Ethics in the Age of Gen AI

Important Links

Key Takeaways


Three frameworks while suggesting clients:
  1. Responsible data practices
    1. What is the source of the training data?
    2. What has been done to reduce bias in the data?
    3. How might the data we're using perpetuate historic bias?
    4. What opportunities exist to prevent biased decision-making?
  1. Boundaries on safe and appropriate use and
    1. Who is the target population for this tool?
    2. What are their main goals and incentives?
    3. What is the most responsible way to achieve these goals?
  1. Robust transparency
    1. How did the tool arrive at its output?
    2. What other ways do we have of testing fairness?
    3. Can decision makers easily understand the input-analysis-output process?
    4. Have you engaged with a broad range of stakeholders?

Preparation for AI Ethics analysis

Ethical data organization can be divided into three parts:
  1. Prioritizing privacy
    1. Conduct a privacy audit
    2. During a privacy audit, you build a comprehensive understanding of:
      1. what data your organization has,
      2. how it was collected,
      3. how it's stored, and
      4. how it's administered.
    3. The results of a policy audit inform recommendations to create or adapt your existing privacy policy to protect sensitive data.
    4. Create a training curriculum for employees.
  1. Reducing bias and
    1. Who are the end users
    2. Is the input data set diverse enough?
    3. Consult who interpreted the data
  1. Promoting transparency

⚠️Disclaimer: All the screenshots, materials, and other media documents used in this article are copyrighted to the original platform or authors.