Michael I. Jordan: evaluation

Transcript context

…And you did some analysis on those 200,000,000 predicted proteins and and and you found they were very good, but there was something missing, but you could robustify them. You could robustify them. That's correct. And I think that's a good example. So I'm a big admirer of AlphaFold. I don't think it's like an LLM. Think it's targeted. It was for a particular set of problems and it does it very well. The issue that we found empirically was that when you ask certain kinds of questions, in particular we did 1 where we were looking whether quantum fluctuations in a protein were associated with phosphorylation, meaning the protein was active or not in the cell. And you might think that these fluctuations which lead to strands hanging off are kind of like bad proteins, evolution wouldn't use them. But it turned out that a lot of them seemed to be phosphorylated, meaning they're reactive in the cell. That suggests a hypothesis test. Is there an association between yesno phosphorylated and yesno quantum fluctuation? So that's a little 2 by 2 table and you do a statistical test on that. And the problem is that if you just use known protein data, there's a crystal structure known, you don't have enough data to test that hypothesis with high power. And so you can't reject the null hypothesis, there's no association even though there looks like there is. If on the other hand, use 200,000,000 proteins out of alpha fold, you can test hypothesis with high power and you reject the null hypothesis. But what we found is that the confidence interval on that statistic of that 2 by 2 table was extremely narrow and way far from the truth, the true value of the gold standard value. And we found this in domain after domain. So why is that? Well, what's happening there is that there's probably not many examples in the training set of proteins with quantum fluctuation because it's not been that studied in the past and it's hard to crystallize. And so not many examples means that it's quite possible AlphaFold won't give out a great answer, but it won't tell you that. It doesn't give you out error bars and it doesn't specifically on the question you're asking. That's where I want the error bars. And it didn't know about that question when it was built and designed. So now I have a good statistical question. What if I add a little bit of ground truth data to the 200,000,000? Can I shift the error bar so it stays somewhat narrow, so I have high power, but it covers the truth? And the answer is, yeah, there's a methodology. We developed something called prediction powered inference that does exactly that. And so it'll cover the truth just like in a classical statistical setting, but it's using this rather highly biased architecture. And it's now it's not biased overall. In fact, its accuracy is high overall. But for the question I'm asking, it might be very biased. And that's gonna happen a lot in science because scientists are rarely interested in just studying the past over again. They're interested in brand new things on the edge of knowledge. And that's where specifically these foundation models will be most poor and most highly biased. d in just studying the past over again. They're interested in brand new things on the edge of knowledge. And that's where specifically these foundation models will be most poor and most highly biased. So there needs to be around any foundation model the ability to maybe collect a bit of ground truth data to merge it in with some procedure like this and then to give out a more trustable answer. That's all not science fiction. That's what can be done and what really needs to be done. And I'm sure the AlphaFold people are on board with that, that they would not find that weird or surprising. But a lot of other people out there talk about bias and all that and they either don't worry about it, I say it'll go away. We have enough data. Or they just critique the architectures and critique the outputs, but they have no scientific method in mind that'll help us go forward. So that's kind of the state we're in.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.