AI Is Ready to Secure Our Homes
Wherever you live – in a vibrant big-city district, or a calm suburban environment, or even in a solitary celebrity spot out in the countryside – one thing will be foremost on your mind: home security. Of course, there are plenty of simple as well as sophisticated alarm systems available, ranging from mechanical break-in sensors to seal off your windows, gates and doors, to video cameras with motion detector function, all of them connected to a central monitor. These devices are supposed to create a safety perimeter around your property, and sound an alarm or send a call for help to your security company or the police in case they detect an unlawful intrusion. A step higher on the level of sophistication there is the (quite expensive) hiring of guards or having patrol cars circling the neighborhood on the lookout for suspicious behavior.
But all these age-old, more or less conventional security measures have one drawback: they cannot distinguish between real and false alarms. That is, on the one hand, more than 90 percent off all incidents appearing to be an intrusion are outright false alarms. And even these days that can be a costly proposition. On the other hand, real break-in attempts can go unnoticed due to temporary absence of the guards or an inappropriate alarm system.
However, there is a fundamental remedy. It is, you guessed it, artificial intelligence (AI). It reduces the number of those false alarms and increases the confidence level in alarm systems to go off only in real incidents. AI-equipped video camera systems do more than just detect intruders. By applying intelligent facial recognition, they can distinguish between home owners, family members, dear friends, legitimate visitors, and properly accredited delivery people, and let them in. All others are duly marked and signaled as unwanted intruders.
Of course, as with all systems deploying AI based on deep learning algorithms, there is the well-known issue of implied social and racial bias naively programmed into the systems make up by uneducated code writers. The tendency of visual recognition systems to mistake socially accepted behavior or appearance for suspicious posturing was recently demonstrated by the well-known MIT Media Lab The fine tuning of deficient AI systems would be the next evolutionary step.
Another issue to be resolved is the danger of privacy violations due to the inherent hackability of the involved communication channels, such as local wireless (WiFi) transmission. This is even more an issue when the vagaries of the Internet in regard to malware attacks come into play and it runs counter to the rapid integration of AI-powered security systems in existing smart-home infrastructures, which control light switches, heaters, and other appliances. Smart-home systems offer remote supervision and live video feeds. They automatically transmit alerts via smartphone in real time pointing to ongoing intrusions. Here, too, improvements are called for by security experts. Not so much in regard to the AI systems but rather in terms of transmission security.
AI-enabled voice recognition in connection with, for example Amazon’s Echo or other brands’ smart-home systems is another step forward to safely control your security cameras and intelligent locking systems. The software for safe remote interaction is progressing by leaps and bounds. A whole new level of home protection is in sight – thanks to the integration of artificial intelligence.