Priscilla Chan and Mark Zuckerberg: Frontier AI + Virtual Biology To Solve All Diseases

One of the core principles for us at CZI was simply getting started. We received advice that philanthropy and scientific endeavors, like any other discipline, require practice; proficiency isn't achieved overnight. Therefore, we decided to dive in, experiment with different approaches, and identify areas we enjoyed and where we could make an impact. As you mentioned, November marks the 10-year anniversary of CZI's inception. We are incredibly proud of the work we've accomplished, particularly in education and community support. However, upon reflection, we believe our contributions to science have yielded the greatest impact and are significantly accelerating progress.

Basic science is a distinct part of the innovation funnel. Our view is that the federal government, primarily through NIH, significantly outweighs other investors. However, their investment pattern enables individual investigators to conduct research. Historically, major scientific advances are often preceded by new tools or observation methods. For instance, the telescope revolutionized astronomy, and the microscope transformed our understanding of biology. We are now at a point where many new tools are emerging, including computational tools and methods to instrument and understand the body in novel ways.

We are accelerating scientists' work towards this goal. Interestingly, our initial timeframe is by the end of the century. When you ask biologists, they often question if this ambitious goal is achievable. However, when you ask AI experts, they suggest it should be easy, wondering why we are so unambitious. Given the rapid pace of AI improvement, I believe it might be possible significantly sooner. While I don't think it's worth setting a specific date, the first decade focused on projects like the cell atlas. This work aimed to understand the specifics and data regarding all the different configurations of every cell in the body.

It's still quite early, but the first step, which is often overlooked, is simply bringing people together, as Priscilla mentioned. This is worth discussing because while it seems obvious, it's somewhat novel in science. Traditionally, funding goes to individual teams with relatively small grants, leading to independent research. However, it's amazing how much progress can be made when people from different disciplines collaborate. Throughout my career, both at Meta and here, I've seen teams not working together or disagreeing. Physically bringing them together has proven incredibly helpful.

With BioHub, we aim to explore what happens when frontier biology and frontier AI are developed in sync. This involves designing tools on the frontier biology side to specifically collect and learn data types. This data then feeds into models built to understand cells and the body at various resolutions. This integrated approach allows us to design necessary tools, ultimately leading to more grounded insights. It moves beyond simply letting AI experts work with whatever biological data is available.

Our genetics are incredibly diverse. Each of us is unique, with both known and unknown indicators of disease. I find variants of unknown significance to be the most interesting and frustrating. Imagine a loved one with a diagnostic mystery. Genetic analysis might reveal unusual findings, but their significance is unclear. This leaves you wondering whether to panic or what steps to take. What we really need, and what I believe these models will provide, is the ability to examine those variants and model their impact on different cells. This includes understanding how they influence cellular behavior and whether they are linked to a disease pathway.

As you mentioned, there are too many dimensions for us to hold in our brains. That's why we're so excited about this intersection at this moment. It's now possible to consider many more dimensions, matching the complexity of biology.

The virtual immune system is a subset of a more generalized model we'll eventually achieve. It's fascinating for several reasons. Firstly, it involves individual cells interacting, many of which we don't fully understand, such as B cells, T cells, and NK cells. Our current technologies allow us to study these cells at a more granular level, which is exciting from a biological perspective. The clinical impact of understanding the immune system is immense because biology has already provided mechanisms to maintain bodily health. However, it can also malfunction, leading to diseases like autoimmune conditions. It's a complex system that must remain in balance; if it deviates in either direction, illness results.