Thoughts on Generative AI
I’ve been building tech companies for the past 25 years and I have yet to see anything like what has occurred over the past few months with generative AI, and that means a lot given that I’ve seen the birth of the Internet and the smartphone.
Arguably, there has never been a technology advancement more profound than what we have seen over the past several months with generative AI and Large Language Models (LLMs), as we now have tools to not only use artificial intelligence to interrogate unimaginable amounts of data, but we have also built interfaces between humans and machines that make that interrogation seamless.
Interfaces that write papers and books, make movies and videos, tell stories, argue and debate. Write code, develop software. Interfaces that talk to you, listen to you, and learn.
Entire industries will likely be transformed by these technologies, healthcare among them. I’ve spent the last 7 years, head down, building Tempus with the singular objective of bringing the power and promise of artificial intelligence to diagnostics, and thus to healthcare, to help fight diseases like cancer.
We originally set out to connect providers to our technology in real time, allowing us to help physicians make data-driven decisions and help researchers be more efficient. As our database grew (it’s currently approaching 150 petabytes), we focused on structuring the multimodal, de-identified healthcare data we were collecting as the first step to deriving insights. Generative AI allows us to rethink our approach and instead deploy AI-enabled Agents across our data, creating self-learning models across disparate datasets, delivering insights and analytics without the need to put all of our data in neat little rows and columns.
This isn’t just a big deal, it’s THE deal. When you can query inherently messy data and find patterns that produce insights without having to clean and structure data, it’s transformative. Imagine being able to look across an entire healthcare system’s data and instantly identify patients who would benefit from targeted interventions, or being able to build incredibly complex cohorts of interest in seconds just by talking into your phone. All without having a PHD.
A few years ago, we launched a product called Tempus One, that was designed for this very purpose. It was meant to build a bridge between a single physician making a single treatment decision and the totality of wisdom that exists across the entire medical field to ensure that decision is optimal.
When we designed One, we knew we were ahead of our time and it would be an uphill battle. ChatGPT, Bard, and other advancements in the field have shortened our climb. It is now not only possible, but highly probable, that these types of technologies will become ubiquitous across healthcare.
Within a few years, patients will be able to upload their test results after a routine physical and it will be as if the 100 smartest primary care providers in the world spent thousands of hours pouring over the results, along with the patient’s entire medical record, to come up with a series of highly curated recommendations to optimize health and wellness. And these insights will be rendered in seconds, right through your phone, allowing you to interact with your doctor on an entirely different, and inherently more level, playing field.
Tempus was designed to bring AI to diagnostics, because diagnostics sit at the epicenter of life and death decisions. Physicians rely on test results (blood tests, CAT Scans, MRIs, genomic tests, etc.) at each pivot point. When that data is connected to medical records for the patient for whom it was ordered, we have the necessary ingredients to contextualize the diagnostic and personalize it, resulting in more tailored and optimal therapies.
Over the past 7 years, Tempus built the chassis, but we were missing fuel to broadly accelerate the adoption of our platform. Generative AI is like the plutonium in Back to the Future that powers the Delorian time machine, giving us the “88 gigawatts” of power we needed.
I know what this means for us and our mission to help patients live longer and healthier lives. I can only imagine what it means for healthcare and other industries more broadly. It’s going to be a profoundly wild ride, so buckle up.
As Dr. Emmett Brown said to Marty Mcfly “Roads? Where we’re going, we don’t need roads.”