⚡ Updated 11.21.23
Large language models (LLMs) will have a profound impact on how to build technology in healthcare, and we want to share our perspective on their use cases. We have spent the first few months of 2023 testing where LLMs make sense in the Oscar tech stack, and in healthcare more broadly, by building and launching applications. Through our rapid iteration, we’ve been able to amass good hands-on insights.
Working with language models almost requires a philosophical shift, not just the use of a new tool. Because the research is moving so fast and it is so hard to predict, it is our belief that the only way to deal with it is to constantly run experiments, pool all of the insights centrally, and push them back out to the org to decentrally run more experiments. We're referring to this approach as a continuous hackathon.
As a tech-driven health insurer with lots of well-organized data, we have a unique purview. By sharing our findings, we believe we may be able to help and enable others in the space. This page is our commitment to sharing our knowledge. So: first, we will share some learnings and thoughts. Second, our framework and our ongoing product launches using language models. Third, our live notes tracking the technological developments in the space.
Our current perspective on applying large language models in healthcare
Specifics on our approach and application
The important concepts to consider when deploying LLMs
04 AI Insights & Reading Notes
Our up-to-date knowledge repository