Facts
Stated by the speakers. Each quote matches the transcript, which shows what was said, not that it is true.
Daphne Koller defines artificial intelligence as recreating human intelligence and machine learning as methods that learn complex tasks from data.
“I think artificial intelligence is the task of recreating intelligence as we know it in people, whereas machine learning is methodologies that learn to perform complex tasks by learning from data.”
0:02:50 · Daphne Koller · AI definitions
Daphne Koller says she finished high school at 16 and college at 17 in Israel.
“I finished high school at 16 and college at 17 and had always been interested in disentangling complexity.”
0:04:18 · Daphne Koller · Early life
Daphne Koller's father was a biology professor, and she avoided biology as a child partly for that reason.
“My father was a professor, actually a professor of biology as it happens, which is one of the reasons why I steered as far away from biology as I could as a child”
0:05:19 · Daphne Koller · Family background
Daphne Koller served as a military intelligence analyst in Israel, where she first encountered Bayesian reasoning.
“I was a military intelligence analyst. So it was actually kind of fun because I got to, it was my first encounter with Bayesian reasoning.”
0:08:55 · Daphne Koller · Military service · 1980s
Daphne Koller says artificial intelligence has cycled through boom and bust periods because of overblown promises, and she began her PhD at the end of a mid-1980s boom.
“AI has gone through peaks and troughs multiple times in its history, generally caused by the fact that certain groups of people made hyperbolic promises about what artificial intelligence would be able to accomplish”
0:14:53 · Daphne Koller · AI winters · 1980s
Daphne Koller says machine learning took off as a discipline within computer science and engineering in the early 1990s.
“But no, it really sort of took off as a discipline within computer science and engineering in the early 90s.”
0:17:08 · Daphne Koller · Machine learning history · 1990s
At Berkeley, Daphne Koller's postdoc mentor asked what three smart undergraduates should build from her PhD work, a question she says changed her approach toward useful applications.
“if I gave you a group of like three smart undergrads to work with you and build a system that would be useful based on the work that you did, what would you have them do?”
0:18:49 · Daphne Koller · Career turning point · 1990s
Daphne Koller co-developed influential work using gene expression data to uncover gene regulatory networks.
“how do you use gene expression data to uncover gene regulatory networks. And that was, I think, work that was quite influential. The paper is still very highly cited.”
0:24:05 · Daphne Koller · Early biology work · 2000s
Daphne Koller spent a sabbatical in Jonathan Weissman's biology lab at UCSF working on yeast gene knockouts.
“my sabbatical in a biology lab at UCSF with Jonathan Weissman, who's one of the leaders in the functional genomics CRISPR field, incredible scientist.”
0:25:48 · Daphne Koller · Sabbatical
Daphne Koller worked on some of the earliest machine learning analyses of cancer biopsy images, but struggled to get it adopted in hospitals from within academia.
“had worked on some of the earliest machine learning-based analyses of cancer biopsy images under the microscope.”
0:28:02 · Daphne Koller · Pathology machine learning
Daphne Koller's Coursera journey began with the first Stanford MOOCs in fall 2011 and lasted almost exactly five years.
“from the very beginning of the journey, which was in the fall of 2011, which is when we launched the first MOOCs while still at Stanford, up until the time that I left, it was almost exactly five years.”
0:32:18 · Daphne Koller · Coursera · 2011
Daphne Koller resigned a chaired, tenured Stanford faculty position to stay with Coursera after her leave of absence ended.
“So I decided and I ended up resigning what was a chaired, tenured faculty position at Stanford, because I felt like I really needed to see that Coursera journey through.”
0:31:24 · Daphne Koller · Leaving Stanford · 2014
Daphne Koller sought advice from Art Levinson in 2016, and he instead pitched her on joining Calico.
“it was a very memorable meeting because he said, look, I can answer your question, but let me tell you about Calico.”
0:34:24 · Daphne Koller · Calico · 2016
Daphne Koller left Calico after about 18 months because she wanted to build a broad drug discovery platform, which did not fit Calico's focus on aging biology.
“I wanted to build a platform and it didn't seem to make sense to build it there”
0:35:14 · Daphne Koller · Leaving Calico · 2018
Daphne Koller's conversations with the investors who founded insitro began in early 2018 as she was transitioning out of Calico.
“they really started in early 18 as I was kind of starting to transition out of Calico.”
0:38:47 · Daphne Koller · insitro founding · 2018
Jim Tannenbaum of Foresight Capital was insitro's first investor, followed soon by Krishna Yeshwant of GV and Bob Nelson of Arch Venture Partners.
“So the first investor was Jim Tannenbaum. And then very soon thereafter, Krishna Yeshwant from GV and Bob Nelson from Arch.”
0:37:53 · Daphne Koller · insitro investors · 2018
insitro's initial investor group had five major investors, including Third Rock Ventures.
“So we ended up with five major investors, actually, in that initial group, with Third Rock being the fifth.”
0:38:07 · Daphne Koller · insitro investors · 2018
insitro raised a $143 million Series B financing in May 2020, led by Andreessen Horowitz.
“as it raised a $143 million Series B financing in May 2020, led by Andreessen Horowitz, and which included earlier investors like ArchVenture Partners, Foresight Capital, GV, and Third Rock Ventures.”
0:00:41 · Luke Timmerman · Funding · 2020
Daphne Koller says an excellent data set with an average algorithm beats the best algorithm on a poor data set.
“if you ended up giving me the tradeoff between the best machine learning algorithm on a crappy data set versus an amazing data set with kind of like an okay machine learning algorithm, the second will win every day.”
0:39:15 · Daphne Koller · Data quality
Daphne Koller says data scientists in most big pharmaceutical companies sit in small siloed groups, one reason she chose not to build insitro inside one.
“the data scientists in most of these companies are usually this kind of smallish siloed group that is sitting in some corner somewhere.”
0:42:32 · Daphne Koller · Pharma culture
Daphne Koller says many insitro hires are roughly 70 percent life scientists who can program in Python, an approach she calls recruiting 'damp' scientists.
“They are 70 % life scientists, but they know how to program in Python and analyze their own data.”
0:47:12 · Daphne Koller · Hiring
insitro makes induced pluripotent stem cells from donors' cells and differentiates them into cell types such as neurons or liver cells for experiments.
“where we can take cells from you or me, convert them into pluripotent status. stem cells and then differentiate them into a Daphne neuron or a Daphne hepatocyte or any one of the set of lineages that the community has developed differentiation protocols for.”
0:52:38 · Daphne Koller · Stem cell platform
insitro's liver disease work came out of its partnership with Gilead Sciences on NASH, non-alcoholic steatohepatitis.
“It came out of the partnership that we have with Gilead on NASH, non-alcoholic steatohepatitis.”
0:55:25 · Daphne Koller · Gilead partnership
Daphne Koller says many central nervous system diseases are highly heritable, citing autism at up to roughly 70 percent.
“the heritability is you know 30 40 50 70 percent for instance in autism so it's very high but also very complex”
0:56:03 · Daphne Koller · Neuro genetics
Daphne Koller notes that BRCA1 and HER2 breast cancers are treated with completely different treatments, as an example of precision oncology.
“a BRCA1 breast cancer is very different from a HER2 breast cancer and are in fact treated with completely different treatments right now.”
1:00:24 · Daphne Koller · Precision oncology
Daphne Koller cites PCSK9 as a target discovered through rare familial cholesterol variants whose drugs also help patients without those variants.
“A great example of that is the poster child here, which is PCSK9, where it was discovered in a rare familial variant in hypercholesterolemia on the one hand, familial hypercholesterolemia, familial hypocholesterolemia on the other, sort of showing this beautiful dose response.”
1:01:42 · Daphne Koller · PCSK9
Daphne Koller predicts that disease classification will change substantially over the next 10 to 15 years, especially in neuroscience.
“I think over the next 10 to 15 years, we'll actually end up with a very different disease taxonomy than we have today, especially in the neuroscience area.”
1:02:25 · Daphne Koller · Disease taxonomy