Sarah Rose Siskind shares her experimental project FetusGPT—an LLM trained exclusively on the audio environment surrounding her unborn child, from Seinfeld reruns to sleep sounds—highlighting how narrow, personal data shapes AI behavior much like early human language acquisition. She demonstrates how AI fuels her comedy writing by simulating brainstorming sessions and generating satirical content, though it often lacks the surprise essential to humor. Beyond creativity, AI plays a practical role in health: Siskind used ChatGPT to interpret medical data, optimize nutrition during pregnancy, and navigate emotional anxieties, including fears about pregnancy symptoms. These cases illustrate AI’s growing role in personalized health and emotional support, especially where traditional systems fall short. The conversation also touches on deeper themes, like gendered differences in knowledge production and the relational trust needed in both therapy and human-AI interactions, underscoring that while AI reflects what we feed it, its value ultimately depends on human judgment, context, and intention.