Phone calls to and from inmates are regularly recorded and monitored, but some companies are adopting AI speech-recognition technology, semantic analytics and machine learning to flag phone calls in near real-time that contain conversations indicating violence or criminal behaviour.
A firm that offers Artificial Intelligence (AI) services to U.S. prisons and jails, uses cloud-based natural language processing to build a customised lexicon based on keywords, code words and local slang. Its software identifies discussions among inmates and their outside conversation partners focusing on weapons, contraband, threats to inmates, gangs, homicides, assaults or suicide.
Investigators notify law enforcement when the system picks up suspicious language or phrases that signal criminal intent, enabling officers to take action before a problem erupts. The company recently signed a contract with the Georgia Department of Corrections for its phone monitoring transcription services, which is hosted on a web-services platform.
The phone monitoring transcription service supports “non-biased phone call analysis and transcription, enabling keyword-based searches and alerts”. It uses translation so corrections officers can easily toggle between Spanish transcripts and English translations. Investigators leverage the information the transcription service collects and help prison systems shut down criminal activity that threatens inmates, staff, and surrounding communities.
A House panel recently asked the Justice Department for a report on the use of AI to monitor prisoners’ calls with an eye toward using it in the federal arena. Local law enforcement officers were able to solve cold homicide cases after prisoners were flagged on the phone talking about committing the murders. The technology also helped prevent suicides. If the federal government starts using it, they are going to prevent a lot of inmate deaths.
However, inmates, their families and advocates say relying on AI to interpret communications opens up the system to mistakes, misunderstandings and racial bias. The advocacy group Surveillance Technology Oversight Project reported last year that the technologies’ AI platform that the New York State Department of Corrections and Community Supervision uses had the potential to automate racial profiling.
A former Secretary of the California Department of Corrections and Rehabilitation said the technology is saving lives both inside and outside of the correctional environments they monitor. Because they listen to all communications, they do not target a race, gender or protected group. Several state and local facilities, including in Alabama and Georgia, already use the technology.
As reported by OpenGov Asia, To reduce bias in AI algorithms, U.S. researchers have developed a new Artificial Intelligence(AI) programming language that can assess the fairness of algorithms more exactly, and more quickly, than available alternatives. Their Sum-Product Probabilistic Language (SPPL) is a probabilistic programming system.
Probabilistic programming is an emerging field at the intersection of programming languages and artificial intelligence that aims to make AI systems much easier to develop, with early successes in computer vision, common-sense data cleaning, and automated data modelling. Probabilistic programming languages make it much easier for programmers to define probabilistic models and carry out probabilistic inference — that is, work backwards to infer probable explanations for observed data.
SPPL gives fast, exact solutions to probabilistic inference questions. These inference results are based on SPPL programmes that encode probabilistic models of what kinds of applicants are likely, a priori, and also how to classify them. Fairness questions that SPPL can answer include “Is there a difference between the probability of recommending a loan to an immigrant and nonimmigrant applicant with the same socioeconomic status?” or “What’s the probability of a hire, given that the candidate is qualified for the job and from an underrepresented group?”.
