Do Androids Dream of Human Doctors?
summary
Shows real failures and edge cases: noisy ECGs and Holter outputs that miss or flag issues (one trace yields 90 candidates), a pregnant ECG where fetal heartbeat confuses automated algorithms, and a COVID case where ChatGPT confirmed treatment but produced false reassurance. Demonstrates prompt testing with Llama 2 system message roles, cites a literature review with about 5–6 studies and only one formal evaluation, and points to EmotionBench for empathy testing. Argues to prioritize doctor-facing, interoperable AI tools, plus regulation, privacy, and measurable validation rather than patient-facing full replacements.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 1:06Medical AI history
- 3:39Medical scoring systems
- 6:28AI in medicine limits
- 7:52AI edge cases
- 8:34ChatGPT patient case
- 10:43Patient vs doctor AI
- 11:45Patient-doctor power
- 13:10Holistic care and AI
- 14:16ELIZA psychotherapy
- 16:04Psychiatric diagnosis limits
- 18:13Skiing trip case study
- 20:23LLM empathy roles
- 22:53Empathy in medical advice
- 24:18AI empathy testing
- 25:47Turing test critique
- 27:14AI patient problems
- 29:22Two AI dreams
- 30:28Empower doctors with AI
related talks
39mHuman Values in Clinical Algorithms, Old and New
21mGen AI in Health: What's New in 2026?
01h 00mKeynote Panel: AI in Healthcare
36mThe Case for AI Pessimism in the Age of AI Optimism
28mThe Future of Healthcare – AI’s Promise and the Challenges Ahead
13mWhen AI Gets a White Coat
04mBlack Box: How AI will Amplify the Best and Worst of Humanity
29mAI for COVID-19: An online virtual care approach