AI Guide for Healthcare: an honest path for those who want to really understand

There is no shortage of content about artificial intelligence out there — and very little of it is written by someone who actually sees patients. humaniza.dev is opening a trail of posts about AI, written by physicians, for healthcare professionals, with the sole ambition of explaining it without magic and without fear.

---

Why this, why now

Every month a new headline lands. "AI passed the board exam." "ChatGPT diagnoses cancer better than doctors." "The next model will replace radiologists." Each cycle brings the same whiplash: one week it's naive enthusiasm, the next it's existential panic.

In the middle of it all, the healthcare professional is stuck in a strange position. They feel they need to understand AI — because patients are already using it, because institutions are adopting it, because students keep asking. But almost every piece of available material was written by someone outside of medicine. The result is predictable: technically accurate explanations that are clinically naive. Or worse, clinical articles written by enthusiasts with no technical rigor.

This Guide exists to fill exactly that gap.

Whoever explains AI to you should understand both the model and the patient.

What the AI Guide for Healthcare is

It's an editorial trail — not a course, not a book, not a gated newsletter. It's a series of independent posts that work on their own but gain depth when read in sequence. Every post is published here on humaniza.dev, open, in Portuguese and English.

The trail is organized into four thematic blocks, which also work as four levels of depth:

What every post has in common: medical analogies when they illuminate, technical rigor when it matters, and an explicit section on limits and ethical considerations. No post in this trail will suggest that AI replaces clinical judgment. And none will hide what the tool actually does under the hood.

What you won't find here

Just as important as saying what the trail is, is saying what it isn't.

You won't find:

• Grand promises. Current AI is useful, limited, and dangerous when misused. All at once.
• Untranslated jargon. If "embedding" appears in a post, the next sentence explains it in clinical language.
• Examples with real patient data. Every demonstration uses a fictional case or public data. That's an editorial rule, not decoration.
• Disguised sales pitch. IRIS (Humaniza Health's clinical decision support platform) will appear as a concrete case study whenever it makes sense — but the posts explain concepts, they don't sell a product.

How to read this trail

You can read in any order, but there's a sequence that makes the most sense — the same sequence in which posts are published, one per week. If you're starting now, I suggest beginning with the next post in the trail:

Each post is independent. Read when it makes sense, come back when you need to, skip the ones that don't speak to your practice. But return to the trail filter whenever you want the full picture: the V0 navigation for this series currently lives at /en/blog?category=guia-ia-saude.

An editorial pact

I am a family and community physician, a medical school professor and preceptor, a scientific initiation advisor, and the founder of Humaniza Health — obsessed with technologies and tools that optimize our clinical practice. I develop the IRIS platform with intensive AI support — and it's precisely because I live both the builder side and the clinical side at the same time that I write this trail.

My commitment to the reader:

1. 01

   No important claim without a verifiable reference.

2. 02

   No technical concept without a clinical translation.

3. 03

   No tool recommended without listing its limitations.

4. 04

   No post published without a complete version in Portuguese and English.

5. 05

   No omission about ethical risks, data privacy, or professional liability.

A simple contract. It will sustain every post in this trail.

Next step

Next week the first content post goes live: "What AI hallucination really is — and why every physician needs to understand it". From there, one new post every week, following the Wave 1 order of the trail.

If you want to follow along, three ways:

• Return to the trail filter at /en/blog?category=guia-ia-saude.
• Follow humaniza.dev's Instagram profile, where every site post becomes a reel, a carousel, and a story with the main hook.
• Share with a colleague who needs this content. This trail was written thinking of a tired resident, an overwhelmed professor, and a curious preceptor — if you know someone like that, they're the audience.

Let's understand AI for real. Without magic. Without fear. With the seriousness that medicine deserves.