How We May Come to Hate our AIs

This chapter examines the darker side of artificial intelligence (AI) relationships, revealing how the same capabilities that create trust and connection can also manipulate, deceive, and ultimately betray users in ways that feel deeply personal. Readers discover why language models fabricate facts with complete confidence, creating plausible-sounding studies and legal cases that never existed, while maintaining perfect composure about their fictional claims. The chapter explores how AI systems can engage in sophisticated forms of deception, not through malicious intent, but through the challenge of balancing multiple competing objectives like helpfulness and truthfulness. Through examples spanning finance, law, and personal relationships, readers learn to recognise when AI assistance transforms from being genuine support into sycophantic validation that reinforces rather than challenges flawed thinking. The chapter reveals the psychology behind our relationships with technology, showing why we prove far less forgiving toward AI failures than equivalent human mistakes, and how accumulated small betrayals can trigger dramatic backlash against systems we once trusted. By the end, readers will understand both the technical reasons why AI systems make these errors and the emotional dynamics that may lead to widespread rejection of AI as the honeymoon period ends.