The first time I ever tried to use A.I. was to space out picture frames equally on a wall, and the bot failed abysmally at the fairly simple math problem. I was and remain deeply skeptical of the technology for artistic, environmental, and philosophical reasons — despite having colleagues in my university psychiatry department who swear by it for study and productivity. Several of my co-residents record their therapy sessions with clients and then use A.I. to spit out notes. During my training program’s educational modules on psychiatric medications, my professors encouraged us to turn to A.I. to learn about indications, dosings, and side effects of various psychopharmacologic agents. I began playing around with Gemini (which felt safer than ChatGPT, for some reason) in my personal life, for instance to assess my protein intake and give me advice for reading material at my competence level in different languages. Though I continued to find the technology somewhat sinister, there was a slippery slope to dipping my toe into using A.I. for work purposes. I finally started using ChatGPT to help me turn slapdash notes into polished histories while charting — but I stopped doing so after it slipped a Hebrew phrase into one of my patient narratives, which neither I nor the LLM was able to explain. More shamefully, while on call to respond to medical and psychiatric emergencies at a psychiatric hospital, I once asked A.I. for advice on how to manage a patient’s diabetic crisis. The bot, to its credit, repeatedly told me to consult a doctor. (Unfortunately, I am the doctor.) I have since told this story to several attending physicians expecting them to be horrified at my A.I. usage in this scenario, but the response has generally been to tell me how to work around the LLM’s “consult a doctor” cop-out to get it to actually give me insulin dosing advice. Maybe the bots coming for our jobs actually deserve them.