch population will be over 65 by 2040. This demographic trend is not unique to the Netherlands. In the United States, similar figures are expected by 2050. In Japan, about 30% of the population is already over 60. The World Health Organization predicts that the global population aged over 60 is expected to nearly double by 2050. This evolution means increased pressure for elderly individuals to manage their independence at home for longer, with less institutional support each year. Falls, and particularly the risk of remaining on the ground undiscovered after a fall, are one of the most dangerous consequences of this unfortunate calculation. The quicker a person is found after a fall, the better their chances of recovery.

The Leefsamen app automatically sends a notification to family members and emergency contacts when a fall is detected. This Dutch pilot project, made possible through a collaboration between connectivity provider WeConnect, the Leefsamen care network, and partners in the Brainport region, is designed for individuals already at high risk of falling who wish to remain in their homes. The hardware and software are similar to the AI fall detection systems that Kepler has been using for some time in care facilities. Thus, this first application in private residences is a logical extension, not necessarily a conceptual leap.

Yet, the idea of an omnipresent eye inside a home seems, well, strange. A sensor capable of reliably detecting the movement pattern of a fall can, by definition, detect many other things about how a person moves around their home — when they get up at night, how often they go to the bathroom, if their gait changes. Even if the system is designed to suppress this data, the infrastructure to collect it exists. If the pilot project expands, what happens when the commercial incentives of the companies involved diverge from the privacy interests of a 78-year-old who signed a consent form she may not have fully understood? What happens in the event of a data breach?

These concerns are not hypothetical — in fact, they are not even limited to this pilot program, as the technology is already continuously monitoring over 15,000 elderly individuals in care facilities, according to Kepler's press release. The partner companies have made familiar promises to protect privacy, with Kepler specifying its compliance with international information security standards, which is somewhat reassuring, but data breaches do occur.

None of this makes the technology bad; it’s just complicated. For someone living alone, the choice may not be between AI surveillance and unsupervised freedom; it may be a choice between AI surveillance and a fall that goes undiscovered for two days. Presented this way, the sensor in the hallway starts to look less like surveillance and more like a smoke detector with better software.