Justice Trends 2: Automated Justice Get the Gist of the future for technology in justice

Automating Justice

Information Technologies (IT) have developed to the point where automation is affecting knowledge sector work both positively and negatively. Here is a list of useful terms:

Machine learning: the programming of algorithms to allow a machine to “learn” from the information it is given. The more algorithms a machine is modified with and the more information it is given, the more complexity it can develop. Some machines are programmed with specific instructions for finding a certain type of answer. In deep learning, which is used by self-driving cars and photo recognition, the machine is built to replicate the human brain and filter information without specific instructions. Deep learning uses algorithms that learn from experience just as people learn from practising a taskFootnote 2. They perform tasks repeatedly, tweaking the performance a little each time to modify or improve the outcome.

Cognitive computing: a term coined by IBM’s researchers who merged cognitive science with deep learning, computer vision, and different fields to increase the complexity of the artificial intelligence. IBM's Watson, which won on Jeopardy, is the quintessential example.

Blockchain:Footnote 3 one of the digital technologies that pundits promise will have a huge impact on the legal industry. Bitcoin and other cryptocurrencies are based on some form of blockchain technology. The blockchain stores data in an immutable ledger that is decentralized from banks, governments, or any single entity. It promises improved transparency for transactions. Blockchain technology does not have to be public, but it can still improve transparency within a closed system. It can underpin smart contracts, which are systems of conditional exchanges, in a transparent and decentralized manner. It is decentralized by relying on the consensus of its users.

Smart contracts: codes that allow an exchange once pre-set conditions are met (i.e., if this happens, then that happens). Smart contracts are largely what is driving the so-called ‘Internet of Things’.Footnote 4 If a homeowner’s car is within 10km of home, then the lights turn on and the heating/ air conditioning is optimized. However, smart contracts can also act based on information provided in a blockchain to automate tasks.

Predictive policing: when law enforcement identifies criminal activity using mathematical, predictive, and analytical techniques. The Los Angeles Police Department (LAPD) has started using predictive policing to predict where crime might occur so that they can direct scarce resources to such locations. However, critics charge that the data the algorithms are using are flawed and will only reinforce racial biasFootnote 5.

Machine learning and cognitive computing have already entered the legal industry and will soon emerge more pervasively in the wider justice system. Blockchain enabled smart contracts (as tools for automation) also have a great deal of potential to affect the justice system as a whole as well as the administration of justice.

Much of the concern commonly expressed in forecasts about automation has centered on the loss of white collar jobs. This assumption relies on extrapolating the effects of automation on blue collar jobs. Technology usually helps to drop the price of performing certain tasks, but this efficiency has historically enabled society to innovate toward new types of work. Some research suggests that digital automation may actually create more employment in the knowledge sector by enabling greater productivity. What will be the implications for reduced costs and increased productivity in the justice sector? How can technology be used both to increase productivity and further protect civil rights?

Automation will affect several domains from banking to education. In the subsections below, you will see that automation is also entering police work, court and evidence management, the legal profession, and the world of crime.

  1. The current development trajectory of AI will lead to some sort of artificial super-intelligence within the century.
  2. Non-routine tasks – whether manual or cognitive – will still be done by humans while routine tasks – even cognitive ones – will be done by machines.
  3. In Canada between 1.5 million and 7.5 million jobs could be at risk of automation in the next 10 to 15 years.
  4. A quoi cela sert-il d'avoir trente avocats en "data room" alors qu'un robot peut le faire? [translation] What is the purpose of having 30 lawyers in a “data room” when a robot could do the work?
  5. If companies were to invest in AI and Human-Machine Collaboration at the same level as the top performing fifth of companies, they could boost revenues by 38 percent and lift employment levels by 10 percent between 2018 and 2022.

General Forecast for the Justice Sector

  1. Avec la simplification de la procédure pénale et de la procédure civile, avec la transformation numérique, l'organisation des juridictions devra nécessairement être repensée. [translation] With the streamlining of the criminal procedure and the civil procedure, given digital transformation, we will need to rethink how jurisdictions are organized.
  2. AI has the potential to improve aspects of the criminal justice system, including crime reporting, policing, bail, sentencing, and parole decisions…while also taking care to minimize the possibility that AI might introduce bias or inaccuracies due to deficiencies in the available data.
    1. Federal agencies that use AI-based systems to make or provide decision support for consequential decisions about individuals should take extra care to ensure the efficacy and fairness of those systems, based on evidence-based verification and validation.

Automated Policing

Predictive policingFootnote 6 has been a growing trend for the past few years. However, people concerned about human rights and fairness have raised concern about bias that may be inherent in an automated system. For such practices to be used, it is essential that users recognize and address the bias in the system.

The potential to use machines as automated decision-makers, however, might be limited at this point. These approaches may be no more accurate than human-based decisions and may even introduce further bias into the courtroom. In 2016, Pro Publica, an American nonprofit newsroom, reviewed the results of an algorithm used to predict risk and found that it was barely more effective than a coin toss: “Of those deemed likely to re-offend, 61 percent were arrested for any subsequent crimes within two years.”Footnote 7 The algorithm also displayed a disturbing level of racial bias - identifying Black defendants as more likely to commit crime and White defendants as less likely even when all other factors were taken into consideration.

However, AI can be used in other ways. The Invisible Institute’s Citizens Police Data ProjectFootnote 8 has suggested that data within a police department can be used to identify officers who abuse their authority. If police can figure out how to use new technologies to reduce crime rates, policy makers will need to consider a balance that allow for innovation while maintaining civil rights. Therefore, the forecasts in this section pertain to justice with regards to privacy, bias, profiling, and security.

  1. Now many are buying programs from tech companies like Hitachi and IBM which claim that analyzing big data can predict crime before it happens.
  2. Law-enforcement officials around the world will use AI to spot criminals, but may also snoop on ordinary citizens.
  3. The use of big data in policing has clear benefits for struggling police forces, but society needs to maintain a critical perspective on moral and ethical grounds.
    1. Using models of risk as a basis for police decision-making means that those already subject to police attention will become increasingly profiled.
  4. If police can divert resources to the right places and proceed automatically to where police and social workers need to be to help people, it would be a fundamental change in the way they approach crime and violence.
  5. Using computer models to determine where crime is most likely to occur could reinforce police biases about neighbourhoods with ethnic or racial minorities.
  6. To achieve even a 5% drop in Chicago’s homicide rate, enormous leaps in both prediction and intervention effectiveness are necessary.
  7. After two especially abominable years of mayhem, Chicago will be a somewhat safer place through 2018 and beyond.

Body Cameras

  1. New capabilities for the cameras could, paradoxically, risk undoing the confidence and trust in the community that cameras are meant to inspire.
    1. If body cams themselves undermine people's willingness to talk to cops, then imagine what it would be like if body cameras with live streaming or face recognition were implemented.
    2. Bolstered by a growing raft of additional high-tech features, the cameras could allow for a new form of high-definition surveillance, one conducted with few safeguards and little oversight.

Automated Courts

In 2018, Quebec announced that it was launching a $500-million project to modernize its legal system by digitizing all court files and records so they can be easily compiled and transferred between police, prosecutors and defence lawyersFootnote 9. The project also includes a legal resource kiosk that will assist citizens through their court process and a plan to digitize provincial inmates’ correctional files. Quebec’s digitization programs are taking shape to make court records uniform and accessible. It is hoped that digitized court files and records could be open to big data analysis and machine learning to help the courts become more efficient. The time necessary to process each case could eventually be estimated with increasing accuracy. Information for each case could be more easily searched and machine learning could enable other services. For instance, the system could recommend cases, evidence, or any indexed data to a user. These recommendations might help with case law reviews and lawyerless courts.

The building blocks for a lawyerless, online small claims court system are now being laid in the United Kingdom (UK). In 2018, the UK completed a pilot project allowing people to file divorce petitions online. Over 1,000 petitions were granted during the pilot phase; 91% of applicants expressed satisfaction; and online petitions were less likely than paper versions to be rejected for errorsFootnote 10. In the UK, algorithms and other IT make the courts more efficient. As the technology advances, a machine might eventually learn to automate a judge’s role. An AI judge could be adopted for civil cases with small claims which should free up court staff’s time for more complex cases.

Lawyers’ Use

  1. Cognitive computing could be used to suggest which arguments in court might play well or badly with a particular judge at a given time of day.

Smart Courts

  1. 85% of Britons are connected to the internet, and there is an expectation that legal services should be available online.
  2. Ethical, moral and legal risks from the growing use of algorithms are under the spotlight as the Law Society launches a public policy commission today on the impact of new technology on the justice system.
  3. The first steps towards cyberjustice will naturally involve modelling and reproducing present paper processes using electronic media.
  4. As our courts become increasingly dependent on technology the impact of any disruption to our core business will be increased.
  5. Court users will benefit from real-time case status screens, SMS and other digital notifications of session times.
  6. Information technology may play an important role in increasing access to transitional justice institutions and in facilitating communication between the institutions and their constituencies especially those in remote areas.
  7. A court system could use intelligent software agents working on behalf of their human and physical (courtroom) counterparts to automatically and intelligently examine and prioritize individual schedules and dynamically assemble a court docket. 
  8. Technology will increase the access, convenience, and ease of use of the courts for all citizens and will enhance the quality of justice by increasing the courts' ability to determine facts and reach a fair decision.

Judicial Prediction

  1. Accepteriez-vous d'être jugés par des algorithmes? [translation] Would you accept being judged by an algorithm?
    1. Est-ce que la justice est faite pour dire à quelqu'un, à partir de statistiques calculées par une machine, qu'il a toutes les chances de récidiver, ou de lui dire qu'il a la capacité de changer? [translation] Is it justice to tell someone, based on machine-calculated statistics, that they are highly likely to re-offend or that they have the ability to change?
  2. An artificial intelligence method developed by University College London computer scientists and associates has predicted the judicial decisions of the European Court of Human Rights (ECtHR) with 79% accuracy.
  3. U.S. courts and corrections departments are experimenting with algorithms to determine a defendant's risk to inform decisions about bail, sentencing, and parole.
  4. If used properly, criminal-justice algorithms offer “the chance of a generation, and perhaps a lifetime, to reform sentencing and unwind mass incarceration in a scientific way.”

Blockchain & Smart Contracts

  1. Judicial enforcement of law could be displaced by blockchain technology.
  2. The Ethereum system powering smart contracts itself envisages a dispute resolution mechanism involving external arbitrators and/or courts, where the contract is frozen pending proceedings, and the award of the court is incorporated into the terms of the smart contract.

Automated Evidence

As trends in automation, mobile technology, and eDiscovery converge, legal teams will be able to access evidence wherever they happen to be—court, client offices, etc. A 2014 report by the Technical Working Group on Biological Evidence Preservation (United States Department of Commence) noted that forensic science laboratories and law enforcement agencies are increasingly using automated identification technology (AIT), such as barcoding and radio frequency identification (RFID), to track and manage forensic evidence, firearms, and personnelFootnote 11. In future, since many forms of evidence will be digitized, machines could eventually be programmed to suggest particular evidence in real time during court proceedings, client consultations, witness interviews, etc.

  1. Technavio's analysts forecast the global ediscovery software market to grow at a CAGRFootnote 12 of 17.36% during the period 2016-2020.
  2. Increased litigation and regulation coupled with expanding use cases for eDiscovery software will continue to drive moderate growth in the worldwide eDiscovery market.
  3. The availability of reliable and effective mechanisms for admitting and displaying digital evidence will have an undoubted impact on the way evidence is gathered.
  4. As a justice system, we need to reach the point where it is expected that a police officer will give evidence by video, taking 10 minutes rather than half of their working day.
  5. New Jersey is developing software that will automatically create risk profiles of people charged with offences.

Other Evidence Technology

Digital technology is not the only technology improving evidence management and gathering. The three forecasts below are added to provide a broader view of potential changes ahead for the role of evidence in justice. With improved fingerprinting, 3D printed evidence, and DNA extrapolation, evidence could be used in new ways to advance the cause of justice. But, would the technology violate privacy or other civil rights?

  1. Fingerprint technology which can detect the brand of hair gel used by a suspect or whether they have handled a condom, food, or illegal drugs could soon be admissible in court. The technology can also detect the gender of the suspect, and it can differentiate whether the suspect has recently touched the blood of a human or an animal.
  2. Craig Venter asserts that your DNA can be used to create a photo-like reconstruction of you that will allow police to pick suspects out of a lineup using a blood spot.

Automated Attorneys

Automation has several benefits for the legal industry. Recently, a service called LawGeexFootnote 13 was shown to be faster (93 minutes for humans versus 26 seconds for the AI) and more accurate (average of 85% for humans versus 94% for the AI) than human lawyers for reviewing nondisclosure agreements (NDAs). This level of efficiency could increase a firm’s profits. Due to the efficiency and improved ways of practising law, lawyers will be able to focus on analysis and improving customer service for their clients. Likewise, they will be able to customize their prices more easily for each client to improve access to legal expertise.

  1. To truly put a law firm in the palm of one’s hand, consumers must have access to a dedicated network of law firms that exclusively serve the members of that community and has a track record of excellent performance.
  2. Looking to the Canadian legal marketplace, the following systems and applications present similar disruptive potential:
    1. cloud-based services that do intelligent deconstruction of documents to facilitate client engagement about contract creation;
    2. legal process and document production portals that enable lawyers to manage document production and document exchange between different parties;
    3. technology that enables lawyers to dispense virtual advice through expert systems in areas with risk or complexity, although the questions may be routine or repetitive;
    4. crowd sourcing and review sites where individuals choose to review companies instead of reGistering disputes;
    5. teleconferencing and web technologies for remote and online legal services;
    6. greater use of e-filing and other court initiatives such as electronic transcripts.

There is a lot of hype around technology in legal services, particularly in light of a wave of new companies touting software that will “replace the lawyer”. The reality is that technology is still a long way off from replicating the people skills, social awareness and intuition required to make a good lawyer. Rather, the technology we will see in the next 10-15 years is enabling software that helps lawyers, by making their job more efficient and automating repetitive, computable tasks – for example, through predictive electronic discovery, intelligent legal research and automated document preparation.

  1. Alors que le gouvernement s'apprête à faire un premier pas vers une justice plus numérique à travers le projet de loi présenté la semaine prochaine, un maGistrat et un chercheur au CNRS y consacrent un livre, Justice digitale... Si cette tendance inquiète une partie de la profession, des bouleversements se font déjà sentir, par exemple avec le développement de cabinets d'avocats entièrement en ligne. [translation] While the government is taking a step toward a more digital justice through the bill to be tabled next week, a judge and CNRS researcher have written a book on the subject, Justice digitale [digital justice]. And while some in the profession are worried about this trend, its effects are already being felt, for example, with the creation of fully online law forms.
    1. Le projet de loi accorde en effet une place de choix à la transformation numérique, avec le développement de la visioconférence, de la prise de rendez-vous ou saisine en ligne, et le recours grandissant aux legaltech, ces technologies numériques appliquées au monde juridique. [translation] The bill effectively puts digital transformation front and centre, with the development of videoconferences, online appointment booking, and growing reliance on legaltech—digital technologies applied to the legal world.
  2. La justice engage sa transformation numérique Objectif de cette transformation: la dématérialisation des procédures, « il faut que les citoyens puissent suivre leurs affaires directement sur Internet ». [translation] Justice is initiating its digital transformation. The objective of the transformation: to dematerialize procedures - "Citizens need to be able to monitor their business directly on the Internet."

Artificial Intelligence

  1. Blue Hill Consulting Group conducted a study that compared traditional legal research tools such as Boolean search and natural language search with the ROSS Intelligence AI-supported platform and found that ROSS had better information retrieval quality, with 40 percent more relevant authorities cited, a 30 percent reduction in research time, and an estimated business impact of $8,466 to $13,067 annual revenue increase per attorney.
  2. ROSS is not a way to replace our attorneys – it is a supplemental tool to help them move faster, learn faster, and continually improve.
  3. Only about 13 per cent of legal work will be taken over by computers within the next five years. So, AI poses less of a threat to legal jobs than some fear, but computers, left unchecked, can have a detrimental impact on the law.
  4. Although AI shows a world with immense potential in the legal arena, it would be highly impossible to replace legal practitioners who are seasoned and can think creatively based on their experience and expertise, while being able to connect with the people as well as be able to use a network to their advantage.

Efficiencies

  1. By 2020, 15% of low-tier, billable legal work will be replaced by smart machines powered by data analytics platforms.
  2. 114,000 legal jobs will likely to be automated in the next 20 years.

Concerns for Justice

  1. The State Department foresees privacy concerns, safety of autonomous vehicles, and AI’s impact on long-term employment trends as AI-related policy areas to watch in the international context.

Blockchain & Smart Contracts

  1. In the future, we are going to hire hackers to look over a smart contract just like we hire lawyers to look over a contract today.
  2. Blockchain based smart contracts have been increasingly deployed across the finance and property sectors in the last two years and even more widespread adoption is expected in the coming years as greater functionality and common standards emerge.

Automated Crime

Because machine learning works best with large quantities of data, much attention and concern has been centered on data monopolies such as Facebook and Google and how they use or plan to use this data. Another concern should be data criminals who use big data solutions for illegal and unethical purposes. According to an article in Datanami, “cybercriminals are increasingly using advanced analytic tools and techniques to more efficiently mine and monetize stolen data”Footnote 14. Criminals are already using automation for online crimes, and they will likely find new ways to exploit data and therefore exploit their victims in both the digital and physical world. As we develop the capacity for AI to be more convincingly human, the potential criminal implications abound, from weaponizing AI and phishing the phone book with an AI avatar to finding loopholes in the law and police avoidance.

  1. Policymakers should collaborate closely with technical researchers to investigate, prevent, and mitigate potential malicious uses of AI.
  2. The use of AI to automate tasks involved in carrying out attacks with drones and other physical systems (e.g. through the deployment of autonomous weapons systems) may expand the threats associated with these attacks. We also expect novel attacks that subvert cyber-physical systems (e.g. causing autonomous vehicles to crash) or involve physical systems that it would be infeasible to direct remotely (e.g. a swarm of thousands of micro-drones).
  3. The use of AI to automate tasks involved in surveillance (e.g. analysing mass-collected data), persuasion (e.g. creating targeted propaganda), and deception (e.g. manipulating videos) may expand threats associated with privacy invasion and social manipulation. We also expect novel attacks that take advantage of an improved capacity to analyse human behaviours, moods, and beliefs on the basis of available data.
  4. A research fellow at Yale University’s Information Society Project is working on a paper that argues robots could be morally responsible and be held criminally liable for their actions, and therefore be subject to “punishment”.
  5. The United States and Europe are ill-prepared for the coming wave of "deep fakes" that artificial intelligence could unleash.
  6. To get ahead of the problem, policymakers in Europe and the United States should focus on the coming wave of disruptive technologies.
  7. Fueled by advances in artificial intelligence and decentralized computing, the next generation of disinformation promises to be even more sophisticated.

AI for Virtual Attacks

  1. The costs of attacks may be lowered by the scalable use of AI systems to complete tasks that would ordinarily require human labor, intelligence and expertise. A natural effect would be to expand the set of actors who can carry out particular attacks, the rate at which they can carry out these attacks, and the set of potential targets.
  2. New attacks may arise through the use of AI systems to complete tasks that would be otherwise impractical for humans, and malicious actors may exploit the vulnerabilities of AI systems deployed by defenders.
  3. Attacks enabled by the growing use of AI will likely be especially effective, finely targeted, difficult to attribute, and likely to exploit vulnerabilities in AI systems.
  4. The use of AI to automate tasks involved in carrying out cyberattacks will alleviate the existing tradeoff between the scale and efficacy of attacks which may expand the threat associated with labor-intensive cyberattacks (such as spear phishing). We also expect novel attacks that exploit human vulnerabilities (e.g. through the use of speech synthesis for impersonation), existing software vulnerabilities (e.g. through automated hacking), or the vulnerabilities of AI systems (e.g. through adversarial examples and data poisoning).
  5. Cyberattacks powered by artificial intelligence will make prevention more difficult.
  6. Advances in computing power and in theoretical and practical concepts in AI research, as well as breakthroughs in cybersecurity, promise that machine-learning algorithms and techniques will be a key part of cyberdefence - and possibly even attack.
  7. [Sixty-two percent] of information-security professionals surveyed by Cylance at Black Hat USA 2017 think that hackers will weaponise AI, and begin using it offensively in 2018.