We develop a transparent and fully auditable LLM-based pipeline for macro-financial stress testing, combining structured prompting with optional retrieval of country fundamentals and news. The system generates machine-readable macroeconomic scenarios for the G7, which cover GDP growth, inflation, and policy rates, and are translated into portfolio losses through a factor-based mapping that enables Value-at-Risk and Expected Shortfall assessment relative to classical econometric baselines. Across models, countries, and retrieval settings, the LLMs produce coherent and country-specific stress narratives, yielding stable tail-risk amplification with limited sensitivity to retrieval choices. Comprehensive plausibility checks, scenario diagnostics, and ANOVA-based variance decomposition show that risk variation is driven primarily by portfolio composition and prompt design rather than by the retrieval mechanism. The pipeline incorporates snapshotting, deterministic modes, and hash-verified artifacts to ensure reproducibility and auditability. Overall, the results demonstrate that LLM-generated macro scenarios, when paired with transparent structure and rigorous validation, can provide a scalable and interpretable complement to traditional stress-testing frameworks.

The rapid progress of Large Language Models (LLMs) has transformed natural language processing and broadened its impact across research and society. Yet, systematic evaluation of these models, especially for languages beyond English, remains limited. "Challenging the Abilities of LAnguage Models in ITAlian" (CALAMITA) is a large-scale collaborative benchmarking initiative for Italian, coordinated under the Italian Association for Computational Linguistics. Unlike existing efforts that focus on leaderboards, CALAMITA foregrounds methodology: it federates more than 80 contributors from academia, industry, and the public sector to design, document, and evaluate a diverse collection of tasks, covering linguistic competence, commonsense reasoning, factual consistency, fairness, summarization, translation, and code generation. Through this process, we not only assembled a benchmark of over 20 tasks and almost 100 subtasks, but also established a centralized evaluation pipeline that supports heterogeneous datasets and metrics. We report results for four open-weight LLMs, highlighting systematic strengths and weaknesses across abilities, as well as challenges in task-specific evaluation. Beyond quantitative results, CALAMITA exposes methodological lessons: the necessity of fine-grained, task-representative metrics, the importance of harmonized pipelines, and the benefits and limitations of broad community engagement. CALAMITA is conceived as a rolling benchmark, enabling continuous integration of new tasks and models. This makes it both a resource -- the most comprehensive and diverse benchmark for Italian to date -- and a framework for sustainable, community-driven evaluation. We argue that this combination offers a blueprint for other languages and communities seeking inclusive and rigorous LLM evaluation practices.

Math anxiety poses significant challenges for university psychology students, affecting their career choices and overall well-being. This study employs a framework based on behavioural forma mentis networks (i.e. cognitive models that map how individuals structure their associative knowledge and emotional perceptions of concepts) to explore individual and group differences in the perception and association of concepts related to math and anxiety. We conducted 4 experiments involving psychology undergraduates from 2 samples (n1 = 70, n2 = 57) compared against GPT-simulated students (GPT-3.5: n2 = 300; GPT-4o: n4 = 300). Experiments 1, 2, and 3 employ individual-level network features to predict psychometric scores for math anxiety and its facets (observational, social and evaluational) from the Math Anxiety Scale. Experiment 4 focuses on group-level perceptions extracted from human students, GPT-3.5 and GPT-4o's networks. Results indicate that, in students, positive valence ratings and higher network degree for "anxiety", together with negative ratings for "math", can predict higher total and evaluative math anxiety. In contrast, these models do not work on GPT-based data because of differences in simulated networks and psychometric scores compared to humans. These results were also reconciled with differences found in the ways that high/low subgroups of simulated and real students framed semantically and emotionally STEM concepts. High math-anxiety students collectively framed "anxiety" in an emotionally polarising way, absent in the negative perception of low math-anxiety students. "Science" was rated positively, but contrasted against the negative perception of "math". These findings underscore the importance of understanding concept perception and associations in managing students' math anxiety.

The accelerated dissemination of disinformation often outpaces the capacity for manual fact-checking, highlighting the urgent need for Semi-Automated Fact-Checking (SAFC) systems. Within the Portuguese language context, there is a noted scarcity of publicly available datasets ( corpora) that integrate external evidence, an essential component for developing robust AFC systems, as many existing resources focus solely on classification based on intrinsic text features. This dissertation addresses this gap by developing, applying, and analyzing a methodology to enrich Portuguese news corpora (Fake.Br, COVID19.BR, MuMiN-PT) with external evidence. The approach simulates a user's verification process, employing Large Language Models (LLMs, specifically Gemini 1.5 Flash) to extract the main claim from texts and search engine APIs (Google Search API, Google FactCheck Claims Search API) to retrieve relevant external documents (evidence). Additionally, a data validation and pre-processing framework, including near-duplicate detection, is introduced to enhance the quality of the base corpora. The main results demonstrate the methodology's viability, providing enriched corpora and analyses that confirm the utility of claim extraction, the influence of original data characteristics on the process, and the positive impact of enrichment on the performance of classification models (Bertimbau and Gemini 1.5 Flash), especially with fine-tuning. This work contributes valuable resources and insights for advancing SAFC in Portuguese.

This study presents a novel integrated framework for dynamic origin-destination demand estimation (DODE) in multi-class mesoscopic network models, leveraging high-resolution satellite imagery together with conventional tra ffic data from local sensors. To extract information from imagery data, we design a computer vision pipeline for class-specific vehicle detection and map matching, generating link-level tra ffic density observations by vehicle class. Building upon this information, we formulate a computational graph-based DODE model that calibrates dynamic network states by jointly matching observed tra ffic counts and travel times from local sensors with density measurements derived from satellite imagery. To assess the accuracy and scalability of the proposed framework, we conduct a series of numerical experiments using both synthetic and real-world data. The results of out-of-sample tests demonstrate that supplementing traditional data with satellite-derived density significantly improves estimation performance, especially for links without local sensors. Real-world experiments also confirm the framework's capability to handle large-scale networks, supporting its potential for practical deployment in cities of varying sizes. Sensitivity analysis further evaluates the impact of data quality related to satellite imagery data. Introduction The widespread availability of spatio-temporal data has created new opportunities for advancing computational tools to model network flows, individual traveler behavior, and travel demand in dynamic transportation networks. Recent developments in sensing technologies and artificial intelligence are revolutionizing traditional models, making them more data-driven, scalable, and e ff ective for complex, large-scale networks. Dynamic Origin-destination Demand Estimation (DODE) is a foundational prerequisite for dynamic network models to accurately reproduce the status quo spatio-temporal network conditions, supporting tra ffic assignment (Pi et al. 2019) and control strategies (Y e et al. 2019, Liu, Ma & Qian 2023, Ke et al. 2025). DODE studies can be broadly categorized into model-based methods, which embed physics-informed tra ffic assignment models, and model-free methods, which formulate the problem using data-driven techniques without tra ffic assignment constraints.

Offensive Robot Cybersecurity introduces a groundbreaking approach by advocating for offensive security methods empowered by means of automation. It emphasizes the necessity of understanding attackers' tactics and identifying vulnerabilities in advance to develop effective defenses, thereby improving robots' security posture. This thesis leverages a decade of robotics experience, employing Machine Learning and Game Theory to streamline the vulnerability identification and exploitation process. Intrinsically, the thesis uncovers a profound connection between robotic architecture and cybersecurity, highlighting that the design and creation aspect of robotics deeply intertwines with its protection against attacks. This duality -- whereby the architecture that shapes robot behavior and capabilities also necessitates a defense mechanism through offensive and defensive cybersecurity strategies -- creates a unique equilibrium. Approaching cybersecurity with a dual perspective of defense and attack, rooted in an understanding of systems architecture, has been pivotal. Through comprehensive analysis, including ethical considerations, the development of security tools, and executing cyber attacks on robot software, hardware, and industry deployments, this thesis proposes a novel architecture for cybersecurity cognitive engines. These engines, powered by advanced game theory and machine learning, pave the way for autonomous offensive cybersecurity strategies for robots, marking a significant shift towards self-defending robotic systems. This research not only underscores the importance of offensive measures in enhancing robot cybersecurity but also sets the stage for future advancements where robots are not just resilient to cyber threats but are equipped to autonomously safeguard themselves.

In this work, we present LengClaro2023, a dataset of legal-administrative texts in Spanish. Based on the most frequently used procedures from the Spanish Social Security website, we have created for each text two simplified equivalents. The first version follows the recommendations provided by arText claro. The second version incorporates additional recommendations from plain language guidelines to explore further potential improvements in the system. The linguistic resource created in this work can be used for evaluating automatic text simplification (ATS) systems in Spanish.

Large Language Models (LLMs) are transformative not only for daily activities but also for engineering tasks. However, current evaluations of LLMs in engineering exhibit two critical shortcomings: (i) the reliance on simplified use cases, often adapted from examination materials where correctness is easily verifiable, and (ii) the use of ad hoc scenarios that insufficiently capture critical engineering competencies. Consequently, the assessment of LLMs on complex, real-world engineering problems remains largely unexplored. This paper addresses this gap by introducing a curated database comprising over 100 questions derived from authentic, production-oriented engineering scenarios, systematically designed to cover core competencies such as product design, prognosis, and diagnosis. Using this dataset, we evaluate four state-of-the-art LLMs, including both cloud-based and locally hosted instances, to systematically investigate their performance on complex engineering tasks. Our results show that LLMs demonstrate strengths in basic temporal and structural reasoning but struggle significantly with abstract reasoning, formal modeling, and context-sensitive engineering logic.

Efforts to ensure the safety of large language models (LLMs) include safety fine-tuning, evaluation, and red teaming. However, despite the widespread use of the Retrieval-Augmented Generation (RAG) framework, AI safety work focuses on standard LLMs, which means we know little about how RAG use cases change a model's safety profile. We conduct a detailed comparative analysis of RAG and non-RAG frameworks with eleven LLMs. We find that RAG can make models less safe and change their safety profile. We explore the causes of this change and find that even combinations of safe models with safe documents can cause unsafe generations. In addition, we evaluate some existing red teaming methods for RAG settings and show that they are less effective than when used for non-RAG settings. Our work highlights the need for safety research and red-teaming methods specifically tailored for RAG LLMs.

One fundamental question for the social sciences today is: how much can we trust highly complex predictive models like ChatGPT? This study tests the hypothesis that subtle changes in the structure of prompts do not produce significant variations in the classification results of sentiment polarity analysis generated by the Large Language Model GPT-4o mini. Using a dataset of 100.000 comments in Spanish on four Latin American presidents, the model classified the comments as positive, negative, or neutral on 10 occasions, varying the prompts slightly each time. The experimental methodology included exploratory and confirmatory analyses to identify significant discrepancies among classifications. The results reveal that even minor modifications to prompts such as lexical, syntactic, or modal changes, or even their lack of structure impact the classifications. In certain cases, the model produced inconsistent responses, such as mixing categories, providing unsolicited explanations, or using languages other than Spanish. Statistical analysis using Chi-square tests confirmed significant differences in most comparisons between prompts, except in one case where linguistic structures were highly similar. These findings challenge the robustness and trust of Large Language Models for classification tasks, highlighting their vulnerability to variations in instructions. Moreover, it was evident that the lack of structured grammar in prompts increases the frequency of hallucinations. The discussion underscores that trust in Large Language Models is based not only on technical performance but also on the social and institutional relationships underpinning their use.