Justin Flechsig

AI agents and workflows are easier (and cheaper) than ever to launch, and they're topping hospital c-suite's list of priorities; with that said, the lack of interest, research, and investment into safety and observability isn't ideal.

Two rulings in the same week (Anthropic, Meta) say training LLMs on copyrighted books qualifies as fair use. Judges call the wins narrow: Meta case lacked market-harm proof; Anthropic still faces a December piracy trial. Honestly surprised both swung that way.

I tried to sign in to my corporate Twitter (oops, X) account today and instead birthed a brand new account. I was gifted with a view into the new user experience: antivax propaganda, hardcore racism, ivermectin clinic ads, and an unironic "sigma grindset" guru. Society is cooked. #Twitter #X

Prof. Mollick, AI's pace is staggering and HealthBench necessary. That said: the narrative of AI > docs is misguided at best, and HealthBench's own methodologies punish brevity (a function of real-world patient encounters). At worst, this narrative fuels unsafe exec staff cuts and staffing ratios.

OpenAI’s HealthBench just dropped. My biggest concern is execs seizing on it as “evidence” to justify deeper clinical cuts. We’re drifting toward AI replacing experts without a real grasp of its limits -- and the 180 on healthcare sentiment toward AI gives me whiplash.

AI takes center stage: What 67 healthcare leaders are investing in this year - Becker's Hospital Review | Healthcare News & Analysis budget, investment, AI, LLM, documentation AI, ambient AI, generative AI, RCM, cybersecurity, money, spend

Becker's asked 67 hospital IT leaders about their top investments for 2025. 94% mentioned AI -- with over 75% ranking it first. Only two highlighted safety & observability as key priorities. What's truly safe and helpful for frontline providers? We analyzed exec's top investments and rationale:

AI development is moving FAST -- what was cutting-edge two weeks ago is outdated today. But with real progress comes a flood of vaporware. Hospitals adopting AI need leadership (or a partner) who truly understands observability and safety. Otherwise, it's just adding complexity and wasting time.

Too many organizations see AI as a way to replace clinicians. That’s the wrong approach. AI should work with providers—not push them out.