Contributions in Merlin

Large Language Models have made remarkable progress in question-answering tasks, but challenges like hallucination and outdated information persist. These issues are especially critical in domains like climate change, where timely access to reliable information is vital. One solution is granting these models access to external, scientifically accurate sources to enhance their knowledge and reliability. Here, we enhance GPT-4 by providing access to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC AR6), the most comprehensive, up-to-date, and reliable source in this domain (refer to the ’Data Availability’ section). We present our conversational AI prototype, available at www.chatclimate.ai, and demonstrate its ability to answer challenging questions in three different setups: (1) GPT-4, (2) ChatClimate, which relies exclusively on IPCC AR6 reports, and (3) Hybrid ChatClimate, which utilizes IPCC AR6 reports with in-house GPT-4 knowledge. The evaluation of answers by experts show that the hybrid ChatClimate AI assistant provide more accurate responses, highlighting the effectiveness of our solution.

The concept of the tragedy of the commons, originally rooted in economics, describes the depletion of shared resources due to self-interested actions by individuals. This work proposes a novel solution to address this economic challenge by leveraging tokens to capture its multidimensional nature. By utilising blockchain and DLTs, this decentralised approach aims to achieve a social optimum while promoting self-regulation. The paper presents a mathematical treatment of the tragedy of the commons, incorporating multi-dimensional tokens and exploring the divergence from the classic optimal solution, highlighting the potential of tokenisation in shaping a sustainable and efficient economy.

When introducing unfamiliar Artificial Intelligence (AI)-based systems, such as conversational agents (CAs), one needs to ensure that users interact with them according to their design. While past research has studied single-user environments, many practical settings involve multiple parties. This study addresses this gap and focuses on financial advisory service encounters and how mental models evolve in multi-party contexts. A multimodal interactive CA is developed and tested in financial consultations with 24 clients. The observations of these consultations and subsequent interviews provide insights into the challenges of using CAs in unfamiliar contexts. The clients have difficulties effectively using the system. This is linked to the institutional setting of financial advisory service encounters and a mismatch between the designer’s conceptual model and the client’s mental model, which we call secondary mental model.

Current approaches to Argument Mining (AM) tend to take a holistic or black-box view of the overall pipeline. This paper, in contrast, aims to provide a solution to achieve increased performance based on current components instead of independent all-new solutions. To that end, it presents the Deployment of Recombination and Ensemble methods for Argument Miners (DREAM) framework that allows for the (automated) combination of AM components. Using ensemble methods, DREAM combines sets of AM systems to improve accuracy for the four tasks in the AM pipeline. Furthermore, it leverages recombination by using different argument miners elements throughout the pipeline. Experiments with five systems previously included in a benchmark show that the systems combined with DREAM can outperform the previous best single systems in terms of accuracy measured by an AM benchmark.

In code reviews, developers examine code changes authored by peers and provide feedback through comments. Despite the importance of these comments, no accepted approach currently exists for assessing their quality. Therefore, this study has two main objectives: (1) to devise a conceptual model for an explainable evaluation of review comment quality, and (2) to develop models for the automated evaluation of comments according to the conceptual model. To do so, we conduct mixed-method studies and propose a new approach: EvaCRC (Evaluating Code Review Comments). To achieve the first goal, we collect and synthesize quality attributes of review comments, by triangulating data from both authoritative documentation on code review standards and academic literature. We then validate these attributes using real-world instances. Finally, we establish mappings between quality attributes and grades by inquiring domain experts, thus defining our final explainable conceptual model. To achieve the second goal, EvaCRC leverages multi-label learning. To evaluate and refine EvaCRC, we conduct an industrial case study with a global ICT enterprise. The results indicate that EvaCRC can effectively evaluate review comments while offering reasons for the grades.

The literature on knowledge spillovers provides evidence that higher education institutions (HEIs) positively affect regional firm location (i.e., start-ups or firms located in a region). However, less is known about how HEIs in different fields of study impact regional firm location in different industries. To investigate this question, we exploit the establishment of universities of applied sciences (UASs)—bachelor’s degree-granting three-year HEIs in Switzerland. We find that the effects of UASs are heterogeneous across fields of study and industries. UASs specializing in “chemistry and the life sciences” and “business, management, and services” are the only UASs that positively affect regional firm location across several industries. Positive effects emerge in service industries characterized by radical service, incremental product, or process innovations. Thus, UASs are not a one-size-fits-all solution for increasing regional firm location. Instead, only UASs specializing in particular fields of study positively influence firm location in certain industries.

We estimate the risk premium for firm-level climate change exposure among S&P 500 stocks and its time-series evolution between 2005 to 2020. Exposure reflects the attention paid by market participants in earnings calls to a firm’s climate-related risks and opportunities. When extracted from realized returns, the unconditional risk premium is insignificant but exhibits a period with a positive risk premium before the financial crisis and a steady increase thereafter. Forward-looking expected return proxies deliver an unconditionally positive risk premium with maximum values of 0.5%–1% p.a., depending on the proxy, between 2011 and 2014. The risk premium has been lower since 2015, especially when the expected return proxy explicitly accounts for the higher opportunities and lower crash risks that characterize high-exposure stocks. This finding arises as the priced part of the risk premium primarily originates from uncertainty about climate-related upside opportunities. In the time series, the risk premium is negatively associated with green innovation; Big Three holdings; and environmental, social, and governance fund flows and positively associated with climate change adaptation programs.

We replicate the finding of Lefgren et al. (2015) showing that professional basketball coaches in the NBA discontinuously change their starting lineup more often after narrow losses than after narrow wins. This result is consistent with outcome bias because such narrow outcomes are conditionally uninformative. As our paper shows, this pattern is not restricted to the NBA; we also find evidence of outcome bias in the top women’s professional basketball league and college basketball. Finally, we show that outcome bias in coaching decisions generalizes to the National Football League (NFL). We conclude that outcome bias is credible and robust, although it has weakened over time in some instances.

In this paper, we aimed to compare the predictive validity of two models of temperament structure: the one proposed within Strelau's Regulative Theory of Temperament (RTT) and the other developed by Strus, Ponikiewska and Cieciuch as a reconceptualization of the RTT fundamental temperament dimensions based on the insights from the temperamental Big Two and the Circumplex of Personality Metatraits. Specifically, we compared the predictive validity of these two temperament models in relation to a set of external variables related to stress (well-being, stress, PTSD, and COVID stress). The study was conducted on a Polish sample (N = 336, age range 17–65). We found that the reconceptualized temperament dimensions allow for better predictions of well-being, stress, and PTSD than the RTT ones.

This study analyzes price discrimination and household switching in the residential mortgage market. Using a unique proprietary micro dataset from Norway, we examine the factors that influence a bank’s choice to counter an offer from a competing bank and the difference between the loan rate paid by current clients when receiving a competing offer from another bank and the concurrent best rate offered to new customers by the current bank. The estimates show that a bank employs internal information to decide how to counter a competing offer and that current clients pay approximately 20 basis points more than new customers. We surmise that new regulations and digitalization enhance transparency and can reduce the rate differential. However, introducing new banking products and changes in the timing of rate differentiation - from immediate upfront to gradually over time - may be used to maintain a constant rate differential.