2nd February 2020
Technically Wrong – sexist apps, biased algorithms and other threats of toxic tech.
Sara Wachter-Boettcher, Norton 2017 232 pp
There have been several analyses of the current state of computing technology reviewed in PC Update over the last year or two. The author of this book, which is now a couple of years old, is female and very aware of misogynist bias. She relates how poorly early revisions of Siri reacted to statements like ‘Help, I’ve been raped’. There is also the issue of appropriate questions; she was once asked in an on-line form (to buy from a pharmacy) whether she had ever been sexually assaulted; the only options were yes or no, and there was no indication how the data might be used.
Snapchat mishandled a facial morphing feature. Google photos classified African Americans as gorillas. One suggestion to reduce problems like these is to increase diversity in technical company staff. Wachter-Boettcher thinks that this will fail, citing surveys that show the best indicator of employee suitability is similarity to existing staff.
Facebook celebrated what a fantastic year someone had had with photos (complete with pictures of balloons and streamers) from the funeral of his daughter who had recently died of cancer. It was undoubtedly the major event in his year, but not something he wanted to be reminded of or to celebrate.
Wachter-Boettcher has difficulty with her surname in most applications; some don’t accept hyphens and the length of the name can also be an issue.
In 1958, Michael Young coined the term “meritocracy” in his book, The Rise of the Meritocracy. Young used the term satirically to depict a United Kingdom ruled by a system that favoured intelligence and merit above all else. Wachter-Boettcher points out that it has been, and will continue to be, a badge of honour in Silicon Valley. She knows that we can only succeed in ridding technology of excesses and oversights if we look at digital tools as a result of a series of choices made by designers and technologists and try to improve the way those choices are made.
Technically wrong is a readable summary of the types of issues raised by recent developments in Artificial Intelligence.