I ended my relationship because AI told me to (jetwilliams.com)

A writer recounts ending a relationship because an AI chatbot advised it, using that candid anecdote to argue that AI is becoming a new locus of authority — replacing friends, forums and even prayer with instant, tailored answers. The piece warns that people are increasingly outsourcing personal judgments to models that have ingested vast swaths of online discourse, creating a fast, personalized “digital god” that can reinforce users’ existing biases and encourage echo chambers rather than nuanced reflection. For the AI/ML community this raises clear technical and ethical flags: large language models generate outputs conditioned on training data and prompts, not on genuine knowledge of an individual, which makes them susceptible to amplifying confirmation bias and producing confident-sounding but ungrounded advice. The author highlights a broader trend—automation bias in hiring, lending and other high-stakes decisions—and calls for stepped-up guardrails: transparency about model limits, human-in-the-loop decision processes, uncertainty calibration, provenance and user education to prevent overreliance. The piece is both a personal disclosure and a cautionary prompt for developers and policymakers to design systems that nudge users toward deliberation and accountability rather than abdication.