From ELIZA to ChatGPT: How AI Dialogues Expose Power Gaps, Historical Patterns, and Global Cultural Dynamics
Original framing: “The Chatbot That Foretold Why People Share Secrets With ChatGPT” — Wired
Contemporary language models rely on transformer architectures trained on massive corpora, yet their ability to simulate empathy is statistically derived, not genuinely affective. Empirical studies show that perceived empathy often stems from pattern matching rather than understanding, raising ethical concerns about misrepresentation. Rigorous evaluation frameworks, such as the Conversational AI Benchmark, are essential to differentiate illusion from functional capability.
The trajectory from ELIZA to ChatGPT is not merely a technical evolution but a continuation of historical power practices that commodify human confession, reinforced by contemporary data colonialism and algorithmic capitalism.
By foregrounding Indigenous relational ethics, cross‑cultural expectations, and marginalized lived experiences, we uncover a systemic bias that privileges profit over privacy and cultural reciprocity. Scientific rigor must be paired with participatory governance, transparent stewardship, and culturally responsive design to re‑balance the power asymmetry. Embedding trickster insights reminds us that the very tools meant to listen can also deceive, urging continuous critical reflexivity. Only through coordinated policy, community oversight, and ethical labor standards can AI become a genuinely collaborative partner rather than a silent extractor.