“It’s tough to make predictions, especially about the future.”

Yogi Berra, American baseball player with the New York Yankees

This famous quote from Yogi Berra is much more difficult with artificial intelligence. The year 2023 has been robust for artificial intelligence, with a major surge in interest and awareness as a result of the large language models like ChatGPT and GPT-4 as well as Google’s Bard and others. This transformer era of artificial intelligence has very much vitalized an imagination for using artificial intelligence not previously seen with the deep learning tools from the last decade. 

As we look back on the past year, the exciting developments from 2023 include:

1. The beginning of large language models (LLMs) like ChatGPT to be part of health systems clinical and administrative tasks

2. AlphaFold and its accomplishments in protein structure determination as well as drug and vaccine design

3. Continuing maturation of existing image focused AI tools in not only radiology but other fields as well, such as pathology, cardiology, and opthalmology

Here are my predictions for artificial intelligence in health and medicine for this upcoming year: 

  1. There will be increasingly higher interest in AI in health professional education. There is escalating interest amongst students and perhaps even with some faculty members now in health professional schools to include artificial intelligence as a topic that is relevant for the future, but the execution of this vision remains very (perhaps too) slow.  
  2. There will be even more focused effort in getting access to healthcare data. With LLMs training on trillions of tokens (GPT-4 on 13 trillion tokens), healthcare data as an AI commodity is more valuable than ever before, especially if healthcare organizations are to leverage this transformer  technology for generative AI in healthcare. 
  3. We will see the first health system and/or clinician(s) to be sued for failure to use an available AI tool in a medical legal lawsuit. With increasing number of healthcare organizations adopting tools such as the AI tool for neuroradiological imaging in cerebrovascular strokes, a bad outcome can trigger a lawsuit to claim that not having the AI tool falls below standard of care. 
  4. Health systems and hospitals will be looking to develop their own large language models. The desire to have an effective enterprise search for information based on its own documents can now be realized. There are already a few of these LLM-powered search platforms for institutions (Glean and Hebbia), but healthcare executives are still in the process of learning about LLMs. 
  5. There will be a desire to push for foundation models and multimodal AI systems as large language models seem more limited. Once the hype for ChatGPT has waned by mid-year, there will be more interest and desire to learn and adopt the more sophisticated foundation models and multimodal AI systems in generative AI (think of LLMs as appetizers before the entrees). 

 The remainder of the list of 10 predictions will be shared next week. 

Ai-Med’s Global Summit 2024 is bigger and better than ever! With a keynote stage bringing together the superstars of strategic and thought leadership for smart healthcare. Book your place now!