In UNICEF’s guidelines for innovation, #8 is a particular favourite of mine: “do no harm.” It harkens back to Google’s old motto to “don’t be evil”, but Google has since changed their slogan since, and it feels like the rest of the tech world has as well. UNICEF’s version even goes a step farther than google: whereas evil is ambiguous and up for interpretation, harm can be measured. To do no harm is a difficult goal and requires a lot of forethought and consideration of ethics, which are ideas explored by this week’s speakers and within the required reading for Doing Good is Good Business. How do we use data to do good and not do harm?
In fairness, Donald Trump supporters would not characterize Cambridge Analytica as doing any harm, but if “The Data That Turned the World Upside Down” is completely factual, their work is easily described as harmful. Although this, and many other articles I’ve read on the subject, act as though the use of Big 5 and personality psychology was Cambridge Analytica’s chief innovation, from my reading, it’s not too different from what Barack Obama’s team innovated in 2008 and 2012. The true difference is the use of targeted advertising and segmentation for voter suppression: discouraging potential Hillary Clinton supporters from voting. This is certainly harmful, and certainly not what Michal Kosinski intended with his data collection and analysis– let alone what Mark Zuckerburg and other technologists intended with the web applications they created.
Similarly, personalized newsfeed and recommender systems were created with the best of intentions. You wouldn’t want to waste your time with things you weren’t interested in. But to be fed a diet of only things you’re interested in can be harmful to your soul, and harmful to your understanding of the world. I’m not always very sympathetic to conservatives– I don’t think there are always two sides to every story, and the so-called “other side” should not be part of the conversation sometimes (for instance, I do not believe creationism has a place in science classes.) But keeping Palestinian and Israeli networks completely separated seems to be doing tangible harm.
In a prime example of the blindness of the tech industry, there is no discussion of ethics within “Social Media Fingerprints of Unemployment”. Is it truly ethical to use social media data to determine economic statistics? One might argue that it isn’t too different from data collected in traditional ways– such as censuses or unemployment surveys. But there is no obvious connection in social media to unemployment, and people who use social media do not explicitly consent to their posts being used in this way. If we do use big data for these purposes, would it affect the accuracy of the data once this information collecting is fully disclosed?
Some questions I might have for our panel discussion next class:
- How can websites and applications do more to inform their users of the information being collected on them and how it’s being used?
- How common is the practice of buying and selling data amongst companies? Or amongst companies and non-profits?
- What is done to compensate for underrepresentation amongst minority in Big Data?
Subscribe via RSS