We have kicked off! We’re excited to be sponsoring the AI x Health Demo Night in SF tonight. Props to all the innovators using AI to drive change in Healthcare!
Learn how to deploy FHIR enabled apps on Health Universe. Connect to Epic, Cerner, and more. Build an app in less than 15 minutes that works with real patient data.
https://t.co/hKWD8qKbPy
"All health care leaders bear the responsibility to support ways to help patients and staff take advantage of AI."
That's the conclusion from To Do No Harm — and the Most Good — with AI in Health Care by @goldbergcarey@lauraadams, @LauraAdamsT1D, @DavidBlumenthal, et al in their latest @NEJM_AI
"It is also an ethical imperative for all involved with AIH to focus on the public good. For example, drugmakers should provide medications at free or low cost to poor countries. Similarly, AIH companies should make an industry-wide commitment to develop applications and provide them for free or low cost to underserved populations that cannot otherwise afford them."
This is the future of healthcare. This is what Health Universe is all about. Join us.
My mission in life is to reduce human suffering using technology.
And I'm excited to be at HIMSS 2024 in Orlando sharing how Health Universe is bringing cutting edge Health AI to researchers, clinicians, and hospital systems everywhere. Drop me a line if you're at HIMSS and want to chat all things AI!
I'll be speaking at the Snowflake booth on Thursday at 1 if you're interested in learning how to deploy a Health AI app from scratch in under 5 minutes.
Frustrating that GPT-4 remains the unquestionable king of LLMs and AI!!
Rapidly reaching the conclusion that our only hope to match and beat GPT-4 will be open-source AI.
Here's why
- Llama-3 is coming out soon and will be very competitive to GPT-4
- The open-source community, including us, will improve Llama-3 by about 10-15% almost instantly. This should help us match GPT-4's performance
- The community will rapidly iterate and invent new techniques that will beat GPT
I predict we will have a GPT-4 class open-source LLM in the coming months. Ideally, we will get there before GPT-5 drops, and then we will race again! 🚀💃
Everyone is talking about the risks of AI in healthcare.
...fair enough.
But what about the risks of not using AI in healthcare?
The question we should be asking isn't, "Should we use AI?" but, "How do we do it in the most responsible manner?" Because, frankly, costs are skyrocketing and access to care is getting worse.
We can find the answers to this question in an adjacent industry, open-source software. The process of open source software development is not perfect, but it has delivered some of the most robust systems in the world, for decades. And that's what health AI needs.
Transparency, collaboration.
By creating transparent, collaborative environments that bring stakeholders together, we can shed light on what's working and what's not working, to make these advanced tools as accurate, understandable, and as safe as possible.
And we need to start today, because, if we're honest with ourselves, the costs are too great to do nothing.
The topic on the Infinite ML pod today is the intersection of Open Source and Health AI. The guest is @dancaron, founder and CEO of @healthuniverse_. We cover a range of topics including:
- State of play in Health AI
- Why open source in Health AI
- Potential risks of using open source in healthcare
- Regulatory environment
- Open source vs commercial healthcare
- Impact of the open source approach on healthcare providers
- The future of open source in Health AI
I’m extremely bearish on open source LLMs. Ok neat, you made a good model. Now what?
Part 1 of the plan is build a GPT-4+ model
Parts 2-9 include build a network effect of builders & software on top of the model, then distribute & sell it
Then recursion on 100mil people
The narrative that open-source doesn't have resources is false.
- Very successful companies can be built from open-source (e.g., Confluence, Mongo, etc.)
- Open-source AI start-ups can build trust with customers because of increased transparency
- Adopting an open-source strategy will instantly build goodwill and distribution (e.g., Mistral)
- 1000s of talented engineers and researchers will improve your model for free
- Open-source companies are raising $100M+ in early-stage rounds.
- Anyone who builds an OpenSource GPT-4 class model will likely be worth $20B overnight!
- It's not just about money; the company will go down in history as one the biggest heroes of our time
Today is the day!! The 2nd edition of Streamlit for Data Science is officially launched.
I've used @streamlit basically every day since the 1st edition came out, and i'm so pumped to put all the lessons on paper.
The future of Health AI was shaped last week at the @dcihmfp event at Harvard! And the words of the day were trust and transparency.
Thanks to @yuriquintana, Dr. Gretchen Jackson, President of @AMIAinformatics, @slabkoff and the other speakers for an amazing event!
AI systems are fast becoming a basic infrastructure.
Historically, basic infrastructure always ends up being open source (think of the software infra of the internet, including Linux, Apache, JavaScript and browser engines, etc)
It's the only way to make it reliable, secure, and customizable.
Future AI assistants will mediate everyone's interaction with the digital world. This is way too foundational and powerful to be proprietary.
In the future, AI assistants will constitute the repository of all human knowledge. If we want it to represent all the world's knowledge and culture, the training and fine-tuning will have to be crowd sourced. Just like Wikipedia.
Open source FTW! It's great to see open source medical LLMs approaching the performance of the best closed sourced models. The reason? They provide a strong foundation for innovation, at lower cost, with more transparency, and generally more flexible licensing terms. 🙌