This article investigates gender bias in narratives generated by Large Language Models (LLMs) through a two-phase study. Building on our existing work in narrative generation, we employ a structured methodology to analyze the influence of protagonist gender on both the generation and classification of fictional stories. In Phase 1, factual narratives were generated using six LLMs, guided by predefined narrative structures (Hero's Journey and Heroine's Journey). Gender bias was quantified through specialized metrics and statistical analyses, revealing significant disparities in protagonist gender distribution and associations with narrative archetypes. In Phase 2, counterfactual narratives were constructed by altering the protagonists’ genders while preserving all other narrative elements. These narratives were then classified by the same LLMs to assess how gender influences their interpretation of narrative structures. Results indicate that LLMs exhibit difficulty in disentangling the protagonist's gender from the narrative structure, often using gender as a heuristic to classify stories. Male protagonists in emotionally driven narratives were frequently misclassified as following the Heroine's Journey, while female protagonists in logic-driven conflicts were misclassified as adhering to the Hero's Journey. These findings provide empirical evidence of embedded gender biases in LLM-generated narratives, highlighting the need for bias mitigation strategies in AI-driven storytelling to promote diversity and inclusivity in computational narrative generation.
LINK
Post-training quantization reduces the computational demand of Large Language Models (LLMs) but can weaken some of their capabilities. Since LLM abilities emerge with scale, smaller LLMs are more sensitive to quantization. In this paper, we explore how quantization affects smaller LLMs’ ability to perform retrieval-augmented generation (RAG), specifically in longer contexts. We chose personalization for evaluation because it is a challenging domain to perform using RAG as it requires long-context reasoning over multiple documents. We compare the original FP16 and the quantized INT4 performance of multiple 7B and 8B LLMs on two tasks while progressively increasing the number of retrieved documents to test how quantized models fare against longer contexts. To better understand the effect of retrieval, we evaluate three retrieval models in our experiments. Our findings reveal that if a 7B LLM performs the task well, quantization does not impair its performance and long-context reasoning capabilities. We conclude that it is possible to utilize RAG with quantized smaller LLMs.
MULTIFILE
With the proliferation of misinformation on the web, automatic methods for detecting misinformation are becoming an increasingly important subject of study. If automatic misinformation detection is applied in a real-world setting, it is necessary to validate the methods being used. Large language models (LLMs) have produced the best results among text-based methods. However, fine-tuning such a model requires a significant amount of training data, which has led to the automatic creation of large-scale misinformation detection datasets. In this paper, we explore the biases present in one such dataset for misinformation detection in English, NELA-GT-2019. We find that models are at least partly learning the stylistic and other features of different news sources rather than the features of unreliable news. Furthermore, we use SHAP to interpret the outputs of a fine-tuned LLM and validate the explanation method using our inherently interpretable baseline. We critically analyze the suitability of SHAP for text applications by comparing the outputs of SHAP to the most important features from our logistic regression models.
DOCUMENT
Developing a framework that integrates Advanced Language Models into the qualitative research process.Qualitative research, vital for understanding complex phenomena, is often limited by labour-intensive data collection, transcription, and analysis processes. This hinders scalability, accessibility, and efficiency in both academic and industry contexts. As a result, insights are often delayed or incomplete, impacting decision-making, policy development, and innovation. The lack of tools to enhance accuracy and reduce human error exacerbates these challenges, particularly for projects requiring large datasets or quick iterations. Addressing these inefficiencies through AI-driven solutions like AIDA can empower researchers, enhance outcomes, and make qualitative research more inclusive, impactful, and efficient.The AIDA project enhances qualitative research by integrating AI technologies to streamline transcription, coding, and analysis processes. This innovation enables researchers to analyse larger datasets with greater efficiency and accuracy, providing faster and more comprehensive insights. By reducing manual effort and human error, AIDA empowers organisations to make informed decisions and implement evidence-based policies more effectively. Its scalability supports diverse societal and industry applications, from healthcare to market research, fostering innovation and addressing complex challenges. Ultimately, AIDA contributes to improving research quality, accessibility, and societal relevance, driving advancements across multiple sectors.
Insecten hebben het al decennia lang moeilijk. De aantallen insecten en verscheidenheid in insecten neemt af (Hallmann, 2017). Door een afnemende biodiversiteit neemt de stabiliteit van onze ecosystemen af. Gewas bestuivende wilde insecten hebben een grote rol in bestuiving van open geteelde gewassen. De landbouw is afhankelijker geworden van bestuivende insecten. Honingbijen en wilde bijen zijn beiden belangrijke bestuivers in de landbouw. Er is echter een tekort aan voedsel voor alle bijen. Dit leidt nu tot een gepolariseerde discussie over concurrentie om voedsel. Deze discussie is zelden gebaseerd op feiten die passen bij de situatie. We willen de polarisatie doorbreken met antwoorden over de voedselvoorziening voor bijen in Nederlandse landschapssituaties. Daartoe is onderzoek nodig naar de dracht van verschillende landschapstypen. Kort samengevat: hoeveel bijenvolken kunnen in een bepaald gebied staan? Imkersverenigingen willen dit weten. We willen dit meten met online meetapparatuur bij honingbijenvolken. Tegelijk meten we het effect van de aanwezigheid van honingbijen op de biodiversiteit en dichtheid van andere bestuivende insecten. Dit zal leiden tot een gedegen opzet voor onderzoek, adviezen voor plaatsing in verschillende landschapstypen in Nederland en een aanvraag voor vervolgonderzoek met als doel een verbeterde inrichting en gebruik van landschapstypen in Nederland ten behoeve van honingbijen en wilde bijen.
Promoting entrepreneurship is an enabler of smart, sustainable and inclusive growth and it is one objective EU regions have pursued since the EC included it into 2020 Strategy. Entrepreneurship development has economic and social benefits, since it is not only a driving force for job creation, competitiveness and growth; it also contributes to personal fulfillment and to achieve social objectives. That is why the EU encourages entrepreneurial initiatives and to unlock the growth potential of businesses and citizens. However, only a 37% of Europeans (Eurobarometer 2012) would like to be self-employed. The Entrepreneurship Action Plan adopted by the EC in 2013 to reignite Europe’s entrepreneurial spirit includes initiatives for educating young people on entrepreneurship. To ensure that EU economy remains globally competitive, young generations of Europeans need to be inspired to develop their entrepreneurial mindset. EU 2020 Action Plan argues that young people benefitting of a specialised entrepreneurial education are more likely to start-up a business and to better tackle challenges in their professional career and life in general. Hence, there is good reason to ensure better quality of entrepreneurial education. Most approaches in recent years have focused on improving the skills or competences youngsters should obtain only within the education system. However, an integrated approach is needed, where the school, their friends, family and the social environment, shall play each one a relevant role, contributing to generate a more adequate atmosphere to boost their entrepreneurial mindsets, intrapreneurial attitudes and innovation capacities. This project will identify and exchange – through a quadruple helix approach- good practices for creating friendlier entrepreneurial ecosystems and actions to boost entrepreneurship in young people mindsets. The good practices and lessons learnt will be transferred into Action Plans to be included in regional policies.