50 Michael Kearns¶

American computer scientist

Michael Justin Kearns is an American computer scientist, professor and National Center Chair at the University of Pennsylvania, the founding director of Penn's Singh Program in Networked & Social Systems ...

Website: https://www.cis.upenn.edu/~mkearns/

Source: Wikipedia

• Born: California
• Doctoral students: Jennifer Wortman Vaughan
• Other notable students: John Langford (postdoctoral visitor)
• Education: Harvard University and University of California, Berkeley
• Affiliation: University of Pennsylvania
• Research interests: Machine Learning, Algorithmic Game Theory, Network Science, and more

The Main Arguments¶

• Algorithmic Fairness and Bias: Kearns emphasizes the critical need to address bias in algorithms, especially in sensitive areas like lending and criminal justice. He argues that algorithms can reinforce existing societal biases if fairness is not prioritized in their design. This highlights the ethical responsibility of data scientists and engineers to create algorithms that do not perpetuate discrimination.

• Trade-offs in Fairness: The discussion reveals that achieving fairness often involves trade-offs, such as between accuracy and fairness. Kearns explains that optimizing for one aspect may lead to compromises in another, which is crucial for stakeholders to understand when implementing algorithms in real-world scenarios.

• Group vs. Individual Fairness: Kearns discusses the tension between group fairness (ensuring fairness across demographic groups) and individual fairness (treating individuals equitably). He notes that while group fairness can be easier to measure, it may overlook the nuances of individual circumstances, which is vital for creating truly ethical algorithms.

• The Role of Human Values: Kearns argues that human values must be integrated into algorithmic design. He stresses that engineers should not only focus on technical performance but also consider the societal implications of their algorithms. This perspective is significant as it calls for a more holistic approach to technology development.

• The Need for Behavioral Research: Kearns advocates for more behavioral research to understand how people perceive fairness and ethics in algorithms. He suggests that current definitions of fairness are often based on theoretical constructs rather than real-world perceptions, which can lead to misalignment between algorithmic outcomes and societal expectations.

Any Notable Quotes¶

• "Most people are good and want to do right, and deviations from that are usually due to circumstance."

• This quote encapsulates Kearns' optimistic view of human nature and sets the tone for discussions about ethical algorithms.

• "Algorithmic fairness is going to be kind of a harder problem and it will have to revisit things that have been thought about by many generations of scholars before us."

• This highlights the complexity of fairness in algorithms and the need for interdisciplinary approaches.

• "Fairness is a much more loaded topic than privacy."

• Kearns points out the contentious nature of fairness discussions, which often involve deep societal values and beliefs.

• "We are kind of looking where the light is."

• This metaphor illustrates the tendency to focus on easily measurable aspects of fairness while potentially ignoring deeper, systemic issues.

• "The algorithms themselves should not decide what the right social norms are."

• Kearns emphasizes the importance of human oversight in defining ethical standards for algorithms, reinforcing the need for accountability in technology.

Relevant Topics or Themes¶

• Ethics in Technology: The episode delves into the ethical implications of algorithmic decision-making, particularly in high-stakes areas like finance and criminal justice. Kearns argues for a framework that incorporates ethical considerations into algorithm design.

• Algorithmic Transparency: Kearns discusses the need for transparency in algorithms, suggesting that stakeholders should understand how algorithms make decisions. This theme connects to broader societal demands for accountability in technology.

• Interdisciplinary Collaboration: The conversation highlights the importance of collaboration between computer scientists, ethicists, and social scientists to address the complexities of fairness and ethics in algorithms. Kearns advocates for a more integrated approach to research and development.

• Public Perception of Algorithms: Kearns emphasizes the gap between technical definitions of fairness and public perceptions. He argues that understanding how people view fairness is crucial for developing algorithms that align with societal values.

• The Future of Algorithmic Governance: The episode raises questions about how society will govern the use of algorithms in the future. Kearns suggests that as algorithms become more pervasive, there will be a growing need for frameworks that ensure ethical use and accountability.

• Political Discourse and Algorithmic Influence: Kearns discusses the role of algorithms in shaping political discourse and the importance of exposing users to diverse viewpoints. He suggests that platforms like Facebook could experiment with showing users content outside their usual interests to foster tolerance and understanding.

• Differential Privacy: Kearns explains the concept of differential privacy as a stronger notion of privacy compared to traditional anonymization techniques. He discusses its significance in protecting individual data while still allowing for useful data analysis.

Overall, the episode provides a comprehensive exploration of the ethical challenges posed by algorithms, emphasizing the need for a thoughtful and inclusive approach to their design and implementation. Kearns' insights into the intersection of technology, ethics, and societal values offer a roadmap for navigating the complexities of an algorithm-driven world.