Source: Universitat Oberta de Catalunya (UOC) - news

A project by two researchers at the Universitat Oberta de Catalunya examined the impact of the European AI Regulation (AI Act) on alternative and online dispute resolution (ADR and ODR) systems and concluded that AI can significantly improve access to justice while posing substantial limitations and risks if misused. The authors — Professor Aura Esther Vilalta and Marian Gili Saldaña, affiliated to the UOC-DIGIT research centre and published open access in Computer Law & Security Review — propose a new categorization of AI-assisted dispute resolution mechanisms according to degrees of autonomy and risk, a framework that has received the Elías Campo Villegas Award from the Catalan Association of Notaries. The researchers focused on ADR and tech-mediated ODR tools such as legal chatbots, document analysis/generation systems and decision-making assistants, and noted platform automated dispute systems (for example those used by large e-commerce sites) as familiar instances of automated resolution.

The proposed four risk categories range from minimum risk (auxiliary administrative tasks like translation or transcription) through limited risk (AI that interacts with parties or generates synthetic content, e.g., complaint triage and chatbots) to high risk (AI that assists and produces results with possible legal effects, such as drafting legal documents or reviewing agreements) and unacceptable risk (fully autonomous systems that generate legal decisions without human supervision). The researchers identify benefits including speed, lower costs and greater accessibility, and warn of threats such as the "black box" effect, loss of empathy, bias and the risk to rights like the ability to be heard and to receive reasoned decisions, noting there is a real risk AI could compromise fair legal outcomes. To mitigate these risks they recommend non‑exclusively technical solutions: external audits of training datasets, assuring data and result quality, guaranteeing a human expert makes final decisions, algorithmic transparency, impact assessments before deploying tools like chatbots, operator training, and a compliance guide for administrations and dispute-resolution operators.