PDF Publication Title:
Text from PDF Page: 033
The Problem in Focus: Factors and Remedies 23 The court case involved dueling statistical experts debating the find- ings of Baldus’s study. The court proceedings even included protracted discussions on detailed statistical concepts, such as multicollinearity. The Supreme Court finally held that the sentencing was valid because the study did not demonstrate deliberate bias in McCleskey’s case. This was the final decision in spite of the carefully demonstrated 4-to-1 racial disparity in sentencing outcomes. The court’s justification was that, however true the Baldus study was, it did not demonstrate that race was a causal factor in McCleskey’s particular sentencing. If we are to rely on algorithms for autonomous decisionmak- ing, they need to be equipped with tools for auditing the causal fac- tors behind key decisions. Algorithms that can be audited for causal factors can give clearer accounts or justifications for their outcomes. This is especially important for justifying statistically disproportionate outcomes. Algorithmic Literacy and Transparency Combating algorithmic bias would benefit from an educated public capable of understanding that algorithms can lead to inequitable out- comes. This is not the same as requiring that users understand the inner workings of all algorithms—this is not feasible. Just instilling a healthy dose of informed skepticism could be useful enough to reduce the effect size of automation bias. There is hope on this front. The sheer amount of time we spend interfacing with algorithms may make algo- rithmic missteps more noticeable. For example, online dating users (a rapidly rising percentage of the population) routinely question the results of date matching algo- rithms. Journalism and documentaries on the 2008 subprime mort- gage financial crash have also helped foster a healthy, more-informed cultural skepticism about the efficacy of complex algorithms. Consider recent reports of public outcry over the SketchFactor app (Marantz, 2015). The app used crowdsourced data and aggregation algorithms to calculate a neighborhood’s “sketchiness” score. There was significant negative reaction to the app on the ground of cultural insensitivity and potential for discriminatory abuse. The public was able to clearly artic-PDF Image | Intelligence in Our Image Risks of Bias and Errors in Artificial Intelligence
PDF Search Title:
Intelligence in Our Image Risks of Bias and Errors in Artificial IntelligenceOriginal File Name Searched:
RAND_RR1744.pdfDIY PDF Search: Google It | Yahoo | Bing
Cruise Ship Reviews | Luxury Resort | Jet | Yacht | and Travel Tech More Info
Cruising Review Topics and Articles More Info
Software based on Filemaker for the travel industry More Info
The Burgenstock Resort: Reviews on CruisingReview website... More Info
Resort Reviews: World Class resorts... More Info
The Riffelalp Resort: Reviews on CruisingReview website... More Info
CONTACT TEL: 608-238-6001 Email: greg@cruisingreview.com (Standard Web Page)