Replacing prisons with high tech systems capable of detaining prisoners in their own homes and the use of artificial intelligence to predict and prevent imminent offenses may sound the stuff of science fiction, but rapid advances in technology surrounding IoT makes such a vision a possibility worth discussing.
A system that effectively turns prisoners into internet nodes using IoT wearables with the ability to deliver electric shocks would have significant social impact. These ramifications need to be taken into account along with engineering design and legal considerations.
This application of IoT falls at the intersection of engineering, technology and law, and as such, needs an interdisciplinary approach.
The case for technological incarceration
Big data, IoT, and AI can be useful in reforming and improving certain aspects of the legal system. One candidate is the prison system, which has remained largely unchanged for hundreds of years.
In Australia and many other Western countries, the rate of incarceration is increasing as governments respond to voter pressure to be tough on crime. This comes at a high social and financial cost.
In the US, the cost of running prisons is tens of billions of dollars per year. In Australia the annual cost is in the billions. It would actually be cheaper (although not practical) to assign an individual police officer to each prisoner.
For prisoners, incarceration causes effects following release including diminished life expectancy, prolonged unemployment and reduced income. This leads to further costs to the public purse. In addition, a disproportionate number of underprivileged and minority groups are imprisoned, including Indigenous Australians and African Americans.
One of the main arguments for incarceration is to deter people from committing crimes. Research has shown that a more effective deterrent than fear of prison is the belief by a potential criminal that their crime is likely to be detected, and that prisoners with a harsh sentence reoffend at a marginally higher rate than those dealt with leniently.
Protection of the community through incarceration of violent criminals is also limited to the length of sentences.
How could technological incarceration work?
Technological incarceration has the potential to punish criminals and keep the community safe while reducing the financial and social costs of traditional incarceration.
One proposal is to implement a variant of home detention which uses electronic bracelets or anklets along with an IoT system to achieve:
- real-time tracking of offenders’ locations
- constant surveillance of offenders’ actions
- immediate immobilisation of offenders who are committing a crime or escaping
One challenge of technological incarceration is that GPS tracking with wearables is not an adequate substitute for prison because it cannot prevent offenders from harming others in their location or if they escape. To solve this issues, the wearables need to be able to report to a central location in real-time.
For constant surveillance, and prevention of harm to the public, the cost of corrections officers viewing CCTV for every offender is too expensive. Therefore a computer-monitoring solution needs to be found.
The final challenge is how to immobilise offenders who are reoffending or escaping. This could be achieved by incorporating a device such as a taser into the offender’s anklet, which could be remotely activated if incapacitation was required.
Technological incarceration could be perceived as “soft” by the community, and education might be needed to convince the public that deprivation of liberty is a harsh punishment in itself. Conversely, some may see it as too harsh, due to complete loss of privacy and the risks of tasering. It could be argued that these concerns are not as great as the current ramifications of traditional incarceration.
Technological incarceration would also place a burden on families, be vulnerable to technological failure, and present privacy concerns to family members and engineers and technicians involved in maintenance of the incarceration equipment.
An important question is the number of technology triggered taser-related deaths, or failures of tasers leading to public danger that society is willing to tolerate, similar to the issues of driverless vehicle-caused fatalities and casualties. This needs to be put in context with current issues including deaths and violent attacks in prisons, and crimes committed by offenders on bail.
Another question is whether technological incarceration would be made available to every offender, or only those who are not violent or dangerous.
As the offenders would be imprisoned in their own homes, provisions would also have to be made for accommodation for homeless offenders.
The electronic anklet is existing technology. There are two forms: one uses RF tracking capability and the other GPS. The GPS version has the capability to accurately track offenders to within around 10 centimeters. They are fitted with an alarm for tampering, and cost around a sixth to a tenth of traditional imprisonment.
In existing devices, fibre optic technology is used to provide tamper-proofing: a beam is interrupted when offenders try to remove their device.
However, this technology is only used currently for offenders on parole or with a non-custodial sentence.
To solve the more complex problem of monitoring and incapacitating offenders in real time if they are posing a danger to others, proponents of technological incarceration have proposed the use of sensor vests in conjunction with computer-based monitoring with technologies such as machine vision.
Rather than installing fixed sensors (infrared temperature sensors (IRT), audio sensors and cameras) in offenders’ homes, these sensors could be installed in modified police vests. This has already been trialled with cameras in vests to provide police accountability.
Machine vision has the potential to detect suspicious movements such as fast hand and leg movement, or picking up implements.There is also a lot of promise in using sensors and machine vision interpretation with convolutional neural networks (ConvNets or CNNs) which have proven effective in image recognition and classification in driverless cars and robot vision.
One issue is the transmission of sensor data (particularly high definition video) in real time for analysis. This could be resolved by analysing the data locally on the vest, and transmitting interpretations, however, it is yet to be determined if available interpretation technology is small enough to be mobile.
Another area for further investigation is how integrated audio, visual and other sensor data can be used to gain a picture of the offender's activities than high definition video alone. Biosensors (which are used in the monitoring of athlete’s condition) could also be used to monitor offenders’ emotional state.
Stable communications are also necessary for the transmission of real time data and triggering of tasers. This would require a reliable 4G, or preferably signal in the offender’s home. If the data connection is lost, police officers would need to be called in. This is another argument for only using technological incarceration for lower risk offenders.
Low battery charge levels on the tasering device would also trigger a police visit.
Facial recognition technology also has the potential to allow monitoring of the gradual reintegration of offenders into society after their sentence has been served.
Technological incarceration using IoT systems is feasible, but its implementation is limited by social and legal concerns and challenges.
Once these challenges and concerns have been addressed, it might be possible to trial technological incarceration on less dangerous offenders (elderly, female and white collar) in controlled conditions.
If society does go down the path of technological incarceration, it is unlikely that people would be completely removed from offender management. In the case of a suspicious movement, an alarm could alert corrections officers and provide them with a visual feed to make a decision on the appropriate response.
Once the technology has been proven, it might be possible to hand over more control of the response to the AI system, in a similar way that we are now allowing driverless cars to make judgement calls on the road.
The manufacturing and supply of devices that could be used in technological incarceration is primarily based in the US at the moment, but there is potential for it to expand to Australia and other nations if society accepts its implementation.
The content on this page was primarily derived from the following:
Webinar titled “The Internet of Incarceration” by Professor Dan Hunter, Dean, Swinburne Law School