Current challenges of artificial intelligence – a Budapest - Rijeka dialogue between technologists and lawyers - workshop

Workshop organised jointly by the Institute of the Information Society, University of Public Service, Budapest, and University of Rijeka Center for Artificial Intelligence.

Date: 01/25/2023 14:00–17:30
Venue: Online, (Teams) and University of Public Service, Side Building (Hybrid event)



Technology panel
14:00–14:15 Arian Skoki: Bread and Butter for Injury Prediction in Soccer
Abstract: What data can we use for training our models? How can AI help in determining increased injury risk in players and what are the biggest issues regarding data collection and processing?
14:15–14:30 Anna Maria Mihel: Overview of applied Machine Learning for estimating discharge in tidal rivers and estuaries
Abstract: Areas of tidal rivers and estuaries are affected by different forcing mechanisms, which makes hydrological processes complex, non-linear, and therefore challenging for modeling. Hydrological and hydraulic models can explain the dynamics of the river flow, but with two key drawbacks, namely the need for a large data set with a high cost of measurement and the necessary model calibration. In recent years, research aimed at predicting discharges in these areas has achieved great success with the help of machine learning (ML) methods. The advantages and limitations of the ML approach (single and hybrid models) are highlighted, along with suggestions for future research.
14:30–14:45 András Horváth: Why is understanding crucial? Intriguing failures of machine learning
Abstract: Deep Neural networks are commonly used in various tasks and enabled the solution of many practical problems. These approaches usually result sufficiently high accuracy, but the robustness and thorough investigation of failure cases of these approaches in critical applications is still under investigation. Domain shift, dataset bias and adversarial attacks pose some of the most significant challenges. In  this talk I would like to briefly introduce these problems and summarize the current state of the art solutions which can be used to mitigate them.
14:45–15:00 Pál Vadász: Some aspects of the future of AI in the field of legaltech
Abstract: The presentation is the extract of the study prepared for the EU in 2021 on the future of small legal firms and small languages. After a review of NLP technologies relevant legaltech applications are examined. Human and technical boundaries and opportunities a presented.
15:00–15:15 Q&A
15:15–15:30 Break


Law panel
15:30–15:45 Franko Hrzic: Importance of medical data in AI: Challenges and legal issues
Abstract: In artificial intelligence, one of the main focuses when developing various models and solutions is the data from which valuable information is extracted. The ill-defined or mislabeled dataset can reduce the models' efficiency to the point of the model's failure. Hence, generating high-quality datasets is imminent. In order to obtain them, especially in medicine, the help of experts is required. However, even when a high-quality dataset is obtained, making it publicly available brings additional legal challenges. This presentation will briefly discuss the challenges of obtaining a high-quality medical dataset and making it publicly available.
15:45–16:00 Ana Pošćić - Adrijana Martinović: Challenges in Teaching Legal Aspects of AI
Abstract: The presentation will explore challenges in teaching legal aspects of AI based on the development of the elective course “Artificial Intelligence, Technology and Law” at the fifth year of Integrated undergraduate and graduate university Study of Law at the Faculty of Law in Rijeka. This innovative interdisciplinary course will be implemented in cooperation with colleagues from the Faculty of Engineering in Rijeka and it aims to highlight the challenges that artificial intelligence and new technologies impose to the existing regulatory framework. Given the unprecedented impact of AI and new technologies in all spheres of life, the law has to be able to respond to the new challenges and risks of these disruptive and transformative technologies. The question what should be regulated and how, or if at all, can be addressed only with adequate knowledge and understanding of basic principles of functioning of AI, distributed ledger technologies, such as blockchain, and autonomous systems. This requires a synergy among experts in the field of engineering, computing, ICT and lawyers. It is necessary to find a proper balance in regulation, that will be able to promote benefits and curb possible risks, without stifling innovation and progress. This entails the responsibility of law faculties to prepare the students for the digital age and integrate the teaching of AI and new technologies in the curriculum. The approach taken by the Faculty of Law in Rijeka is based on an interdisciplinary perspective and collaboration with the Faculty of Engineering in Rijeka, bearing in mind that it is impossible to understand the complex and novel legal issues without proper understanding of the technological aspects, implications, operation, limits, benefits and risks of AI-related technologies. The idea is to build a course which can be taken by future lawyers and engineers alike, which will generate better mutual understanding of fundamental concepts and contribute to lively class discussions. This presentation aims to discuss possible teaching methods and teaching outcomes that would best serve the aims of this elective course. 
16:00–16:15 Kinga Sorbán: AI to counter hate speech: regulatory challenges in AI-based hate speech moderation
Abstract: The hate speech filtering - and in this context, the restriction of freedom of expression - is still widely debated in academia and policy discussions, as the constantly evolving online environment provides a rich supply of dilemmas to be solved. Prominent among these dilemmas is the use of artificial intelligence (AI) in content filtering. The use of artificial intelligence in moderation implies relying on computer algorithms to decide whether a piece of online speech is valuable or worthless, permissible or prohibited. Artificial intelligence performs particularly poorly when it comes to filtering hate speech: while Meta's Transparency Report shows that in Q4 2021, Facebook acted on 17.4 million posts and comments as hate speech, 95.90% of which was identified and classified by the platform’s AI, information leaked from whistleblower Frances Haugen suggests that 95% of hate speech shared through the service goes undetected.   
In the presentation I will examine what risks the unrestricted use of artificial intelligence poses to freedom of expression. I argue that while content-filtering AI is a necessary tool for content moderation, is not yet capable of fully replacing human effort, as its imperfect functioning could cause unforeseeable damage to the prevalence of online freedom of expression.
16:15–16:30 Q&A
16:30–16:45 Márton Domokos: Compliant integration of the AI into HR processes
Abstract: Employers have been using software to facilitate performance management and disciplinary proceedings for many years, but AI has provided new opportunities. However, automated decision-making is expected to be fair and objective, and free of any allegation of bias.
There may also be regulatory issues – besides the EU’s upcoming general AI Regulation, there are country-level rules to comply with. In France, a bill relating to the algorithmic organisation of work would require employers to demonstrate that the algorithm they use does not discriminate. Spain also published guidelines on the use of algorithmic data in a labor environment. There are legislative efforts in the USA too. Employers in Illinois must gather and report certain demographic information to analyse whether the data discloses a bias in the use of the AI. New York established that the use of automated employment decision tools are subject to an impact assessment.
The aim of the presentation is to investigate whether it will ever be acceptable culturally for a machine to decide in employment matters. Will technology ever be sophisticated enough to understand and rationalise mitigating factors in a way that a human might? How would a court view a decision made solely by a machine? How can employers integrate AI into their HR processes, to ensure consistency of approach?
16:45–17:00 András Hárs: Understanding or Hostility - Surveying the Attitude of International Law Experts towards AI
Abstract: There has been a significant amount of surveys measuring the attitude of artificial intelligence (or AI) experts in the field of computer science but the attitude of scholars and practitioners of international law have mostly been left outside of the scope of observation. It would be an important tool in aiding decision–makers to know how those who are well-versed in the field of international law and more specifically human rights law understand AI. In light of the proliferation of international agreements and recommendations by the EU, UNESCO and OECD, it is safe to assume that the opinion and attitude of legal experts will play an important role in the years to come. Through an international online survey conducted at the University of Public Service, professionals’ attitude towards AI-based technologies, their effect on human rights and possible measures taken by governments in order to sooth any possible hostility or fear will be analysed in detail. The goal is to find the common denominator between law and computer science as to provide a more comprehensive legal framework which would in turn help create a safe, transparent and human-centred regulation both on the domestic and on the international level.
17:00–17:15 András Pünkösty: How do data ethics foster correct trust in AI solutions?
Abstract: Trust is a crucial element for the operation of any complex system where a human agent acts and makes decisions. In the rapid digital transformation of our societies, AI solutions may bring excellent opportunities for certain problem matters but might pose challenges for the vulnerable human nature if inflexible technologies are not designed carefully to achieve a ‘good AI society’ (C. Cath et al., 2017.). The technological advancement resulted in the adoption of new laws to limit the risk of emerging AI technologies; however, an adequate legal definition for AI is still lacking. Nevertheless, data is a core input for machine learning approaches that are intrinsically needed and used for the new technology. Data ethics effectively contribute to vital questions of the big datasets used by AI approaches and to avoid the bias and or inaccurate patterns thereof. In my presentation, I will shortly discuss how data ethics can contribute to principled AI that deserves trust in a ‘mature information society’ (C. Cath et al.)
17:15–17:30 Q&A and wrap up (Zsolt Ződi


Bios of the presenters

Arian Skoki received MS degree in Computer Engineering from the Faculty of Engineering, at the University of Rijeka in 2019 with the Master's thesis "Automatic Music Transcription for Traditional Woodwind Instruments Sopele". After finishing his studies, he moved to Graz, Austria, and worked as a Full Stack Developer. In 2020 an opportunity came to combine work with his other passion - Sports. Therefore, he returned to Rijeka to work as a Teaching assistant at the Faculty of Engineering and pursue a Ph.D. in Computer Science. His research involves data analysis and machine learning mainly in the field of football/soccer for determining the game intensity, injury prediction, physiological, and psychological player profiling, etc.

Anna Maria Mihel was born in Levanger, Norway on March 21, 1996. She completed undergraduate and graduate studies in Computer Science at the Faculty of Engineering in Rijeka. In November 2021, she enrolls in a doctoral study in Computer Science also at the Faculty of Engineering and has been employed there as an assistant since December of 2021. 

András Horváth is an Associate Professor at the Faculty of Information Technology and Bionics at Pázmány Péter Catholic University. His research is mostly focused on computer vision and artificial intelligence, especially on the efficient implementation of modern machine learning algorithms with emerging devices. He took part in the DARPA-UPSIDE (Unconventional Processing of Signals for Intelligent Data Exploitation) project between 2012 and 2018 in a consortium with Intel, MIT, which aimed the development of an object recognition pipeline with oscillatory based computing, implemented on emerging devices (e.g.: spin-torque and resonant body oscillators) and was involved in multiple international research grants sponsored by the European Union and ONR. He is author or co-author of more than 50 publications which appeared in various international journals. He is an active Reviewer for various peer reviewed journals. (e.g.: IEEE Transaction on Signal Processing, IEEE Transactions on Circuits and System, etc.). He is a member of the IEEE Circuits and System Society and the IEEE Computational Intelligence Society and the secretary of the Cellular Nanoscale Networks and Array Computing Technical Committee. 

Pál Vadász was graduated as a mathematician in Eötvös Loránd University and has been active in the IT industry for more than 40 years. He is a senior researcher in Institute of the Information Society, Eötvös József Research Centre, University of Public Service. Besides his academic position he is the CEO of Montana Knowledge Management Ltd., a company dealing with various NLP and legaltech projects. 

Franko Hržić was born in 1993 in Rijeka. In 2022 he was promoted to the doctor of science with the "Thèse en cotutelle" from the Medical University of Graz, Austria, and the Faculty of Engineering University of Rijeka. His areas of scientific interest and research include but are not limited to machine learning, medical image analysis, explainable artificial intelligence, computer vision, and computer-aided diagnosis (CADx).

Ana Pošćić: Associate professor and Head of Department of European Public Law at the University of Rijeka, Faculty of Law as well as head of the Inter-University Centre of Excellence Opatija. Her research interests include various areas of EU law, EU competition law, internal market and consumer protection law .Co-holder of various EU law courses at the Faculty of Law in Rijeka, at undergraduate, graduate, postgraduate and doctoral law studies, including Lectures in European Law I, EU Social Security Law, Fundamentals of EU Law, Internal Market and Fundamental Freedoms, EU Labour Law, Corporate Acquisitions and Restructuring, European Company Law and Fundamental Freedoms, Competition Law and State Aid. 

Adrijana Martinović: Associate professor at the Department of European Public Law, University of Rijeka, Faculty of Law. Deputy-Head of the Jean Monnet Inter-University Centre of Excellence Opatija. Scientific interests and research include various areas of EU law, EU citizenship, social security law, competition law, and gender equality and non-discrimination law. Co-holder of various EU law courses at the Faculty of Law in Rijeka, at undergraduate, graduate, postgraduate and doctoral law studies, including European Law I, EU Social Security Law, Fundamentals of EU Law, Internal Market and Fundamental Freedoms, EU Labour Law. 

Kinga Sorbán is a research fellow at the Institute of Information Society of the University of Public Service Hungary, her main research topics are platform regulation, media law, and content-related cybercrime. Until June 2018 she worked for the National Media and Infocommunications Authority where she was responsible for representing the Authority’s standpoint related to media regulation in the European Union and international organizations. She graduated as a lawyer from Eötvös Loránd University in 2013 and received an LL.M. title as a lawyer specializing in ICT from the University of Pécs. In 2022 she defended her Ph.D. thesis on the role of internet intermediaries in policing cybercrime. In the same year, she completed the Hungarian bar exam.

Márton Domokos is a senior counsel in the commercial team at CMS Budapest focusing on data protection, cybersecurity, intellectual property, commercial transactions and the TMT sector. He is also Coordinator of the CEE Data Protection Practice. He has substantial experience in drafting and negotiating general commercial and IT contracts, advertising, sponsorship and marketing, outsourcing, distribution and franchise agreements, commercial regulatory matters as well as data protection, privacy law and internet law issues. Dr. Domokos continuously monitors regulatory developments in the field of AI. He regularly gives presentations and publishes articles on AI issues. Dr. Domokos is a co-author of the CMS Expert Guide to AI strategies in CEE. He is also an active member of several working groups of Hungary’s Artificial Intelligence Coalition and he is the president of the Data Protection Board of the Direct and Interactive Marketing Association (a FEDMA member). Dr. Domokos is recommended by Legal 500 and Chambers and Partners in the TMT category. 

András Hárs is senior lecturer at the National University of Public Service in Budapest, Hungary with 9 years of teaching experience in international law and related subjects. His field of research includes sexual exploitation and abuse in United Nations peace operations, international criminal law and artificial intelligence and international law - the former of which was the subject of his doctoral thesis, successfully defended in 2021. He has substantial experience as guest lecturer in Spain, the Netherlands, Italy, Kenya, Turkey, Cyprus, Portugal, Finland, India and Sudan among others as well as being an avid presenter in numerous domestic and international conferences. He has written extensively in his field of research in peer-reviewed journals, including an online encyclopaedia entry regarding the resolutions of international organizations for the Hungarian Academy of Sciences and being the co-author the book titled Interstate Relations, which is widely used in higher education by students of international relations. He serves on the organizing committee of the Hungarian Case-Solving Competition for International Law and he is co-organizer of the War and Peace in the 21st Century international law conference. His professional memberships and contribution to academic societies include ESIL, ILA, IBA, AYICL and the UN Association of Hungary.

András Pünkösty is a researcher at the Institute of the Information Society at the Ludovika University of Public Service and at the Competition Law Research Centre at Pázmány Péter Catholic University, where he works as an assistant professor at the European Law Department. András is an editor of the journal of the Hungarian Competition Authority (GVH) titled ‘Competition Mirror’. András worked for the GVH from 2103 till 2018, mainly dealing with merger control. His most recent publication is ‘Competition Law Sanctions in Hungary’, published by Cambridge University Press in Tihamér, Tóth (ed.) The Cambridge Handbook of Competition Law Sanctions (pp. 407-459.) András' research focuses on competition law-related aspects of digital transformation and ethical issues in the informational society.

Zsolt Ződi is a senior researcher in Institute of the Information Society, Eötvös József Research Centre, University of Public Service. He graduated as a lawyer in University of Miskolc, earned PhD in legal informatics in University of Pécs, and habilitated in Eötvös Loránd University. He had been working between 1996 and 2016 in the commercial publishing industry in different companies and positions. Recently his field of research is regulation of artificial intelligence, use of advanced technologies in lawyering (legaltech), and regulation of internet platforms. He is author of 2 books and more than 100 articles, editor of 6 books and member of the editorial board of 3 legal journals. He is the ambassador of the European Legal Technology Association in Hungary.