A Report on “Tethics”: Ethical Challenges in a Data Saturated World
In 1942, Isaac Asimov published a short story called “Runaround¹, a future (2015) in which a robot malfunctions while accompanying a couple of scientists to Mercury to restart a mining mission.
The malfunction centers around the implementation of three Laws of Robotics, which Asimov introduces for the first time in this story. Speedy, the robot in question, is caught between law two and three:
“…One, a robot may not injure a human being, or, through inaction, allow a human being to come to harm.”
“Two,” continued Powell, “a robot must obey the orders given it by human beings except where such orders would conflict with the First Law.”
“And three, a robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.”
Asimov’s is far from the only ethical reflection stirred up by tech acceleration. The fallout from Cambridge Analytica’s role in the 2016 election is still being felt today.²
In an article for Slate magazine, Microsoft CEO Satya Nadella quotes New York Times journalist John Markoff: “The best way to answer the hard questions about control in a world full of smart machines is by understanding the values of those who are actually building these systems.”³
As a field, data science is relatively new and its power has come to touch every facet of our lives in ways never dreamed of in Asimov’s time. The combination of power coupled with rate of change heightens the possibility of mismanagement, mistake or abuse. Greater regulation will likely be a key step to take but aside from its political feasibility (very much in doubt in today’s bifurcated climate), will almost certainly be insufficient in our global context. International cooperation? Unlikely as the political will necessary to maintain international relationships wanes in the face of the rising tide of nationalism.
While there is much room for pessimism, it is my hope that highlighting some of the positive steps being taken will be food for thought for those who will become the next generation of data engineers and analysts, project managers and executives.
Princeton University began WebTAP in 2014 to inform the public about the uses and implications of enhanced data collection. The key feature of WebTAP lies in OpenWPM, an open source software conducting automated web privacy measurements. 2019 alone saw publications on fingerprint surface-based detection of web bots detectors⁴, a privacy analysis of the web-porn ecosystem⁵, and comparisons of web privacy protection techniques⁶, to name a few.
Stanford University takes another approach. The One Hundred Year Study on Artificial Intelligence anticipates emergent trends in privacy and machine intelligence, AI and warfare, and criminal uses of AI.⁷
Finally, there are corporate efforts as well to move the needle on data security issues. While these can be viewed as efforts to stymie the call for regulation by demonstrating a capacity for self-policing, it is impossible to make any headway on these crucial issues if those agents most responsible for technological innovations remain on the sidelines. While certainly they are at the forefront of protecting theirs and their customers data from malicious attacks, they have been noticeably silent when taken to task over their own data collection practices. It is hard to see how these larger societal issues will be addressed in anything close to a cohesive way if that silence continues unabated.
A Microsoft policy paper lays out 4 general areas where ethical grey areas arise in data science:
- Bias in the data- datasets can reflect the biases of those collecting the data.
- Interpretability of algorithmic models- a model that is too complex to explain simply can occult when the model is improperly suited for a project.
- Controversial uses of data science- example: utilizing credit scores of a borrowers facebook friends to ascertain credit worthiness.
- Unsupervised system decisions and interactions.
From their conclusion:
“At Microsoft, we feel that the first step is recognizing that these thorny issues exist; some data scientists are aware of them and others are not. For those who are aware there is a clamoring for guidance on how to think through these issues and design to avoid harm. We are currently at the stage of articulating guiding principles and triage processes teams can use, but there is a need to test them in practice and to vet and further develop them with our teams and with others in industry.”⁸
[1] Asimov, Issac. “Runaround.” “I, Robot (1950) p 20. chrome-extension://oemmndcbldboiebfnladdacbdfmadadm/http://ekladata.com/-Byix64G_NtE0xI4A6PA1--o1Hc/Asimov-Isaac-I-Robot.pdf
[2] “The Great Hack”, Karim Amer & Jehane Noujaim, Netflix.com, July 24, 2019
[3] “The Partnership of the Future”, Slate.com, June 28, 2016, https://slate.com/technology/2016/06/microsoft-ceo-satya-nadella-humans-and-a-i-can-work-together-to-solve-societys-challenges.html
[4] chrome-extension://oemmndcbldboiebfnladdacbdfmadadm/http://www.open.ou.nl/hjo/papers/ESORICS19.pdf
[5] chrome-extension://oemmndcbldboiebfnladdacbdfmadadm/http://www1.icsi.berkeley.edu/~narseo/papers/pornweb2019_preprint.pdf
[6] chrome-extension://oemmndcbldboiebfnladdacbdfmadadm/https://arxiv.org/pdf/1712.06850.pdf
[7] chrome-extension://oemmndcbldboiebfnladdacbdfmadadm/https://ai100.stanford.edu/sites/g/files/sbiybj9861/f/ai100report10032016fnl_singles.pdf
[8] Cantrell, Bethan; Salido, Javier & Van Hollebeke, Mark . “Industry needs to embrace data ethics: here’s how it could be done.” chrome-extension://oemmndcbldboiebfnladdacbdfmadadm/http://datworkshop.org/papers/dat16-final38.pdf
