Recent reports have revealed that several companies are currently pushing “intelligent street lights” that are capable of being loaded with various kinds of sensors including, as Reuters reported late last month,
sensors for moisture, ambient light, seismic activity, radiation, wind, temperature, air quality, audio, MAC address tracking to detect cellphones, and yes, audio and high-def video, all joined together via a 5 gig wireless system that supports both point-to-point and mesh networks, connects to intermittent Internet backbone, and uploads the largest trove of planetary data the world has ever known up to the cloud.
That’s the vision of a company called Sensity Systems, which wants to build a “billion-node network of global sensors” based in street lights. And in another report by a local TV station we learn (via C-NET) that Las Vegas is already installing audio- and video-capable street lights made by a company called Illuminating Concepts.
Meanwhile, Nextgov.com recently reported on a contracting notice the FBI has put out for video recognition technology. The agency wants technology that can “scan crime scene footage against tapes of known people, places and objects to derive names and possible whereabouts.”
The desired connect-the-dots abilities include computing the degree of similarity among pedestrians, graffiti designs, buildings in the background of photos, and other recurring images in videos and stills….
FBI officials said they are interested in technology that would automatically "cluster" or group the images—for example, programming a computer to cluster all media with the same t-shirt logo.
A function called “tracking and re-identification,” would follow a specified individual across multiple videos to eventually help find the person’s name.
Another feature would follow certain behaviors. The "automated recognition of individuals based on behavior" would use algorithms to analyze similarities in "gait, expression, voice," and other mannerisms, according to terminology for the video analysis project.
As I have observed before, accurate interpretation of complex physical and social spaces is a very difficult artificial intelligence problem, but plenty of work is being done on video analytics, and such work promises to greatly intensify the assault on our privacy.
Both of the above stories represent different dimensions through which the privacy of our public spaces is being eroded. I see that challenge taking place along three major dimensions:
- Increased distribution and density of sensors. We are currently seeing our public spaces filled with more cameras, license plate recognition systems, RFID readers, and other sensors. All of these devices will produce oceans of data about individuals.
- Increasingly intelligent analysis of data. The FBI’s hunt for video analytics is but the tip of the iceberg here. Increasingly, the need for analysis will impel the deployment of artificial intelligence techniques to sift through all the data and make judgments about where security resources should be deployed, who should be subject to further scrutiny, etc.
- Expansion of purpose. The third dimension of surveillance-expansion is an ever-growing scope of purpose to which all this data and all this analysis is likely to be devoted. Sold as helping catch terrorists and other high-importance wrongdoings, they will inevitably expand to cover far pettier offenses, from unpaid parking fines to speed limit-violators. Not to mention political protesters and dissidents.
If allowed to continue, these three dimensions in the expansion of privacy invasions will in turn contribute to three significant dynamics that threaten our way of life:
- Automated enforcement. The increasing ubiquity of sensors in our public spaces, combined with constant analysis of their data, will increasingly lead to attempts at setting up automated law enforcement systems. An early example that we are already seeing is automated red light and speed cameras. The problems with automated enforcement include the fact that computers lack the judgment to fairly evaluate the diverse circumstances surrounding a violation, or may be susceptible to bugs or hackers. They may also fail to fairly and properly translate the state of the law into computer code.
- Total enforcement. Automated enforcement, in addition to raising issues of fairness and due process, may also make possible 100% enforcement of all our traffic laws, and many other laws as well. As I discussed in this post on “Extreme Traffic Enforcement,” we are rapidly reaching the point where this is becoming technologically possible. It may sound like a good idea at first blush, but most laws have a fair amount of “play” in them—gray areas and extenuating circumstances where human beings with good judgment would not enforce them. Total enforcement of laws would prove disruptive and problematic in many ways.
- Chilling effects. Ultimately, the fear is that we become a society riddled with rigid enforcement of petty rules, and where our public lives become weighed down by an oppressive sense of constant monitoring and control.
Right now we are in a situation where, as my colleague Catherine Crump put it, capability is driving policy, but if we do not restrain the capabilities of our technology in order to stay consistent with our values, we may find ourselves living in a country we scarcely recognize.