Database Reference
In-Depth Information
industry is actually a bit ahead in this regard. An executive from Target—
the proverbial or perhaps apocryphal Target predictive application—doesn't
necessarily care about the underlying causal process giving rise to someone
buying diapers versus beer. What the executive cares about is: “Does this
model have predictive power and does it let me then go do something else?”
I think you're going to see much more of that trend in the life sciences.
Gutierrez:
For someone who wants to start working in this area, what
material should they be consuming?
Jonas:
On the computational bio side, there are lots of blogs out there
actually. For instance,
Nature
and
Science
both run blogs. However, one of
the most useful resources, which I didn't appreciate as an undergraduate, are
review articles. A review article is an up-to-date survey of some particu-
lar subfield. This is something no one tells you about when you're 21 and
struggling through some material. It would be so much easier if someone said,
“Guess what? Some poor graduate student out there has written a 15-page
article on the state of the art in extreme but accessible detail because it's
designed for the wider scientifically literate audience. You should go and read
it to understand the material.” Both
Nature Reviews Neuroscience
and
Nature
Reviews Genetics
are both great sources for having these sort reviews. Those
are my two go-to resources.
The other thing that people really don't appreciate is that graduate students,
perhaps because of an adherence to sunk cost fallacy, often write really great
surveys of the field at the beginning of their PhD thesis. Often, when I want to
get into a new field or an adjacent field, I go find a recent grad student's thesis
and then read the first chapter. This is because they're going to talk about the
review of their field in a way that keeps in mind a general audience for that
section. So they're often great pedagogical tools. I have lots of printed-out
PhD theses around my living room from random graduate students.
Gutierrez:
Let's switch back to your startup experience with Prior
Knowledge. What was it like to be a guest lecturer at Peter Thiel's startup
class at Stanford?
Jonas:
The whole experience was great and the Stanford students were very
enthusiastic. To the class I guest lectured specifically, the experience was very
fun, as the class was very much focused on AI and not data analytics. What
made it a very interesting experience was that I was a guest lecturer with Scott
Brown, the CEO of Vicarious, and Bob McGrew, the director of engineering of
Palantir. These two companies are pretty much on opposite sides of the space
of possible data/AI companies.
Palantir is a very successful company that does nothing AI-related in the
slightest. In fact, they tend to be very AI-agnostic. They build tools to let
human analysts do better data analysis. On the other hand, you have Vicarious,
