Module: Adversarial testing of Generative AI
Facilitator: Rashid Kisejjere & Arthur Kakande, Pollicy
Session Description:
Desired Learning Outcome:
Understand the purpose and process of adversarial testing to identify vulnerabilities and biases in generative AI systems.
Learn how to design and execute tests to assess AI-generated content for misinformation, bias, and ethical concerns.
Explore examples of adversarial scenarios relevant to African journalism, including cultural misrepresentations and gender biases.
Gain hands-on experience in applying adversarial testing techniques to evaluate and improve the reliability and fairness of AI outputs in journalism.