What are the four categories of unethical AI use described in the chapter?

The four categories of unethical AI use are AI-disinformation, AI-electoral-interference, AI-HR-manipulation, and unethical-AI-culture. AI-disinformation involves fake or misleading content to deceive or influence voters. AI-electoral-interference is the manipulation of democratic or voting processes. AI-HR-manipulation is the misuse of AI in human resource staffing to create inequality or discrimination. Unethical-AI-culture exploits cultural and personality factors to manipulate behavior.

The chapter introduces these four categories as distinct risks for unethical AI in democratic systems. First, AI-disinformation concerns fake or manipulated media such as deepfakes aimed at distorting the public's view of political figures or events. Second, AI-electoral-interference involves actively controlling or tampering with electoral processes, for example by deleting or creating data, suppressing voter credentials, or making it harder for people to vote. Third, AI-HR-manipulation refers to using AI improperly within human resource functions, such as biased recruitment that discriminates or unfairly promotes certain people. Fourth, unethical-AI-culture means applying AI to exploit cultural traits like uncertainty avoidance and psychological personality factors to manipulate citizens or employees.