Naslov
Artificial intelligence as an auxiliary tool to assess the credibility of testimony

Autori
Golub, Jura

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, neobjavljeni rad, znanstveni

Skup
The role of courts and access to justice in the digital era

Mjesto i datum
Nijmegen, Nizozemska, 09.06.2022. - 10.06.2022

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Podatak o recenziji nije dostupan

Ključne riječi
artificial intelligence ; civil proceedings ; cross-border taking of evidence ; deception detection ; video conferencing

Sažetak
Aiming to strengthen judicial cooperation and promote immediacy and low costs of civil proceedings, the Evidence Regulation No. 1206/2001 opened up the possibility of using videoconferencing in taking evidence in civil and commercial matters. Moreover, its revised Regulation No. 2020/1783 introduces the use of videoconferencing as the "gold standard" for direct taking of evidence by hearing persons in another MS. ELI/UNIDROIT European Rules of Civil Procedure promote direct court hearing with physical presence for a number of valid reasons. At the same time they single out credibility of testimony, cost-effectiveness and efficiency as arguments favouring videoconferencing in cross- border disputes. In comparison to indirect taking of evidence, direct cross-border taking of evidence by videoconference significantly fosters the basic principles of civil procedure. Due attention should be given to the principle of immediacy, which requires that the same court perceives the nature and the contents of the evidence as well as decides on its probative value. In the context of oral testimony, nature of the evidence sets new requirements for the court. In addition to logical, psychological criteria should also be applied. Aiming to establish the factual situation as completely as possible court should consider gestures, colour and tone of voice, determination or indecision in testimony, security or confusion. Some researches show that 55% of communication is body language, 38% is the tone of voice, while the content or spoken words make up only 7% of communication. The application of AI systems can contribute to the assessment of testimony credibility. Available statistics indicate that AI-based systems achieve an accuracy of 85% - 90%. Research furthermore shows that in civil proceedings judges have a conservative approach to videoconferencing. It signals a risk that testimonies may be assessed less favourably if taken through videoconferencing, only because from the judge's perspective, face-to-face testimonies are more credible. If above is taken in conjunction with available studies that in merely 54% of the cases a person can independently detect deception, there are viable arguments that AI has a potential to strengthen judges' confidence in the use of videoconferencing. Considering the legal admissibility of the possible application of the AI system in the detection of emotions, the Proposal of the EU Artificial Intelligence Act accepts in principle the use of the AI in the judiciary, but with certain conditions. In addition to transparency, a number of additional requirements emphasizing the existence of human control, accuracy and resilience follow in order to minimize risks to fundamental rights and security. Regarding the question of the possible admissibility of the use of AI, the proposed Act emphasizes that the application of AI should be achieved in accordance with Charter of Fundamental Rights, secondary EU law and national law. It is of utmost importance to consider the undesirable side effects that may arise from the application of AI in assessing the credibility of testimonies. Possible violation of the right to a fair trial, the right to an effective remedy and the right to equality before the law, as well as the danger of discriminatory treatment deserve due attention. The primary danger to a fair trial is the possibility of erroneous results in assessing the credibility of testimony, which gets more pronounced if judges routinely accept AI results, absent critical and individual approach to each participant. There is also a danger of violating the principle of equality before the law if only those participants who participate by videoconference, and not those participants who are physically present at the court, would be subject to AI. Cultural differences shouldn’t be ignored, nor the fact that AI may lead to algorithmic discrimination towards certain groups.

Izvorni jezik
Engleski

Znanstvena područja
Pravo

POVEZANOST RADA