This thesis explores the relationship between bureaucratic accountability and their disposition toward utilizing algorithms in their decision-making processes. Drawing upon the literature on...Show moreThis thesis explores the relationship between bureaucratic accountability and their disposition toward utilizing algorithms in their decision-making processes. Drawing upon the literature on government accountability and aversion to algorithmic decision-making, it hypothesizes that the more public officials are aware of the chains of accountability they are tied to, the less favorable they will be to utilizing algorithms. The hypothesis is tested through a case study of the Chilean Institute of Social Services, which employs multiple algorithms to automate eligibility decisions for pension and other social benefit applications. To trace the organization’s bureaucratic accountability chain, data collection is based on semi-structured interviews of public officials from different hierarchical levels. The results confirm several theoretical expectations on reduced discretion, muddled authority over the algorithm and algorithmic opacity, leading to blame avoidance within the organization. However, the results also disprove the hypothesized negative relationship, revealing that officials with high awareness and perceptions of individual accountability instead favor using algorithms to automate decisions. Further analysis of the dependent variable reveals that a favorable disposition toward algorithm use is overwhelmingly tied to the perception of trust. The individual descriptions of bureaucrats convey clues for an alternative explanation of the outcome, suggesting that stringent evaluation and audit practices can help circumvent algorithm aversion resulting from opaque algorithms or reduced discretion. Such a potential explanation implies that bureaucratic accountability chains could serve as a substitute source of trust, allowing public servants to hold the algorithm to account by proxy. The qualitative accounts in this thesis offer insights into how bureaucrats feel personally accountable for the algorithms they use, expanding the literature of public officials’ reliance on algorithmic decision-making.Show less
