Ethics

Lecture 23

Dr. Mine Çetinkaya-Rundel

Duke University
STA 101 - Fall 2023

Warm up

Announcements

• Course evaluations - On DukeHub
• TA evaluations - Email from Dr. Joan Durso

10:00

Misrepresentation

Cost of gas

Cost of gas - revisit

COVID-19 Cases

What is wrong with this graph? How would you correct it?

COVID-19 Cases - redo

Source: https://livefreeordichotomize.com/posts/2020-05-17-graph-detective

Brexit

• What is the graph trying to show?

• Why is this graph misleading?

• How can you improve this graph?

Source: https://humansofdata.atlan.com/2019/02/dos-donts-data-visualization

Exercise and cancer

Exercise Can Lower Risk of Some Cancers By 20%

People who were more active had on average a 20% lower risk of cancers of the esophagus, lung, kidney, stomach, endometrium and others compared with people who were less active.

Alice Park. Exercise Can Lower Risk of Some Cancers By 20%. Time Magazine. 16 May 2016.

Exercise and cancer - another look

Exercising drives down risk for 13 cancers, research shows

[…] those who got the most moderate to intense exercise reduced their risk of developing seven kinds of cancer by at least 20%.

Melissa Healy. Exercising drives down risk for 13 cancers, research shows.
Los Angeles Times. 16 May 2016.

Spurious correlations

• What does this graph show?

• Work with your neighbor to write a misleading news article title based on this graph. (Submit on canvas as today’s check in. Access code: spurious)

Source: https://www.tylervigen.com/spurious-correlations

Catalan independence

On December 19, 2014, the front page of Spanish national newspaper El País read “Catalan public opinion swings toward ‘no’ for independence, says survey”.

Alberto Cairo. The truthful art: Data, charts, and maps for communication. New Riders, 2016.

Catalan independence - with uncertainty

Alberto Cairo. “Uncertainty and Graphicacy: How Should Statisticians Journalists and Designers Reveal Uncertainty in Graphics for Public Consumption?”, Power from Statistics: Data Information and Knowledge, 2017.

Algorithmic bias

Google translate - a few years ago

Google translate - today

Criminal sentencing

There’s software used across the country to predict future criminals.
And it’s biased against blacks.

Julia Angwin, Jeff Larson, Surya Mattu, and Lauren Kirchner. Machine Bias. 23 May 2016. ProPublica.

A tale of two convicts

ProPublica analysis

• Data: Risk scores assigned to more than 7,000 people arrested in Broward County, Florida, in 2013 and 2014 + whether they were charged with new crimes over the next two years.

• Results:

  • 20% of those predicted to commit violent crimes actually did

  • Algorithm had higher accuracy (61%) when full range of crimes taken into account (e.g. misdemeanors)

  • Algorithm was more likely to falsely flag black defendants as future criminals, at almost twice the rate as white defendants

  • White defendants were mislabeled as low risk more often than black defendants

How to write a racist AI without trying

Thomas Lumley. How to write a racist AI in R without really trying.
Biased and Inefficient. 27 September 2018.

Chat GPT

• Do you use Chat GPT? How?

• How, if at all, does Chat GPT (or other similar Large Language Models) reflect human bias?

• How would you feel if a company use AI-assisted chatbots to respond to your product return requests? How about if your professors responded to your emails with a similar chatbot? How about if your therapist responded to your inquiries with a similar chatbot?

Data privacy

Web scraping

A data analyst received permission to post a data set that was scraped from a social media site. The full data set included name, screen name, email address, geographic location, IP (Internet protocol) address, demographic profiles, and preferences for relationships. The analyst removes name and email address from the data set in effort to deidentify it.

• Why might it be problematic to post this data set publicly?

• How can you store the full dataset in a safe and ethical way?

• You want to make the data available so your analysis is transparent and reproducible. How can you modify the full data set to make the data available in an ethical way?

Further reading

Further reading

• Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence by The White House

• Machine Bias by Julia Angwin, Jeff Larson, Surya Mattu, and Lauren Kirchner

• Ethics and Data Science by Mike Loukides, Hilary Mason, DJ Patil (Free Kindle download)

• Weapons of Math Destruction How Big Data Increases Inequality and Threatens Democracy by Cathy O’Neil

• Algorithms of Oppression How Search Engines Reinforce Racism by Safiya Umoja Noble

• Why Pokémon Go’s plan to 3D-scan the world is dangerous by Loren Smith

• How Charts Lie: Getting Smarter about Visual Information by Alberto Cairo

• Calling Bullshit: The Art of Skepticism in a Data-Driven World by Carl Bergstrom and Jevin West

Time permitting

Project questions?

• Any content questions?

• Any formatting questions?

• Any requests for live review?