Featured Lendület Researcher: Boldizsár Szentgáli-Tóth

Modern technology, and AI in particular, which has been developing at breakneck speed in recent years, is playing an increasingly significant role in our daily lives, as well as in science, the economy, healthcare and education – practically in every domain of human activity.

Creating flexible frameworks

Social life and legal relationships are no exception to this, but legislation and legal practice are not always able to keep pace with these rapid changes. “The law is a relatively conservative system of rules that is slow to change. In contrast, today’s social reality is changing extremely rapidly and is highly dynamic. These are challenges to which we must respond by creating flexible frameworks,” said Szentgáli-Tóth, head of the Momentum research group on Perspectives and Limits of Algorithmic Constitutionalism. “We can try to apply our existing legal concepts to emerging new legal relationships, but this alone does not necessarily lead to the right outcome.

Boldizsár Szentgáli-Tóth Photo: mta.hu / Tamás Szigeti

Therefore, we must also seek innovative tools that ensure new technological conditions do not alter our daily lives and that the protection of human rights is not compromised.”

The researchers of this Momentum group are contributing to this work primarily by applying the tools of constitutional law. The research group leader explained that as the world faces new challenges, both the focus of constitutional law and the discourse surrounding the field evolve accordingly. In recent years, for example, the researchers have also examined the challenges posed to human rights by the public health crisis arising from COVID-19 and the wider pandemic, as well as the possible legal responses available within this special legal framework.

It is therefore legitimate to ask what direct or indirect impact AI – and modern technology more broadly – may have on constitutionalism, the legal system, legal practice and human rights. “In some countries, artificial intelligence is already present in courtrooms, and it may soon appear in parliaments as well, where it could perform a wide range of functions,” continued Szentgáli-Tóth. “AI can, for example, prepare transcripts of hearings or parliamentary sessions, and can also produce drafts of legal documents. It is quite possible that lawyers will soon be using generative AI applications to draft documents, while judges themselves may consult artificial intelligence regarding the substance and wording of judicial decisions.”

The issue of legal certainty always arises when using AI, whether in healthcare, elderly care, or online communication.

AI can easily create fake profiles, which it can then use to post comments on social media. This brings to the fore issues related to privacy rights, freedom of speech, and the protection of personal data. The results of last year’s presidential election in Romania were annulled due to such forms of manipulation, meaning that the right to vote and the entire democratic institutional system could be compromised.

Three Responses from the Legal System

The impact of artificial intelligence on the legal system is not necessarily negative, as it can also facilitate the effective and fair administration of justice. AI can help predict likely developments and can also help in terms of a more effective allocation of resources. However, for these positive effects to prevail, the legal system must adapt to changing conditions. The law can essentially provide three types of responses to these new challenges: normative regulations can be changed, meaning new legislation can be adopted; non-binding legal norms can also be created (the so-called “soft law” approach); and changes in judicial practice can also be used to respond to the altered circumstances. In the first area, steps have already been taken, as the European Union has adopted an AI Act that seeks to regulate emerging issues through normative means, which will then need to be transposed at the national level.

The researcher noted that we are also in relatively good shape in the realm of soft law, as numerous non-binding international agreements, guidelines, and national strategies have already been developed in recent years regarding artificial intelligence. Where truly significant progress is needed is in the areas of case law and judicial interpretation, as here even the application of existing regulations from new perspectives could effectively address the issues that arise. During the first phase of the Momentum project, the research team is collecting as much data as possible on the legal documents and norms that exist in various countries and international organisations regarding the relationship between artificial intelligence and fundamental rights, as well as on the judicial responses different legal systems have developed thus far in addressing these issues.

They focus on three areas in particular:

the right to freedom of expression, the right to a fair trial, and the right to a healthy environment.

They examine the results of tests conducted so far regarding the integration of AI into judicial practice, as well as how to develop responses using legal instruments to ensure adequate protection of human rights in this new era.

According to the research group leader, AI has now become part of our lives, so we should no longer focus on whether its effects are positive or negative, or whether its expansion should be permitted on that basis. Instead, we should concentrate on how we can best coexist with it. “We could ban AI, but that would only result in it finding its way into the system through various indirect channels.

Artificial intelligence is neither good nor bad; rather, it has certain capabilities and characteristics which, if we learn to work with them effectively, can help us in many ways,”

says the constitutional lawyer. “There have been many occasions throughout human history when a new invention or technological discovery emerged, was tested, and then was gradually integrated into daily life. The same thing happened when railroads and various machines appeared during the Industrial Revolution: at first they seemed dangerous and alien, but later they found their place in the world.”

AI’s Tendency Towards Discrimination

In legal practice, according to the researcher, there are numerous repetitive, time-consuming, and labour-intensive processes in which AI could prove very useful (such as drafting records or reviewing massive legal databases). At the same time, the added value of human lawyers remains indispensable. Any documents drafted by AI must always be reviewed by a lawyer; otherwise, we could face very damaging consequences.

Since AI fundamentally makes decisions on a statistical basis, it is highly prone to discrimination, which is an issue requiring special attention in legal practice. In this sense, it is a myth that AI can render more objective judgements than humans, who are unable to free themselves from their biases. If the algorithm determines that the proportion of offenders is higher among certain social groups, it will be more likely to issue a guilty verdict for individuals from that community, regardless of the specific context. In addition, we must prevent people with better digital skills from gaining an unfair advantage in legal proceedings that utilise AI and other electronic tools. “Those who, for whatever reason, have not received technical training or do not have the appropriate tools are at a very serious disadvantage,” said Szentgáli-Tóth. “So there are both advantages and disadvantages. Our task is to identify the tools that allow us to capitalise on the advantages while at the same time eliminating the disadvantages as much as possible, or at least keeping them within manageable limits.”