After 5 months, the same medical vignette is back, and 12 out of 13 LLMs we tested still can't solve it. So why are we calling this a success? You have to read the whole study to find out. 😉
This was a super interesting article! But I suspect the differences you're seeing are mostly from specific product decisions in the chat app, rather than any differences between individual SOTA models.
Which especially makes sense in the context of healthcare! Sure, each individual doctor is smart and well trained, but the "intelligence" of the system lies in the org structure, context management, and processes:
1/ The triage nurse: The initial prompt engineering and intent routing.
2/ The chart and patient history: The context window + maybe RAG pipeline
3/ The differential diagnosis: The explicit agentic loop mapping out multi-step reasoning as well as the system prompt
4/ Lab techs and specialists: The external tool use and continuous verification.
@Sergei, just saw Alibaba’s latest Medical AI product launch (H+) this week in China. Quite impressive, but low key media coverage abroad (only launched in China). Thought I’d share here.
It’s for Doctors as reliable AI research & clinical assistant, with new architecture and medical evidence data/refresh cadence to help redefine the std of Healthcare AI. Its core design principle is: Medical Evidence + Evidence-Based Medicine + AI = A More Reliable Medical AI Assistant.
Only through this combination—building upon a foundation of both rapid retrieval and precise accuracy—can the authority to verify evidence be fully and completely restored to the doctors themselves.
Below is a summary from Gemini (still very limited western media coverage):
“Alibaba Health's AI assistant for doctors, Hydrogen Ion (H+) is an evidence-based medical large language model (LLM) designed as a "GPT for doctors". It is purpose-built to act as an intelligent clinical and research assistant, focusing on low hallucinations and rigorous traceability.
Core Capabilities
-Traceable Evidence: Every response provides clickable citation tags so clinicians can verify statements, trace clinical conclusions, and directly access original literature.
-Dynamic Evidence Localization: Tracks global medical guidelines and literature daily, ensuring doctors receive "living evidence" rather than static references.
-Workflow Integration: Provides seamless bilingual (Chinese and English) medical Q&A, full-text reading, online translation, and literature analysis tailored to clinical workflows.
Architecture
The platform is built on a specialized four-layer AI architecture to ensure medical accuracy:
1. Evidence Comprehension: Understands and structures guidelines based on PICO frameworks and GRADE standards.
2. Retrieval-Augmented Generation (RAG): Enhances data retrieval to ensure outputs have the lowest possible hallucination rates in the medical domain.
3.Model Fine-Tuning: Trained extensively on medical scenarios to enforce strict standards of safety and clinical accuracy.
4.Expert Review System: Includes a closed-loop review system to maintain high quality.
Key Partnerships
-BMJ Group: Hydrogen Ion secured an exclusive content partnership in China, gaining licensed access to a decade of content and multimedia resources from 70 medical journals published by the BMJ Group.
-Expert Committee: Alibaba Health established a Medical AI Expert Committee featuring over 300 clinical experts and top academics to develop medical evaluation standards and datasets.”
Exactly, it’s about the right product success eval criteria (in this case, Accuracy and How it arrives at conclusion both matter a lot in Medical Diagnosis), training and eval,, architecture all are important when it comes to scalable solution.
Thank you both for sharing the thorough testing and thought-provoking analysis. Perhaps you could create a medical diagnosis benchmark arena with such medical 1st principles/reasoning and cases when Causality should be prioritized vs other metrics. Otherwise, LLM’s working mechanisms are basically pattern matching (probability). As you pointed out, MAD may not work if the right success criteria are not selected for the agents for different use purposes. Not suprised Google Gemini now has hardcoded temporal reasoning as they may have worked with most healthcare practitioners and recognized the proper success criteria for medical diagnosis.
I agree. We need a solid benchmark for medical diagnosis / clinical reasoning. As our exercise showed, strong pattern recognition alone may not be enough.
Btw, I’m in AI product for a different industry vertical (retail) and have done quite some Causal Inference Reasoning ML dev in additional to traditional pattern-learning ML. Really interested in Care space and would like to explore potential partnership in this space as side project. if you are interest, let me know!
This was a super interesting article! But I suspect the differences you're seeing are mostly from specific product decisions in the chat app, rather than any differences between individual SOTA models.
Which especially makes sense in the context of healthcare! Sure, each individual doctor is smart and well trained, but the "intelligence" of the system lies in the org structure, context management, and processes:
1/ The triage nurse: The initial prompt engineering and intent routing.
2/ The chart and patient history: The context window + maybe RAG pipeline
3/ The differential diagnosis: The explicit agentic loop mapping out multi-step reasoning as well as the system prompt
4/ Lab techs and specialists: The external tool use and continuous verification.
@Sergei, just saw Alibaba’s latest Medical AI product launch (H+) this week in China. Quite impressive, but low key media coverage abroad (only launched in China). Thought I’d share here.
It’s for Doctors as reliable AI research & clinical assistant, with new architecture and medical evidence data/refresh cadence to help redefine the std of Healthcare AI. Its core design principle is: Medical Evidence + Evidence-Based Medicine + AI = A More Reliable Medical AI Assistant.
Only through this combination—building upon a foundation of both rapid retrieval and precise accuracy—can the authority to verify evidence be fully and completely restored to the doctors themselves.
Below is a summary from Gemini (still very limited western media coverage):
“Alibaba Health's AI assistant for doctors, Hydrogen Ion (H+) is an evidence-based medical large language model (LLM) designed as a "GPT for doctors". It is purpose-built to act as an intelligent clinical and research assistant, focusing on low hallucinations and rigorous traceability.
Core Capabilities
-Traceable Evidence: Every response provides clickable citation tags so clinicians can verify statements, trace clinical conclusions, and directly access original literature.
-Dynamic Evidence Localization: Tracks global medical guidelines and literature daily, ensuring doctors receive "living evidence" rather than static references.
-Workflow Integration: Provides seamless bilingual (Chinese and English) medical Q&A, full-text reading, online translation, and literature analysis tailored to clinical workflows.
Architecture
The platform is built on a specialized four-layer AI architecture to ensure medical accuracy:
1. Evidence Comprehension: Understands and structures guidelines based on PICO frameworks and GRADE standards.
2. Retrieval-Augmented Generation (RAG): Enhances data retrieval to ensure outputs have the lowest possible hallucination rates in the medical domain.
3.Model Fine-Tuning: Trained extensively on medical scenarios to enforce strict standards of safety and clinical accuracy.
4.Expert Review System: Includes a closed-loop review system to maintain high quality.
Key Partnerships
-BMJ Group: Hydrogen Ion secured an exclusive content partnership in China, gaining licensed access to a decade of content and multimedia resources from 70 medical journals published by the BMJ Group.
-Expert Committee: Alibaba Health established a Medical AI Expert Committee featuring over 300 clinical experts and top academics to develop medical evaluation standards and datasets.”
Product link (they’d recommend doc user to register) https://yds.ali-doctor.com/app/doctor-msg-app/home?undefined
Super interesting. Thank you for pointing this out. 🙏
this is fantastic thank you
Merci. I'll share something soon focused more on the larger implication which is a bit concerning.
Glad I use Gemini LOL
Personally, I don’t think the hard-coded override Gemini used in this particular case is a long-term solution. But hey, it worked here.
Exactly, it’s about the right product success eval criteria (in this case, Accuracy and How it arrives at conclusion both matter a lot in Medical Diagnosis), training and eval,, architecture all are important when it comes to scalable solution.
I don't ask it for a medical diagnosis, tbh
Thanks for your constructive vigilance in pushing AI to be a more meaningful tool to improve patient care.
🙏
Thank you both for sharing the thorough testing and thought-provoking analysis. Perhaps you could create a medical diagnosis benchmark arena with such medical 1st principles/reasoning and cases when Causality should be prioritized vs other metrics. Otherwise, LLM’s working mechanisms are basically pattern matching (probability). As you pointed out, MAD may not work if the right success criteria are not selected for the agents for different use purposes. Not suprised Google Gemini now has hardcoded temporal reasoning as they may have worked with most healthcare practitioners and recognized the proper success criteria for medical diagnosis.
I agree. We need a solid benchmark for medical diagnosis / clinical reasoning. As our exercise showed, strong pattern recognition alone may not be enough.
Btw, I’m in AI product for a different industry vertical (retail) and have done quite some Causal Inference Reasoning ML dev in additional to traditional pattern-learning ML. Really interested in Care space and would like to explore potential partnership in this space as side project. if you are interest, let me know!
🔥