In healthcare AI, the scarce skill was never coding or medicine. It’s the crossbar between them: knowing what’s worth building and whether it survives the ward. AI tooling didn’t create that bottleneck. It just made it impossible to ignore.

You're not afraid of AI. You're afraid of what it reveals, that you stopped learning a long time ago.

AI is already in daily healthcare work: imaging, documentation, scheduling, decision support. What’s missing isn’t the tools. It’s the training to use them well. That’s what HelloAI is for. It gives healthcare staff the knowledge & skills to use AI efficiently & responsibly. helloaiprofessional.com

Reading Mathias Goyen’s new book this week between airports. The Things We Know to Be Wrong (But Do Anyway). Four chapters in, I keep underlining things.

Tested the 5x AI plan last month. Went back to the normal one today. Lasted 10 minutes before upgrading again. It’s remarkable how fast a new capability stops feeling like a luxury and just becomes… how things should work. That’s true for tools, habits, and probably a lot more.

I've spent years telling clinicians that AI literacy matters. So GE HealthCare made the full HelloAI Professional track available at no cost. 20+ hrs. Real case studies. Responsible adoption frameworks. For anyone who wants to understand what's actually happening when AI touches patient care.

Karen Hao at HIMSS26: AI companies are empires. Just with better PR. Seize data. Automate labor. Dominate research. Wrap it in a moral mission. Colonial empires called it God and civilization. AI companies call it preventing AI hell. Same playbook. @himssglobal.bsky.social @karenhao.bsky.social

AI missed the finding. That's the headline. Nobody writes about AI that never got deployed.

Hospitals do not have an AI adoption problem. They have a control problem. AI shifts decision gravity away from hierarchy and toward data. — From Hard Truths About AI in Healthcare by Jan Beger hardtruths.info#t2

We’ve reached a new era of computing when your laptop is buckled into the back seat because Claude Code is still working and you don’t want to disturb it.

The next decade will be defined by whether we can turn momentum into infrastructure. That's the real deliverable. research.gehealthcare.com/patient-care...

99 Hard Truths About AI in Healthcare is rebuilt. Every truth now lives as a sharable quote card. One click, you save the PNG. Then post it. LinkedIn, X, Bluesky, wherever you post. hardtruths.info

AI will never fully replace human oversight. At least not while humans are held legally responsible for its mistakes.

AI is the greatest EQ stress test humanity has ever faced.

Every $1 of AI tech requires $10 of invisible work. Most health system budgets don't account for it. This week’s Second Opinion is about that gap. janbeger.substack.com/p/the-70-pro...

The question isn't whether AI works. It's whether your organization has built the governance, measurement, and workflow integration to know if it's working for you. Most haven't. That's where the real work is. research.gehealthcare.com/across-the-e...

More AI pilots in US hospitals than actual pilots at United. A third will reach production. The rest stay in committee. Yes, the stakes are higher in healthcare than anywhere else. That’s the reason to do this carefully. It’s not the reason to never finish. Sometimes the answer is on a hat.

US hospitals are running more AI pilots right now than United Airlines employs actual pilots.

Want the logic? Let me know.

We have a literature full of studies showing AI can detect pathology. We have almost no literature on what happens when it doesn’t get used. Non-adoption is the silent harm.

Healthcare needs AI because it needs the human touch. I’ve been saying this for years. Now there’s an economic model that explains why.

My TEDx talk is available now on YouTube. Let me know your thoughts.

6M Americans asked AI chatbots for health advice last month. 4% strongly trust it. That’s not AI adoption. That’s a broken healthcare system pushing people toward whatever works.

Hospital CEOs want to replace radiologists with AI. Most hospitals can't deploy basic AI yet. The ambition at the top. The operational reality at the ground floor. They don't match, and the consequences are real. Let me know what you're seeing in your environment. open.substack.com/pub/janbeger...

Most healthcare AI lives in the radiology reading room. Most of the clinical value doesn't. Hosting a HLTH masterclass with Dr. Amine Korchi on what's actually working beyond imaging. April 22, 11AM ET. Virtual. hlth.com/virtual-roun...

Human review of AI output is least effective precisely when AI is most confident. That's the moment oversight is most assumed and most broken.

The AI tools that are actually running in hospitals didn't win on technical performance. They won because someone could show real-world benefit, clinicians could fit them into their day, and the IT infrastructure was there to hold them up.

Sunday morning. Barbell loaded, Claude Code running. Almost perfect.

Proud to be named one of The Imaging Wire’s Top 40 Radiology Resources again. Means a lot. theimagingwire.com/2026/03/22/t...

This is my entire Monday philosophy. That's it. That's the post. Happy Monday.

Every AI demo shows what the tool can do. Nobody shows what the hospital needs to run it.

Seeking warm intros to editors/agents for a kids’ responsible‑AI book (7–12). Nonprofit, sample spreads ready. DMs open.

Here are the ten papers that captured my attention in February. I hope the insights (and the ideas they spark) are useful to you. www.linkedin.com/pulse/intere...

We talk about AI safety. We rarely talk about AI status loss. The political one stalls more projects. That’s 1 of 95 Hard Truths About AI in Healthcare. Just published the March 2026 update. A lot changed since June. Read it. Tell me which ones are wrong. hardtruth.carrd.co

Vienna bound for ECR. Ready for a few packed days of ideas, familiar faces, new connections, and probably too much coffee. Looking forward to catching up with partners, customers and friends. If we’re already working together, or should be, let’s try to find a slot.

Will I see you at ECR in Vienna this week?

As AI models commoditize, the gap between vendors in the healthcare industry won’t be in the architecture. It’ll be in the three seconds between a finding and the next click.

Is it funny cause it’s true?

Matt Shumer just described what happened to his job in the last couple of months, and it’s the most honest thing I’ve read about where this is actually heading. If you’ve been dismissing AI based on your experience from 2023, this will change your mind.​​​​​​​​​​​​​​​​ shumer.dev/something-bi...

AI has reached a tipping point, moving from narrow tools toward scalable clinical infrastructure in defined use cases. Multimodal and agentic AI can help enable more proactive, connected care, while keeping the human at the center. Thanks TEDx New River for having me.

AI has reached a tipping point, moving from narrow tools toward scalable clinical infrastructure in defined use cases. Multimodal and agentic AI can help enable more proactive, connected care, while keeping the human at the center. Excited to share this perspective today at TEDx New River.

If I can explain it to AI, I understand it. If not, I keep going. Confusion isn’t failure. It’s incomplete thinking.

This is a love story. A drama. A science fiction novel that sometimes feels like fantasy. Or at least, that’s how it feels when you’ve spent most of your professional life where artificial intelligence meets healthcare ... janbeger.substack.com/p/healthcare...

Stop asking whether AI can beat a radiologist on hard cases. Ask whether it can stop a radiologist from missing an obvious, catastrophic finding on a terrible night.

Biggest lesson AI taught me wasn’t about tech. It was about communication. Every bad prompt is a mirror. It shows you exactly where your thinking is fuzzy. AI doesn’t just make you smarter. It shows you where you’re not. 🤓

Chaos. German style. 😉

An AI habit more people should use: force disagreement. When a decision matters, I don’t ask one AI. I set up a few. The yes voice. The no voice. The numbers brain. The context brain. One AI is good at echoing your thinking. A small group that’s built to push back is how you get better answers.

The hardest part of AI transformation is not choosing the right tools. It is redesigning governance systems built for a slower, more static healthcare era.

I don’t use AI to get answers. I use it to argue with my ideas. I’ll toss out something half-baked and say, poke holes in this. What’s weak here? I’m not trying to win the argument. I’m trying to see what survives.