PDF Publication Title:
Text from PDF Page: 014
4 An Intelligence in Our Image The fallibility of algorithms is an easy point to make. This includes systematic algorithmic errors, not just the statistical inaccuracies inher- ent to many algorithms. There are many examples in public policy– oriented applications. As a concrete example of significant error, Google’s Flu Trends tool is famous for repeatedly misdiagnosing nation- wide flu trends (Lazer et al., 2014). Many risk-estimation algorithms were based on incorrect probabilistic models and failed to react prop- erly just before the 2008 U.S. financial crash (Salmon, 2012). One city implemented algorithms intended to optimally detect street potholes based on passively collected data from smartphone users. The demo- graphic breakdown of smartphone users at the time would have led to blind spots, causing some communities to be underserved (Crawford, 2013). This would have had the effect of depriving less-affluent citizens’ access to city repair services. Another city decided to use algorithmic approaches to direct its law-enforcement activities. The justification was that predictive policing algorithms were more objective as they only relied on objective “multi-variable equations,” not on subjective human decisions (quoted in Tett, 2014). Reporting on another crimi- nal justice application, Angwin et al. (2016) demonstrated systematic bias in a criminal risk assessment algorithm used in sentencing hear- ings across the United States. Defining Algorithms It will be helpful to carefully examine what algorithms are as we pro- ceed. The concept has shifted quite a bit over centuries. The medieval Islamic scholar Abu-Abdullah Muhammed ibn-Musa Al-Khwarizmi, who lent his name to the algorithm was more interested in reliable step- by-step procedures for computing solutions to equations (Arndt, 1983). Alonzo Church and Alan Turing (Turing, 1937a; Turing, 1937b) intro- duced the concepts of computability and computable functions to for- malize the idea of an algorithm. The definition amounted to a finite sequence of precise instructions that are implementable on computing systems (including but not limited to human brains). This probably brings to mind involved rote procedures, such as recipes for makingPDF 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)