Would AI gun dectectors protect US schools?

Tuesday, August 7, 2018
by Cameron Shelley

In a recent piece in FastCompany, Jeremy Hsu discusses the pros and cons of a system called Shielded Students, which relies on a high-tech gun detection system to prevent shooters from getting into schools.

A central component of the system is a microwave radar scanner that takes images of people who pass in front of its sensors.  Those images are compared to a library of profiles of guns of various types.  Enough similarity between a scan and a library profile results in a match and, thus, an alert.

Readers of this blog will likely recognize that Shielded Students raises issues similar to those raised by the ShotSpotter gunfire detection system, discussed in a recent post

As Hsu points out, such a system can commit two types of errors:

  1. False positives: Non-guns are flagged as guns, leading to false alarms;
  2. False negatives: Guns are not recognized as such, allowing illicit arms into schools.

As noted in the article, many false alarms would quickly lead to people routinely ignoring alarms, as people ignore most car alarms.  False negatives, of course, defeat the purpose of the system.

As I pointed out in connection with ShotSpotter, at a given level of accuracy, there is a trade-off between errors, such that the fewer of one kind the system allows, the more of the other kind occur.  As purchasers of Shielded Students are likely to be intolerant of false negatives, false positives will likely become the main issue for them. 

It may be very difficult for the system to reach a level of accuracy necessary to reconcile those affected to the number of false alarms they will get.

Another point would be assessing the effectiveness of the system.  While too common, school shootings are still rare.  As Hsu notes, their frequency has been declining since the 1990s.  As such, it will be difficult to tell whether or not an installed system is actually preventing shootings.  In the absence of compelling evidence of improved safety, people may tire of paying to operate a system. 

Gated communities in the US are typically designed with guardhouses but guards are often laid off in order to save money once it becomes clear that the threat to residents is quite small. 

It is also reasonable to ask whether or not money allocated to high-tech gear might be better spent elsewhere.  Better mental health or community outreach programs might prove more cost-effective in this regard, and would provoke less of a siege mentality around schools.  Increasing security at a target might be the obvious way to increase safety but, as in the case of seat belts on school busses, appearances may be deceiving.

There is no doubt that schools need security.  Whether or not Shielded Students is a good way to provide it remains up for debate.