nep-ain New Economics Papers
on Artificial Intelligence
Issue of 2023‒10‒09
eight papers chosen by
Ben Greiner, Wirtschaftsuniversität Wien

  1. Computer says 'no': Exploring systemic hiring bias in ChatGPT using an audit approach By Louis Lippens
  2. Can Unbiased Predictive AI Amplify Bias? By Tanvir Ahmed Khan
  3. Executive AI Literacy: A Text-Mining Approach to Understand Existing and Demanded AI Skills of Leaders in Unicorn Firms By Pinski, Marc; Hofmann, Thomas; Benlian, Alexander
  4. Conducting qualitative interviews with AI By Felix Chopra; Ingar Haaland
  5. Data science, artificial intelligence and the third wave of digital era governance By Dunleavy, Patrick; Margetts, Helen
  6. Novissi Togo - Harnessing Artificial Intelligence to Deliver Shock-Responsive Social Protection By Lawson, Cina; Koudeka, Morlé; Cardenas Martinez, Ana Lucia; Alberro Encinas, Luis Inaki; Karippacheril, Tina George
  7. The future of artificial intelligence in the Arab world The experience of some Arab countries By Bouzid Merouane
  8. TradingGPT: Multi-Agent System with Layered Memory and Distinct Characters for Enhanced Financial Trading Performance By Yang Li; Yangyang Yu; Haohang Li; Zhi Chen; Khaldoun Khashanah

  1. By: Louis Lippens
    Abstract: Large language models offer significant potential for optimising professional activities, such as streamlining personnel selection procedures. However, concerns exist about these models perpetuating systemic biases embedded into their pre-training data. This study explores whether ChatGPT, a chatbot producing human-like responses to language tasks, displays ethnic or gender bias in job applicant screening. Using a correspondence audit approach, I simulated a CV screening task in which I instructed the chatbot to rate fictitious applicant profiles only differing in names, signalling ethnic and gender identity. Comparing ratings of Arab, Asian, Black American, Central African, Dutch, Eastern European, Hispanic, Turkish, and White American male and female applicants, I show that ethnic and gender identity influence ChatGPT's evaluations. The ethnic bias appears to arise partly from the prompts' language and partly from ethnic identity cues in applicants' names. Although ChatGPT produces no overall gender bias, I find some evidence for a gender-ethnicity interaction effect. These findings underscore the importance of addressing systemic bias in language model-driven applications to ensure equitable treatment across demographic groups. Practitioners aspiring to adopt these tools should practice caution, given the adverse impact they can produce, especially when using them for selection decisions involving humans.
    Date: 2023–09
  2. By: Tanvir Ahmed Khan
    Abstract: Predictive AI is increasingly used to guide decisions on agents. I show that even a bias-neutral predictive AI can potentially amplify exogenous (human) bias in settings where the predictive AI represents a cost-adjusted precision gain to unbiased predictions, and the final judgments are made by biased human evaluators. In the absence of perfect and instantaneous belief updating, expected victims of bias become less likely to be saved by randomness under more precise predictions. An increase in aggregate discrimination is possible if this effect dominates. Not accounting for this mechanism may result in AI being unduly blamed for creating bias.
    Keywords: artificial intelligence, AI, algorithm, human-machine interactions, discrimination, bias, algorithmic bias, financial institutions
    JEL: O33 J15 G2
    Date: 2023–07
  3. By: Pinski, Marc; Hofmann, Thomas; Benlian, Alexander
    Date: 2023
  4. By: Felix Chopra (University of Copenhagen, CEBI); Ingar Haaland (Norwegian School of Economics)
    Abstract: Qualitative interviews are one of the fundamental tools of empirical social science research and give individuals the opportunity to explain how they understand and interpret the world, allowing researchers to capture detailed and nuanced insights into complex phenomena. However, qualitative interviews are seldom used in economics and other disciplines inclined toward quantitative data analysis, likely due to concerns about limited scalability, high costs, and low generalizability. In this paper, we introduce an AI-assisted method to conduct semi-structured interviews. This approach retains the depth of traditional qualitative research while enabling large-scale, cost-effective data collection suitable for quantitative analysis. We demonstrate the feasibility of this approach through a large-scale data collection to understand the stock market participation puzzle. Our 395 interviews allow for quantitative analysis that we demonstrate yields richer and more robust conclusions compared to qualitative interviews with traditional sample sizes as well as to survey responses to a single open-ended question. We also demonstrate high interviewee satisfaction with the AI-assisted interviews. In fact, a majority of respondents indicate a strict preference for AI-assisted interviews over human-led interviews. Our novel AI-assisted approach bridges the divide between qualitative and quantitative data analysis and substantially lowers the barriers and costs of conducting qualitative interviews at scale.
    Keywords: Artificial Intelligence, Interviews, Large Language Models, Qualitative Methods, Stock Market Participation
    JEL: C83 C90 D14 D91 Z13
    Date: 2023–09–25
  5. By: Dunleavy, Patrick; Margetts, Helen
    Abstract: This article examines the model of digital era governance (DEG) in the light of the latest-wave of data-driven technologies, such as data science methodologies and artificial intelligence (labelled here DSAI). It identifies four key top-level macro-themes through which digital changes in response to these developments may be investigated. First, the capability to store and analyse large quantities of digital data obviates the need for data 'compression' that characterises Weberian-model bureaucracies, and facilitates data de-compression in data-intensive, information regimes, where the capabilities of public agencies and civil society are both enhanced. Second, the increasing capability of robotic devices have expanded the range of tasks that machines extending or substituting workers' capabilities can perform, with implications for a reshaping of state organisation. Third, DSAI technologies allow new ways of partitioning state functions in ways that can maximise organisational productivity, in an 'intelligent centre, devolved delivery' model within vertical policy sectors. Fourth, within each tier of government, DSAI technologies offer new possibilities for 'administrative holism' - the horizontal allocation of power and functions between organisations, through state integration, common capacity and needs-based joining-up of services. Together, these four themes comprise a third wave of DEG changes, suggesting important administrative choices to be made regarding information regimes, state organisation, functional allocation and outsourcing arrangements, as well as a long-term research agenda for public administration, requiring extensive and detailed analysis. This article has been accepted for publication in the Sage journal Public Policy and Administration, August 2023.
    Date: 2023–08–27
  6. By: Lawson, Cina; Koudeka, Morlé; Cardenas Martinez, Ana Lucia; Alberro Encinas, Luis Inaki; Karippacheril, Tina George
    Abstract: This case study, jointly authored by the Government of Togo and the World Bank, documents the innovative features of the NOVISSI program and posits some directions for the way forward. The study examines how Togo leveraged artificial intelligence and machine learning methods to prioritize the rural poor in the absence of a shock-responsive social protection delivery system and a dynamic social registry. It also discusses the main challenges of the model and the risks and implications of implementing such a program.
    Date: 2023–09–01
  7. By: Bouzid Merouane (UMBB - Université M'Hamed Bougara Boumerdes)
    Abstract: For more than two decades, artificial intelligence has been making major transformations in various sectors: from education, healthcare, to public transportation, business, entertainment, war, and more. Therefore, this sector has turned into a major competition arena among the countries of the world. Arab countries live in different internal conditions, which are clearly reflected in their plans to adopt artificial intelligence in their discourse, strategies, and institutions. Arab countries, especially in the Gulf, hastened to adopt the latest technologies, institutions, standards and plans to localize and use artificial intelligence, which reflected positively on their ranking in global indicators. On the other hand, other Arab countries are still groping their way, with attempts to teach artificial intelligence subjects in some curricula with the aim of laying the foundations for this industry.
    Keywords: intelligence artificial intelligence research centers the strategy innovation decisions. JEL Classification Codes: J23, J24, intelligence, artificial intelligence, research centers, the strategy, innovation decisions. JEL Classification Codes: J23
    Date: 2023–06–04
  8. By: Yang Li; Yangyang Yu; Haohang Li; Zhi Chen; Khaldoun Khashanah
    Abstract: Large Language Models (LLMs), prominently highlighted by the recent evolution in the Generative Pre-trained Transformers (GPT) series, have displayed significant prowess across various domains, such as aiding in healthcare diagnostics and curating analytical business reports. The efficacy of GPTs lies in their ability to decode human instructions, achieved through comprehensively processing historical inputs as an entirety within their memory system. Yet, the memory processing of GPTs does not precisely emulate the hierarchical nature of human memory. This can result in LLMs struggling to prioritize immediate and critical tasks efficiently. To bridge this gap, we introduce an innovative LLM multi-agent framework endowed with layered memories. We assert that this framework is well-suited for stock and fund trading, where the extraction of highly relevant insights from hierarchical financial data is imperative to inform trading decisions. Within this framework, one agent organizes memory into three distinct layers, each governed by a custom decay mechanism, aligning more closely with human cognitive processes. Agents can also engage in inter-agent debate. In financial trading contexts, LLMs serve as the decision core for trading agents, leveraging their layered memory system to integrate multi-source historical actions and market insights. This equips them to navigate financial changes, formulate strategies, and debate with peer agents about investment decisions. Another standout feature of our approach is to equip agents with individualized trading traits, enhancing memory diversity and decision robustness. These sophisticated designs boost the system's responsiveness to historical trades and real-time market signals, ensuring superior automated trading accuracy.
    Date: 2023–09

This nep-ain issue is ©2023 by Ben Greiner. It is provided as is without any express or implied warranty. It may be freely redistributed in whole or in part for any purpose. If distributed in part, please include this notice.
General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.