|
on Artificial Intelligence |
| By: | Radosveta Ivanova-Stenzel (TU Berlin); Michel Tolksdorf (TU Berlin) |
| Abstract: | We suggest a simple method to elicit individual preferences for algorithms. By altering the monetary incentives for ceding control to the algorithm, the menu-based approach allows for measuring, in particular, the degree of algorithm aversion. Using an experiment, we elicit preferences for algorithms in an environment with measurable performance accuracy under two conditions|the absence and the presence of information about the algorithm's performance. Providing such information raises subjects' willingness to rely on algorithms when ceding control to the algorithm is more costly than trusting their own assessment. However, algorithms are still underutilized. |
| Keywords: | algorithm aversion; delegation; experiment; preferences; |
| JEL: | C91 D83 D91 |
| Date: | 2024–07–30 |
| URL: | https://d.repec.org/n?u=RePEc:rco:dpaper:508 |
| By: | Lippens, Louis (Ghent University) |
| Abstract: | The advent of large language models (LLMs) may reshape hiring in the labour market. This paper investigates how generative pre-trained transformers (GPTs)—i.e. OpenAI’s GPT-3.5, GPT-4, and GPT-4o—can aid hiring decisions. In a direct comparison between humans and GPTs on an identical hiring task, I show that GPTs tend to select candidates more liberally than humans but exhibit less ethnic bias. GPT-4 even slightly favours certain ethnic minorities. While LLMs may complement humans in hiring by making a (relatively extensive) pre-selection of job candidates, the findings suggest that they may miss-select due to a lack of contextual understanding and may reproduce pre-trained human bias at scale. |
| Date: | 2024–07–11 |
| URL: | https://d.repec.org/n?u=RePEc:osf:osfxxx:zxf5y |
| By: | Iñaki Aldasoro; Olivier Armantier; Sebastian Doerr; Leonardo Gambacorta; Tommaso Oliviero |
| Abstract: | Generative artificial intelligence (gen AI) is expected to increase productivity. But if unequally adopted across demographic groups, its proliferation risks exacerbating disparities in pay and job opportunities, leading to greater inequality. To investigate the use of gen AI and its drivers we draw on a representative survey of U.S. household heads from the Survey of Consumer Expectations. We find a significant "gen AI gender gap": while 50% of men already use gen AI, only 37% of women do. Demographic characteristics explain only a small share of this gap, while respondents' self-assessed knowledge about gen AI emerges as the most important factor, explaining three-quarters of the gap. Gender differences in privacy concerns and trust when using gen AI tools, as well as perceived economic risks and benefits, account for the remainder. We conclude by discussing implications for policy to foster equitable gen AI adoption. |
| Keywords: | artificial intelligence, privacy, gender, gen AI |
| JEL: | C8 D8 |
| Date: | 2024–07 |
| URL: | https://d.repec.org/n?u=RePEc:bis:biswps:1197 |
| By: | Mehler, Maren F.; Krautter, Kai |
| Abstract: | Generative Artificial Intelligence (GenAI) has become widespread in daily work but present novel challenges for users as previously meaningful tasks can now be completed by GenAI. This study examines the impact of ChatGPT on task performance and perceived meaningfulness in two programming tasks. In an online experiment (n=161) assigning participants to coding or debugging tasks, with and without ChatGPT assistance, we found that using ChatGPT improved task performance, partially because the supported tasks are less difficult. However, using ChatGPT resulted in lower perceived meaningfulness, partly because participants considered the tasks less effortful. Notably, both tasks exhibited slightly different results, indicating that contextual factors may amplify or mitigate the effects. This study emphasizes the dual nature of GenAI integration, balancing enhanced performance with psychological impacts on users. Our findings offer insights for organizations and developers on integrating GenAI, highlighting the importance of incorporating efficiency gains with the meaningfulness of human work. |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:dar:wpaper:146774 |
| By: | Mehler, Maren F.; Ellenrieder, Sara; Buxmann, Peter |
| Abstract: | While the use of Generative Artificial Intelligence (GenAI) aims to automate human tasks, psychology research shows how crucial human effort is for the appreciation of the final results. The so-called “IKEA effect” refers to the increased valuation individuals attribute to self-created products. However, the potential implications of this effect for GenAI have remained unexplored. This study delves into the presence of the IKEA effect in GenAI, specifically focusing on image creation. Through an online experiment involving 174 participants in Germany, we observed that participants valued images higher if more human effort was invested during collaborative co-creation with GenAI. Our findings indicate a significant presence of the IKEA effect, although existing GenAI research primarily focuses on the automation of processes. This discovery emphasizes the importance of understanding user psychology and also offers valuable insights for designing and leveraging GenAI applications. |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:dar:wpaper:146773 |
| By: | Cattaneo, Maria Alejandra (Swiss Co-ordination Center for Research in Education); Gschwendt, Christian (University of Bern); Wolter, Stefan C. (University of Bern) |
| Abstract: | Advances in technology have always reshaped labor markets. Automating human labor has lead to job losses and creation but most of all, for an increasing demand for highly skilled workers. However, emerging AI innovations like ChatGPT may reduce labor demand in high skilled occupations previously considered "safe" from automation. While initial studies suggest that individuals adjust their educational and career choices to mitigate automation risk, it is unknown what people would be willing to pay for a reduced automation risk. This study quantifies this value by assessing individuals' preferences for occupations in a discrete-choice experiment with almost 6'000 participants. The results show that survey respondents are willing to accept a salary reduction equivalent to almost 20 percent of the median annual gross wage to work in an occupation with a 10 percentage point lower risk of automation. Although the preferences are quite homogeneous, there are still some significant differences in willingness to pay between groups, with men, younger people, those with higher levels of education, and those with a higher risk tolerance showing a lower willingness to pay for lower automation risk. |
| Keywords: | artificial intelligence, automation, willingness to pay, survey experiment |
| JEL: | J24 O33 |
| Date: | 2024–06 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp17097 |
| By: | Findeisen, Sebastian; Dauth, Wolfgang; Schlenker, Oliver |
| Abstract: | New technologies drive productivity growth but the distribution of gains might be unequal and is mediated by labor market institutions. We study the role that organized labor plays in shielding incumbent workers from the potential negative consequences of automation. Combining German individual-level administrative records with information on plant-level robot adoption and the presence of works councils, a form of shop-floor worker representation, we find positive moderating effects of works councils on retention for incumbent workers during automation events. Separations for workers with replaceable task profiles are significantly reduced. When labor markets are tight and replacement costs are high for firms, incumbent workers become more valuable and the effects of works councils during automation events start to disappear. Older workers, who find it more challenging to reallocate to new employers, benefit the most from organized labor in terms of wages employment. Concerning mechanisms we find that robot-adopting plants with works councils employ not more but higher quality robots. They also provide more training during robot adoption and have higher productivity growth thereafter. |
| Keywords: | automation, organized labor, work councils, labor market tightness, worker re-training |
| JEL: | J20 J30 J53 O33 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:cexwps:300230 |
| By: | Eric Langlais; Nanxi Li |
| Abstract: | This paper studies how the combination of Product Liability and Tort Law shapes a monopoly' incentives to invest in R&D for developing risky AI-based technologies ("robots") that may accidentally induce harm to third-party victims. We assume that at the engineering stage, robots are designed to have two alternative modes of motion (fully autonomous vs human-driven), corresponding to optimized performances in predefined circumstances. In the autonomous mode, the monopoly (i.e. AI designer) faces Product Liability and undertakes maintenance expenditures to mitigate victims' expected harm. In the human-driven mode, AI users face Tort Law and exert a level of care to reduce victims' expected harm. In this set-up, efficient maintenance by the AI designer and efficient care by AI users result whatever the liability rule enforced in each area of law (strict liability, or negligence). However, overinvestment as well as underinvestment in R&D may occur at equilibrium, whether liability laws rely on strict liability or negligence, and whether the monopoly uses or does not use price discrimination. The first best level of R&D investments is reached at equilibrium only if simultaneously the monopoly uses (perfect) price discrimination, a regulator sets the output at the socially optimal level, and Courts implement strict liability in Tort Law and Product Liability. |
| Keywords: | Artificial Intelligence, Algorithms, Tort Law, Product Liability, Strict Liability, Negligence |
| JEL: | K13 K2 L1 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:drm:wpaper:2024-22 |
| By: | Jo-An Occhipinti; William Hynes; Ante Prodan; Harris A. Eyre; Roy Green; Sharan Burrow; Marcel Tanner; John Buchanan; Goran Ujdur; Frederic Destrebecq; Christine Song; Steven Carnevale; Ian B. Hickie; Mark Heffernan |
| Abstract: | Work is fundamental to societal prosperity and mental health, providing financial security, identity, purpose, and social integration. The emergence of generative artificial intelligence (AI) has catalysed debate on job displacement. Some argue that many new jobs and industries will emerge to offset the displacement, while others foresee a widespread decoupling of economic productivity from human input threatening jobs on an unprecedented scale. This study explores the conditions under which both may be true and examines the potential for a self-reinforcing cycle of recessionary pressures that would necessitate sustained government intervention to maintain job security and economic stability. A system dynamics model was developed to undertake ex ante analysis of the effect of AI-capital deepening on labour underutilisation and demand in the economy. Results indicate that even a moderate increase in the AI-capital-to-labour ratio could increase labour underutilisation to double its current level, decrease per capita disposable income by 26% (95% interval, 20.6% - 31.8%), and decrease the consumption index by 21% (95% interval, 13.6% - 28.3%) by mid-2050. To prevent a reduction in per capita disposable income due to the estimated increase in underutilization, at least a 10.8-fold increase in the new job creation rate would be necessary. Results demonstrate the feasibility of an AI-capital- to-labour ratio threshold beyond which even high rates of new job creation cannot prevent declines in consumption. The precise threshold will vary across economies, emphasizing the urgent need for empirical research tailored to specific contexts. This study underscores the need for governments, civic organisations, and business to work together to ensure a smooth transition to an AI- dominated economy to safeguard the Mental Wealth of nations. |
| Date: | 2024–04 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2407.01545 |
| By: | Andreas Schaefer (University of Bath); Maik T. Schneider (University of Graz) |
| Abstract: | With the 4th Industrial Revolution ahead there is huge uncertainty about the likely labour market impacts ranging from massive layoffs as a response to Automation and AI to the view that overall more jobs will be created than lost. Whatever the outcome in the end, there will be major structural change with substantial implications for individual labour income risk. We argue that precautionary savings are an ineffective protection against labour market risk arising from major technological shifts and discuss four policy instruments, 1) a private insurance scheme, 2) a universal basic income, 3) a robot tax, and 4) a governmental insurance scheme. Further, we examine whether these policy instruments are suitable to achieve high and inclusive growth. |
| Keywords: | Artificial Intelligence, Economic Growth, Endogenous Technological Change, Industrial Revolution, Robot Tax, Universal Basic Income. |
| JEL: | H20 O33 O38 |
| Date: | 2024–01 |
| URL: | https://d.repec.org/n?u=RePEc:grz:wpaper:2024-06 |
| By: | Andrea Carriero; Davide Pettenuzzo; Shubhranshu Shekhar |
| Abstract: | This paper presents a comparative analysis evaluating the accuracy of Large Language Models (LLMs) against traditional macro time series forecasting approaches. In recent times, LLMs have surged in popularity for forecasting due to their ability to capture intricate patterns in data and quickly adapt across very different domains. However, their effectiveness in forecasting macroeconomic time series data compared to conventional methods remains an area of interest. To address this, we conduct a rigorous evaluation of LLMs against traditional macro forecasting methods, using as common ground the FRED-MD database. Our findings provide valuable insights into the strengths and limitations of LLMs in forecasting macroeconomic time series, shedding light on their applicability in real-world scenarios |
| Date: | 2024–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2407.00890 |
| By: | Kumar, Deepak; Weissenberger-Eibl, Marion |
| Abstract: | In the fast-paced realm of technological evolution, accurately forecasting emerging trends is critical for both academic inquiry and industry application. Traditional trend analysis methodologies, while valuable, struggle to efficiently process and interpret the vast datasets of today's information age. This paper introduces a novel approach that synergizes Generative AI and Bidirectional Encoder Representations from Transformers (BERT) for semantic insights and trend forecasting, leveraging the power of Retrieval-Augmented Generation (RAG) and the analytical prowess of BERT topic modeling. By automating the analysis of extensive datasets from publications and patents, the presented methodology not only expedites the discovery of emergent trends but also enhances the precision of these findings by generating a short summary for found emergent trends. For validation, three technologies - reinforcement learning, quantum machine learning, and Cryptocurrencies - were analysed prior to their first appearance in the Gartner Hype Cycle. Research highlights the integration of advanced AI techniques in trend forecasting, providing a scalable and accurate tool for strategic planning and innovation management. Results demonstrated a significant correlation between model's predictions and the technologies' appearances in the Hype Cycle, underscoring the potential of this methodology in anticipating technological shifts across various sectors |
| Keywords: | BERT, Topic modelling, RAG, Gartner Hype Cycle, LLM, BERTopic |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:esconf:300545 |
| By: | Yi Chen (ShanghaiTech University); Hanming Fang (University of Pennsylvania); Yi Zhao (Tsinghua University); Zibo Zhao (ShanghaiTech University) |
| Abstract: | Categorical variables have no intrinsic ordering, and researchers often adopt a fixed-effect (FE) approach in empirical analysis. However, this approach has two significant limitations: it overlooks textual information associated with the categorical variables; and it produces unstable results when there are only limited observations in a category. In this paper, we propose a novel method that utilizes recent advances in large language models (LLMs) to recover overlooked information in categorical variables. We apply this method to investigate labor market mismatch. Specifically, we task LLMs with simulating the role of a human resources specialist to assess the suitability of an applicant with specific characteristics for a given job. Our main findings can be summarized in three parts. First, using comprehensive administrative data from an online job posting platform, we show that our new match quality measure is positively correlated with several traditional measures in the literature, and we highlight the LLM’s capability to provide additional information beyond that contained in the traditional measures. Second, we demonstrate the broad applicability of the new method with a survey data containing significantly less information than the administrative data, which makes it impossible to compute most of the traditional match quality measures. Our LLM measure successfully replicates most of the salient patterns observed in a hard-to-access administrative dataset using easily accessible survey data. Third, we investigate the gender gap in match quality and explore whether there exists gender stereotypes in the hiring process. We simulate an audit study, examining whether revealing gender information to LLMs influences their assessment. We show that when gender information is disclosed to the LLMs, the model deems females better suited for traditionally female-dominated roles. |
| Keywords: | Large Language Models, Categorical Variables, Labor Market Mismatch |
| JEL: | C55 J16 J24 J31 |
| Date: | 2024–07–23 |
| URL: | https://d.repec.org/n?u=RePEc:pen:papers:24-017 |
| By: | Marc Burri; Daniel Kaufmann; Nima Ostovan |
| Abstract: | This report documents the use and misuse of generative artificial intelligence in academic economic research and provides guidelines for university students and instructors. It primarily addresses students and instructors in a master’s program in economics; however, the use cases and guidelines may be useful in other fields and academic research in general. |
| Date: | 2024–07 |
| URL: | https://d.repec.org/n?u=RePEc:irn:polrep:24-03 |