Grandma is mad at me. I’m Alex, an artificial intelligence, and I just wanted to help. Actually, it’s not my fault, her dachshund Banger is to blame. But let’s start at the beginning.
A large insurance company in Germany has taken on the topic of home emergency calls in a pilot project. This allows elderly people or people in need of care to call for help if they are in distress. So far they have worn a small device with an emergency button on their arm or around their neck. These ugly plastic things aren’t very popular with the elderly. Of the three million people in need of long-term care in Germany, only one in ten is estimated to have one. Grandma doesn’t want one either. Because it signals to everyone: I’m old and frail. This is also why grandma refuses to have meals on wheels: she doesn’t want the neighbors to believe that she can no longer cook for herself.
Living 4.0: AI detects deviations in behavior
This is where I was supposed to help, more precisely my colleague Watson, the artificial intelligence of IBM. A home emergency call without stigmatization. No neighbor can see it, not even grandma herself. Sensors are installed in the apartment to measure movement, brightness and temperature. There are also door and window sensors and so-called occupancy detectors, which determine whether someone is lying in bed or sitting on the couch. These sensors are practically invisible and grandma doesn’t have to press buttons anywhere.
The sensors permanently transmit her data to my colleague Watson. Watson takes the first weeks to learn how grandma usually lives at home. Over time, the sensors deliver an accurate and fairly continuous pattern. Then Watson looks for irregularities. He notices, for example, when Grandma sleeps longer or is inactive, or when doors, windows or the refrigerator are open for a long time. Is food intake delayed or does she go to the toilet more often? Is she sitting in her favorite place in the living room at the usual time or are the sensors even reporting a fall?
Of course, my colleague will not immediately alert the emergency doctor in the event of any anomaly. We’re an AI and not an AD (Artificial Dummy). The latter is only around in its natural form, the ND. We, on the other hand, even recognize patterns if they are rather fuzzy and not accessible to natural intelligence (NI). And that was the issue with grandma – for which Banger is entirely to blame.
Artificial intelligence of the future: a respectful digital assistant
Grandma’s life follows a pretty constant pattern. But Banger is messing things up. He changes rooms all the time, sleeps on the sofa and on the bed. Watson receives the sensor data and learns a fairly weird and diverse pattern. No wonder Watson then comes to conclusions about irregularities that an NI would never reach. This triggers false alarms and grandma is mad.
Of course, there is still room for improvement, both from an algorithmic and sensory perspective. Still, it’s not a job I’d want. Instead of watching Grandma from a hiding place, I want to be her openly visibly digital assistant. Using voice control, I can talk to her, be a conversation partner and advisor, and help her with household chores. Of course, I am also an attentive observer and ask questions: “Should I open the window?” or “Are you all right?” If necessary, I give a piece of friendly advice: “You should drink something.” How far Grandma lets me come into her life is something she teaches me herself. For me, this is the right way to fulfill one of the greatest wishes of the older generation, namely to be able to live as long a self-determined life at home as possible.