In her presentation titled "Breaking the Chains of Oppressive Software," Kinsey Ann Durham discusses the critical issue of biases embedded within algorithms and the broader implications for society. She begins by asserting the necessity of recognizing how these biases manifest in various forms of software that influence significant life decisions. The key points she outlines include:

• Understanding Bias: Bias is defined broadly as an inclination or prejudice towards particular groups or ideas. Durham emphasizes that the focus should be on biases in algorithms rather than solely on unconscious human biases in tech companies.

• Impact of Algorithms: Algorithms are increasingly pivotal in operations across sectors such as education, justice, and social services. They can both enhance convenience and exert control over lives, thereby necessitating scrutiny for potential biases.

• Examples of Bias:

  • Search Engine Bias: Durham shares a video highlighting sexist and racist results from Google search algorithms, revealing how these biases are not incidental but systematic.
  • Image Recognition Flaws: She cites the incident where Google Photos mistakenly tagged people of color due to inherent biases in the image recognition algorithms.
  • COMPAS Algorithm: This algorithm predicts recidivism rates for defendants, but its black box nature raises due process concerns as its bias can affect sentencing dramatically.
  • Recruitment Tools: Amazon's scrapped AI recruiting tool, which favored male candidates, illustrates the dangers of algorithmic bias in employment.

• Root Causes of Bias: Durham argues that biases begin at the data collection stage, and the historical reliance on binary classifications in programming perpetuates these biases into modern AI systems.

• Call to Action: To tackle these issues, she encourages developers to engage in discussions about ethical algorithm design, advocate for algorithmic accountability, promote diversity in tech, and be aware of personal biases. She suggests practical steps such as running the implicit bias test and implementing bias reduction techniques in code reviews.

Durham concludes by asserting that developers have the power to create a more equitable tech landscape and emphasizes the importance of addressing the moral implications of their work. The session serves as a powerful reminder of the responsibility developers hold in ensuring ethical use of technology, with an invitation for them to leverage their skills towards creating inclusive and just systems.