Understanding and Preventing Artificial Intelligence Ethics Issues in Financial Services Organizations: Three Studies
Bibliographic record
Abstract
Reports of financial institutions proliferating discrimination, jeopardizing customer privacy, generating decisions from customer data without informed consent, and making other ethical failures in their use of artificial intelligence (AI) are increasing. As the use of AI grows, understanding and preventing the associated ethics implications is of great importance. The goal of the dissertations is to determine what the key ethical issues are arising from the use of AI in financial services organizations and study the use of AI principles (AIPs) to prevent them. AIPs are formal documents developed or selected by an organization that state normative declarations about how artificial intelligence ought to be used by its managers and employees. First, an introduction is provided that discusses the general motivation for the dissertation, a summary of the research studies included within, and the contributions of the thesis in relation to other fields studying AI ethics (Chapter 1). Second, a study is conducted to determine what the key ethical implications are of using AI in the financial services industry, drawing on literature and a semi-structured interview study of 21 employees at a large Canadian financial services institution. This understanding of AI ethics issues is presented as a Code of Conduct (one type of AIP), which has since been adopted by several Canadian financial institutions (Chapter 2). Third, an in-depth technical case study is presented that considers one ethical issue uncovered in Chapter 2: bias and discrimination. A realistically large dataset from a global fintech lender is used to simulate the impact of anti-discrimination regimes and their corresponding data management and model building guidance on gender-based discrimination. In addition, several approaches for organizations to reduce the discrimination are provided, considering the implications for model quality and firm profitability. The regulatory implications for preventing gender discrimination in non-mortgage fintech lending are then discussed (Chapter 3). Fourth, the dissertation returns to the broader organizational study of AI ethics self-regulatory initiatives and examines employee perceptions on the effective adoption of AIPs. 49 interviews were conducted with employees of 24 financial services organizations across 11 countries; the findings from which provide eleven components that could impact the effective adoption of AIPs at organizations (Chapter 4). The conclusions and practical implications of the dissertation are then summarized, followed by a discussion of limitations and future research plans (Chapter 5).
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.034 | 0.062 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.004 | 0.007 |
| Science and technology studies | 0.018 | 0.010 |
| Scholarly communication | 0.007 | 0.006 |
| Open science | 0.002 | 0.007 |
| Research integrity | 0.004 | 0.006 |
| Insufficient payload (model declined to judge) | 0.003 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".