What AI Gets Wrong About Your Health

My Onix co-founder David Bennahum and I joined Dr. Meenal Agarwal on episode 119 of her Top 3 medicine podcast, Uncover Your Eyes, for a conversation about what happens when you hand your health questions to an AI that has never met you.

AI as Processed Food

Dr. Agarwal, a Doctor of Optometry and one of the experts building a Guide on Onix, opened with an analogy that stuck with me: today’s AI is a lot like processed food. It’s convenient, it arrives with confidence, and once a model has aggregated and homogenized its inputs, you’ve lost track of the source. In personal health, that missing context is exactly where the risk lives.

Three Questions

The conversation works through three of them:

• Why mainstream AI struggles with the nuance of healthcare
• Why understanding context matters more than producing an answer
• What a genuinely trustworthy health AI actually requires

A generic model answers everyone the same way because it knows nothing about anyone. Real guidance depends on context: your history, your body, and the expert you actually trust. That’s the difference between an answer and the right answer for you.

These models aren’t failing because they aren’t smart enough. They’re built to sound authoritative about a person they have never met. Onix is built the other way around: high-fidelity models grounded in a specific expert and in your own context, private to you.

Listen

The full episode is out now on Substack, Apple Podcasts, and YouTube. To meet the experts building Guides on Onix, visit onix.life.