Artificial intelligence agency and two methods
- 1. ENCJ Lisbon, March 28, 2019 Filippo Donati Consiglio Superiore della Magistratura (CSM) 1
- 2. Outline AI in the EU AI in Italy Outstanding issues
- 3. Artificial Intelligence (AI) refers to any machine or algorithm that is capable of observing its environment, learning and, based on the knowledge and experience gained, taking intelligence action or proposing decisions. There are many different technologies that fall under this broad AI definition. At the moment, “Machine Learning” techniques are the most widely used. 3
- 4. The methodological developments in AI date back more than 50 years but the recent breakthrough has come from advances in computing power, availability of data, and complexity of the algorithms. The many applications of AI are entering into our everyday lives, from filtering spam, to machine translations, image recognition and music generation, and are increasingly used in industry, government and commerce. We are probably only at the beginning of this process because the development of ubiquitous sensor networks, and the IoT, will increase exponentially the sensing capabilities of AI, the volumes of data available to train the algorithms and the reach of AI through decisions and actions. The opportunities are many, and in some cases not yet foreseen, but so are the challenges.
- 5. 5 Many national strategies and programme (“Villani Report 2018” in France; “Robotics and Artificial Intelligence” Report 2016 in United Kingdom) Cooperation among EU member States and the European Commission (on April 2018, the EU Member States and Norway signed a “Declaration of cooperation on Artificial Intelligence” in which they agreed to work together on the most important issues raised by AI, from ensuring Europe’s competitiveness in the research and deployment of AI to dealing with social, economic, ethical and legal questions). Several member States (Finland, beyond France and United Kingdom) already have AI strategies in place or are in the process of adopting them.
- 6. 6 Most national strategies have similar aims and objectives: embracing AI to become a leading country in the field, supporting innovation and business, promoting the use of AI in the public sector, establishing one or more national centers of excellence and developing strategies to promote the sharing and use of data more widely between the public and private sectors. They also focus on applications aimed at modernizing public administrations, as well as specific sectors as health.
- 7. Communication from the European Commission to the Parliament and the Council on Artificial Intelligence for Europe (2018) sets out a EU initiative on AI, which aims to: - promote the EU's technological and industrial capacity and AI uptake across the economy - prepare for socio-economic changes brought about by AI - ensure an appropriate ethical and legal framework, based on the Union's values and in line with the Charter of Fundamental Rights of the EU.
- 8. Council of Europe initiatives Recommendation (2017) of the Parliamentary Assembly about Technological convergence, artificial intelligence and human rights European Ethical Charter on the use of artificial intelligence in judicial systems and their environment (2018),
- 9. Importance of the Charter, the first European instrument to set out (five) methodological principles that apply to the automated processing of judicial decisions and data based on AI techniques. Developed by the Council of Europe Commission for the efficiency of justice (CEPEJ), it is aimed at private companies (start-ups active on the market of new technologies applied to legal services), public actors in charge of designing and deploying AI tools and services in this field, public decision-makers in charge of the legislative or regulatory framework, as well as legal professionals. The Ethical Charter is already a reference point for applications of AI in the judicial sector
- 10. Italy For the time being judges in Italy do not make any practical and daily use of predictive software or AI tools However, AI tools would likely meet the needs of judges: AI offers a new tool for calculating scales (divorce, dismissal, compensation for personal injury ectc.) through a probabilistic or actuarial approach, offering quick and reliable results AI might contribute to the efficiency and quality of justice in terms of duration of court cases and uniformity of case law 10
- 11. MoCAM system Software developed in 2007 for calculating scales in family cases: amount of child support in cases of separation, divorce or rupture of a de facto union, contribution due in favor of the spouse The software is mainly directed to citizens as a guidance concerning their rights in case of nuptial crisis 11
- 12. 12
- 13. The system requires Registration on the portal Payment of a fee Provision of data Algorithm for calculating the scales is (pre) defined by the programmer No machine learning activity, no big data analytics, no case law automatic research 13
- 14. 14
- 15. Use of Mocam by the Court of Florence both for facilitating dispute settlements and for calculating scales Shortcomings of the system Not tailored in order to address the peculiarities of specific cases Example: the algorithm considers the time spent by children with each parent in terms of quantity ony, without taking into consideration the quality of tasks performed by each parent during that relevant period 15
- 16. Outcome depending on the formula defined by the programmer No consideration of case-law evolution Need of a case by case balancing test of the interests to be taken into account Supreme Court (Corte di Cassazione): scales for the allowances due to spouses and children to be determined through an overall assessment off all relevant circumstances and a balancing of the legal principles, not on the base of a mere mathematical criteria 16
- 17. Initiatives of certain Italian Courts, aimed to providing citizens with guidance on possible duration and probable outcomes of judicial disputes Transparency and predictability tools may encourage out-of-court settlements and limit un-necessary recourse to justice 17
- 18. 18
- 19. Provision to any interested subject, through the Court’s web site, of consolidated case law reports with respect to given matters, in order to facilitate predictability ot the outcome of potential judicial disputes on such matters Reports prepared by trainees under supervision of Court Judges Reports useful for judges and for citizens in terms of transparency and predictability Shortcomings Peculiarities of each case to be taken into account Need of constant updates with respect to case-law evolution (including case law of other Courts, Supreme Court and international Courts)
- 20. 20 AI systems might in the future help to overcome some of the above shortcomings For the time being, the introduction in the Italian judicial system of AI predictive tools seems problematic
- 21. Very advanced stage of digitalization in Italian civil court proceedings The Civil Telematic Process is a project initiated by the Ministry of Justice aimed at improving the quality of judicial services in the civil law sector. A new technological architecture enables online remote execution of operations (such as document filing, transmission of communications and notifications, consultation of the proceedings status using the registry held by the chancery, consultation of the files and case law), which previously were only available by physically visiting the Court chancery Complete data base of Supreme Court decisions available however Criminal Telematic Process not ready Data base of ordinary court decisions not available
- 22. 22 Legal issues Article 102 Italian Constitution: Judicial proceedings are exercised by ordinary magistrates empowered and regulated by the provisions concerning the Judiciary Article 6 ECHR: right to a fair and public hearing within a reasonable time by an independent and impartial tribunal established by law Art. 22 GDPR: The data subject shall have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly significantly affects him or her
- 23. 23 Italian Constitutional Court, decision 187/2018: compensation for unlawful dismissal automatically (pre) determined by the law in relation to length of service alone is unconstitutional In the opinion of the Constitutional Court, the calculation of the amount of damages due to unlawfully dismissed workers must be made by a judge taking into account many criteria, such as number of employees, size of the business, conduct and conditions of the parties etc. Principle of judge’s self-determination
- 24. principles of the Ethical Charter on the Use of Artificial Intelligence in Judicial Systems to be guaranteed Principle of respect for fundamental rights (design and implementation of AI tools and services to be compatible with fundamental rights) Principle of non-discrimination Principle of quality and security Principle “under user control” (users as informed actors and in control of the choices made) Principle of transparency, impartiality and fairness: make data processing methods accessible and understandable, authorise external audits
- 25. 25 automatic decisions making or scale definition require transparency on the criteria used by the algorithm TAR Lazio, sez. III bis, sent. 14 febbraio 2017, n. 3769: the public administration may use softwares for the evaluation of applicants to a given position, provided that each interested party is granted full access to the logic followed by the algorithm AI algorithms based on machine learning methods as a “black box”: we can access the inputs and outputs, but do not understand fully what happens in-between, and how certain outputs, including decisions and actions, are derived
- 26. Final remarks widespread acknowledgement that AI might contribute to the efficiency and quality of justice Technical and legal issues to be addressed 26
Editor's Notes
- #1: 1) © Wikimedia Commons/Ssolbergj 2) © fotolia/FotolEdhar Approfondimenti: L'Unione europea conta nove regioni ultraperiferiche, geograficamente molto distanti dal continente europeo: http://ec.europa.eu/regional_policy/it/policy/themes/outermost-regions/