Summary

In this chapter, we provided an overview of the potential harms of AI and automated decision-making. The chapter reviewed examples of AI harm in hiring and recruitment, facial recognition, biased natural language models, discriminatory impact, attention engineering, social media, and AI’s environmental impact. It also discussed autonomous weapon systems and military use cases. It was important to look at these examples because they highlight the potential negative consequences of using AI and the need for proper governance and risk management. By understanding the potential risks of AI, we can work toward developing more responsible and ethical AI systems.

In the next chapter, the focus shifts toward the methods that make explainable and interpretable AI possible. It covers a taxonomy of machine learning interpretability approaches, including global and local methods, debugging, and audit. The advantages and disadvantages of these techniques will be reviewed, along...