Roman law achieved legal certainty not through legislative accumulation but through structural coherence. The concepts of obligatio and bona fides ensured stability by allowing interpretative flexibility to operate within an intelligible normative framework. Contemporary legal systems, however, function under conditions of legislative inflation, vertical multilayering, and algorithmic mediation of legal information, all of which contribute to normative fragmentation. This article examines whether legislative responsibility in the age of artificial intelligence requires preserving structural coherence before interpretative flexibility. Semantic reconstruction of dispersed normative relations may strengthen legal certainty by improving the intelligibility of legal obligations; it also, however, creates risks associated with formalisation bias, technological asymmetry, and excessive reliance on computational outputs. Refraining from systematic reconstruction entails equally serious risks: fragmented regulatory structures reduce the practical accessibility of law, while interpretative authority may gradually shift towards private technological infrastructures. The article argues that structural coherence is not merely a technical requirement but a precondition for legal certainty and for maintaining the rule of law in AI-mediated legal systems.
Pozostałe osiągnięcia naukoweMonografia (zamknięty dostęp)Monograph Chapter (Conference proceedings)
This article asks what role computational tools can play in the reconstruction of a legal norm without taking over decisions reserved for the interpreter. It starts from a critical analysis of Andrzej Porębski’s study, which sets out the arguments for taking the semantic analysis of general language into account in legal interpretation. The authors accept those findings and undertake the task Porębski deliberately left aside: showing how the postulate can be carried out. The methodological model proposed here rests on a distinction between two ways of using computational tools. The first covers searching, retrieving, summarising and analysing legislation, case law and legal scholarship. The second, more far-reaching, consists in subjecting the corpus of legal texts itself to semantic processing and organising it as a representation of normative material for further reconstruction. The model draws on semantic embeddings, vector search, legal ontologies, large language models and retrieval-augmented
generation. The usefulness of this approach is demonstrated on material from a research activity funded by the National Science Centre, Poland, in which a corpus of more than 150,000 legal acts was searched using a language model trained on Polish. The study showed that the same entity is designated differently across individual normative acts, with no cross-references between them, and it further revealed terminological divergences and gaps that could not have been established by traditional methods given the volume of the material. Combining the computational results with classical doctrinal analysis made it possible to formulate conclusions addressed primarily to those responsible for shaping the regulation. These findings support a conclusion about the limits of computational assistance. Those limits are set by the material on which the tools operate, in the case analysed here the text of an editorial unit of a legal provision. The tools can prepare this resource: point to relations between provisions and mark the places where a decision is required, though that decision should always remain with the one who interprets.
Pozostałe osiągnięcia naukoweArtykuły (zamknięty dostęp)Journal article
This paper examines how IT leaders can use intelligent systems in their work, with particular attention to how these systems can support stress management in their teams. In the context of dynamic digital transformation of ICT in organizational structures, understanding leaders’ needs and the potential role of solutions based on Artificial Intelligence (AI) has become increasingly important. The main objective of the pilot study was to examine the potential use of AI-driven chatbots as supportive intelligent systems for leaders in managing team-related stress and to identify leaders’ expectations regarding AI-supported leadership. Additionally, the study identifies perceived barriers associated with the use of ICT and AI in the workplace. A quantitative pilot study involved 118 participants, including 89 people in leadership roles. The survey included the Digital Transformation Stress Scale (DTSS) and the translated version of the Unified Theory of Acceptance and Use of Technology (UTAUT). The study results showed that leaders with positive views of AI expressed lower levels of digital transformation stress (DTS) than those with negative attitudes. Additionally, IT leaders were more open to adopting AI-based solutions compared to leaders from other business sectors. Based on leaders’ opinions from the questionnaire, a functional prototype of an AI-powered chatbot was developed. This prototype uses Retrieval-Augmented Generation (RAG) to provide stress-related information and educational support designed for the needs of IT leaders.
Pozostałe osiągnięcia naukoweMonografia (zamknięty dostęp)Monograph Chapter (Conference proceedings)
Background
The primary aim was to identify combinations including ≥ 3 behavior change techniques (BCTs) that are jointly used in behavioral interventions for women with/survivors of breast cancer, and result in a simultaneous improvement of physical activity, healthy nutrition behaviour, and healthy body weight maintenance.
Methods
A meta-review integrated evidence from systematic, scoping, and pragmatic reviews. Ten databases were searched using ASReviews software and through manual searches, resulting in the inclusion of 32 reviews. The ROBIS tool was applied to assess bias risk. All combinations of ≥ 3 BCTs were identified in the original studies, and ratios of studies indicating a significant improvement to those with no improvement were calculated for three outcome domains. BCT combinations used in ≥ 10 original intervention studies (with ≥ 3 studies per each outcome domain) and supported in ≥ 85% of intervention studies were identified as the most promising.
Results
A total of k = 17,715 combinations were identified, including 3–12 BCTs. Only 46 combinations met the criteria for the most promising ones (supported by ≥ 85% of studies), with 7 combinations supported in ≥ 90% of studies. The promising combinations contained 3–6 BCTs included the following: goal setting (behaviour), problem solving, demonstration of the behaviour, self-monitoring of behaviour, instruction on how to perform the behaviour, feedback on behaviour, credible source. Interestingly, goal setting (related to behaviour or outcome) was present in 22 of the supported combinations, whereas demonstration of behaviour, problem solving, and self-monitoring were included more frequently (33, 32, and 26 of supported combinations, respectively).
Conclusions
This meta-review offers a novel approach to identifying BCT combinations that are jointly used in interventions resulting in an improvement in physical activity, healthy nutrition behaviour, and healthy body weight maintenance. These evidence-based combinations may serve as examples of good practices for promoting a healthy lifestyle in the context of breast cancer.
Chronic procrastination is increasingly targeted in psychotherapy, necessitating valid measures of procrastination in help-seeking individuals for assessment and to measure intervention outcomes. Confirmatory Factor Analysis was used in a sample of help-seeking students with problematic procrastination (n = 1461) on the Pure Procrastination Scale (PPS), Aitken Procrastination Inventory (API), and Study Problems Questionnaire (SPQ). We confirmed a three-factor solution for PPS. We did not confirm previously established factor structures for API and decided to explore a one-factor solution (API-R). Additional community student samples (n = 817 and n = 996, respectively) corroborated PPS and API-R factor structures. Invariance analyses demonstrated only configural invariance for both PPS and API-R. Researched measures were validated in a help-seeking sample, suggesting their utility in clinical studies. Current results tentatively suggest that used questionnaires capture different facets of procrastination among students seeking help for their problematic irrational delaying, compared with general student samples, indicating a need for broader research on procrastination as a clinical phenomenon.
Pozostałe osiągnięcia naukoweArtykuły (zamknięty dostęp)Journal article
