Data Ethics, Part I

Data Literacy

Background

• A sub-discipline of Ethics of Technology
• World War II and the Atomic Bomb
• How do we build and use technology ethically?
• How do we collect and manage data ethically?
• Just because we can, does not mean we should.

• The atomic bomb was a turning point.
• Now had more devastating weapon than had ever existed.
• How do we manage a world in which technology has such awesome power.

What Happens Without Data Ethics: Positive?

• Increased performance
• Faster decision making.
• Techno-optimism

• Techno-optimism: the belief that technology can solve all problems and improve society.

Difference Between Ethics and Law

• Law says what we must do.
• Ethics, what we should do.
• Example:
  • It’s not illegal to collect publicly available data.
  • Is it unethical to scrape faces/names off social media for facial recognition?

What Happens Without Data Ethics: Negative?

• Discrimination.
• Privacy Issues.
• Accountability.
• Vulnerability of Public Relations.

• Discrimination: Data can be used to discriminate against certain groups of people, such as minorities
  • What we never want is for our analysis to negatively impact one group of people.
• Privacy Issues: Data can be used to invade people’s privacy, such as tracking their online activity or collecting personal information without their consent.
  • Collecting information without consent.
  • Collecting faces off Facebook.
• Accountability: Forces companies to be accountable for their own actions and decisions, which can lead to better decision-making and more responsible behavior.

Why Care About Data Ethics?

• Reputation
• Do not alienate users.
• Loss of revenue.
• Helps make models more accurate.

Balance Utility and Ethics

• Too much utility, you lose ethics.
• Too much ethics, you lose utility.
• The virtue is the mean between the two.

Neutrality Thesis

• Tools have no inherent moral value.
• Their use has moral value.

Embedded Value in Technology

• Technology built for specific purposes.
• Embedded Values create Embedded Consequences.
  • Example: bridge that does not permit buses due to low overhangs.

Co-Shaping of Technology

• Intention flows through the device.
  • Example: Gun extends lethal force to anyone holding it.
  • Example: Instagram extends power to communicate to its user.
    • What about the power of the algorithm?

Data Specific Issues

• Privacy Issues.
  • Example: Facial Recognition
• Discrimination Issues.
  • Example: Facial Recognition
  • Example: AI recognition software.
• How does the neutrality thesis fit here?