Science Friday - Florence Flooding, Algorithms, Dino Demise. Sept. 14, 2018, Part 1

Episode Date: September 14, 2018

Last month, California passed a bill ending the use of cash bail. Instead of waiting in jail or putting down a cash deposit to await trial at home, defendants are released after the pleadings. The catch? Not everyone gets this treatment. It’s not a judge who determines who should and shouldn’t be released; it’s an algorithm. Algorithms have also been used to figure out which incarcerated individuals should be released on parole. Mathematician Hannah Fry and computer scientist Suresh Venkatasubramanian join Ira to discuss how algorithms are being used not only in the justice system, but in healthcare and data mining too.  As Hurricane Florence approaches the Carolinas this week, forecasters and disaster management officials are stressing one key piece of advice to evacuating residents: Take the storm seriously, regardless of the category designation. Once projected to hit Category 4, Florence was at Category 2 as of Thursday morning, but that number only describes the wind speed. Meanwhile, as University of California-Irvine civil engineer Amir AghaKouchak notes, there could be unusually devastating flooding, as storm surge from the ocean meets rainfall from a storm that is projected to pour on the region for days. “Compound flooding” is the phenomenon that left Houston under water after Hurricane Harvey in 2017, and, at its worse, could cause rivers to run in reverse. And, AghaKouchak says, climate change and sea level rise both make such flooding more likely in storms like Florence. The prevailing theory says a meteorite led to the demise of the dinos. But Gerta Keller, a longtime geologist and paleontologist, isn’t buying it, and says volcanoes were the real culprit. The latest episode of Undiscovered tells her story, and asks whether conflict among scientists really makes science stronger. Co-hosts Elah Feder and Annie Minoff join Ira for a preview. Subscribe to Undiscovered wherever you get your podcasts.   Subscribe to this podcast. Plus, to stay updated on all things science, sign up for Science Friday's newsletters.

This is Science Friday. I'm Ira Flato, broadcasting from the studios of K-U-E-R-N-P-R-U-T-R-U-T-R-U-T-R-U-T-R-Lat in Salt Lake City today. Later in the hour, a look at how algorithms are creeping into our everyday lives. Have you had a personal run-in with an algorithm in your day-to-day life? Share it with us. Give us a call. Our number is 844-724-8-255-8-4-Sai-T-Talk, or you can tweet us at SciFri. But first, the EPA is considering. changing its rules for how certain emissions of the greenhouse gas methane are regulated. It's a change favored by oil and gas producers, but not favored by climate change scientists. Here to talk about that and other selected short subjects in science is Amy Nordrum.

News editor at the ICCLEE spectrum in New York. Welcome back, Amy. Hi, Ira. Thank you. You're welcome. So what's going on with these new methane emission rules? Well, the EPA requires oil and gas companies to regularly inspect all of their equipment for methane leaks.

And they also require these companies to report leaks that they do find. Now, with these new rules, the EPA is giving companies more time to report a leak and repair it once they find it, and also more time between mandatory inspections that they have to conduct on their equipment. So basically, they're doubling the amount of time.

So if they had to inspect their wells twice a year for methane leaks, now they'll have to just do it once a year. And if they were previously required to repair a methane leak within 30 days, now they have about 60 days to do that. So a lot of environmentalists are not quite happy with this. Right, yeah, they're saying that this change in regulations could lead to more leaks that are found less frequently, so more methane leaking out into the atmosphere. And methane is a very potent greenhouse gas.

It's 25 times more effective at trapping heat in the atmosphere than CO2. But, of course, the industry is very happy about this regulation. They said it was a lot of red tape that the EPA is getting rid of. And the EPA is estimating that this change would save the industry about $75 million a year. Wow. That comes on the heels of other regulatory shifts involving greenhouse gas emissions and climate change. Yeah, the EPA is doing a bit of a trifecta this year, so there were two previously changes that was also proposing to make. So one has to do with freezing the vehicle emission standards that the Obama administration had put into place for tailpipe emissions.

And the other one has to do with allowing states to monitor the emissions from coal-fired power plants that Obama had under the clean power plan. And other types of emissions, there's news about e-cigarettes. What's going on there? Yes, so at their best, you know, e-cigarettes are a product that could help smokers transition to a less harmful way to consume nicotine and maybe ultimately quit. But at their worst, authorities fear that they can also hook people that have never touched a cigarette onto a new habit and onto nicotine for life. So in this change, the FDA is worried about the latter, especially as it applies to minors,

so those who are under 18 years of age. So this week, the FDA sent out more than 1,000 warning letters to retailers that they had caught selling e-cigarettes to minors. And the FDA also told five manufacturers that they needed to submit plans within 60 days for how to keep their products out of the hands of minors. So the FDA is really taking this possibility that teens and minors would get their hands on these products very seriously

and starting to crack down on it. There are lots of teens using these devices, right? I was really surprised. You know, the FDA said that last year, two million high schoolers and middle schoolers were regular users of e-cigarette products, and they called them numbers an epidemic proportion. Wow. Let's move on.

You have a story out this week about ancient cave art. Right. This week reported in nature. Archaeologists said they'd found the world's oldest line drawing. So this was discovered in a cave in South Africa that has been excavated continuously for a couple decades. It's a series of nine red lines that were drawn with a sort of clay. crayon on a piece of stone that they believe was used to grind different materials back in the middle stone age.

This dates back to about 73,000 years ago. Wow. So does it look like a hashtag? And people have been calling it a hash, the world's oldest hashtag. Yeah, it's like that cross-hatch design. And, you know, curiously, this design, this kind of symbol has been found repeatedly over the course of human history. So it's not just, you know, Twitter these days.

There may have been some early significance of a very similar. looking symbol and it really does kind of look like a red hashtag on the stone. Or they were playing a tic-tac-toe or something like that. Yeah, maybe. Maybe kind of a game. Another story about language and thought bilingualism. Yes, NYU researchers have been studying what happens in a bilingual person's brain when they're forced to switch from one language to the next.

And they're looking at particular at how much cognitive effort it takes to stop speaking one language and start speaking in a new one. So in a study of people who speak both English and also American Sign Language, they found that there's actually more activity and work involved in the brain at stopping a language. So you can think of it as like putting down that vocabulary and stopping access to those words than to start access in a new one. So it's actually easier to pick up a new language in these bilingual brains than to put down the one that you're currently using. Wow, wow. And did they look into like sign language or things like that? Yeah. And they also looked at whether it was easier for, especially in the case of people who can both sign and speak English at the same time, they studied whether it was more difficult to do both of those things than just one of them.

And they found for native English speakers, it was easier for them to both sign and speak English at once than to just sign. So there's also a mechanism involved in actually suppressing your native language and turning that part of your brain off in a sense. And that's more difficult than just speaking and signing in two different languages at one time. Very interesting. And finally, there's the giant pool noodle. Yes. It's being called. Absolutely.

And it looks just like one. Yeah, it's floating out there in the Pacific right now. So this is basically kind of a giant pool noodle built for the ocean and specifically for the, great Pacific garbage patch. You may have heard of this. It is a gyrating mass of mostly plastic material that's out there in the Pacific. One point eight trillion pieces floating around out there. And last weekend a non-profit launched sort of a novel effort to try to clean up part of this patch. So they built a giant pool noodle. It's

600 meters long. It's floating in a U-shape out there in the Pacific. They're starting a two-week trial. And then they'll move it out to the great garbage patch and try to clean some of that up if all goes well in the two-week trial that they've just launched. But not everybody is convinced this is a good thing. How could you be against a giant pool noodle? Yeah, it's a pretty clever idea, but there are critics of this method. So first of all, the technology itself is not proven.

And the nonprofit readily admits that. That's what part of this trial is all about, figuring out if it'll work as intended, and it may not. But there are others who say that this is sort of a distraction from the real problem, which is preventing the pollution and the plastic from getting out there in the first place. And it'd be much easier to just, you know, work on that rather than going out and to clean it all up. Always great stuff, Amy. Thanks for taking time to be with us today. Yep, thanks, Ira. Amy Nordrum News editor at the ICCLEE Spectrum in New York.

Now it's time to check in on the state of science. This is KERNNO. St. Louis Public Radio News. Local science stories of national significance. And for parents around the country, back to school often means a trip to the doctor, sometimes a vaccine required by the school. Parents can seek an exemption from the required vaccines, either for medical or non-medical reasons. And nationwide, around 2% of students get some type of exemption. Today we're broadcasting from KUER studios in Salt Lake City and joining me now to talk about a change in how to state of Utah manages such exemptions.

is Eric Newman, a reporter in KUER's Mountain West News Bureau. Welcome, Science Friday. So what was the exemption policy before, and what is it now? The policy, the law that people can exempt their kids for a philosophical or personal belief is basically the same as what it was before. But where previously parents would have to say they wanted to get their kids exempt from getting vaccines, go to the local county health department, get a note from a nurse, they may have to pay a fee. The information that parents might be getting around the state was kind of variable. And so the health department here decided that they should streamline all that and make it consistent.

So they created this new education module that parents have to watch over the Internet, and it kind of walks them through the risks of exempting their kids from getting vaccines and talks about what they might have to do to keep their kids at home for, you know, a number of weeks if there's an outbreak and things like that. So did they have to sign a document that says, I have watched the training module? My understanding is that it's basically a click on the computer. I did the module myself and didn't do the final signing at the end, but it streamlines the process. And the hope is that it makes the information better. But some people are worried that it's going to make it that much easier for parents that are that much more convenient for parents that don't want to get their kids.

You don't need the doctor's note anymore. You just watch the video and say, I don't want to do it for whatever reason. Yeah. So give us an idea of what Utah's vaccination numbers are like. Is there a particular problem in this state compared to the rest of the unit? From the immunization folks that I've talked to at the Department of Health here, they've described Utah as being relatively, in the middle. It's not that our rates are extremely low for kids getting vaccinated. But there's certain pockets of the state that sort of are, I would say, representative of Utah's libertarian

political perspectives, where there are a lot of parents who don't vaccinate their kids. And so some research that's come out recently has identified certain counties in Utah that have such high rates of exemptions that they're saying this could be at, risk of vaccine preventable disease outbreaks if, you know, measles was introduced here. And you have a lot of kids in Utah. We have one of the highest birth rates in the country. And so it kind of set us up for potential risks. So they're betting that by watching the video, more education will be better for the vaccination program.

But do we know if there's any science behind being more educated makes you want to get a vaccination? You know, it's really hard to tell. I talked to one family who didn't, who was not vaccinating their kids. And they were very informed about information about this. But they might be overstating the risks of their kid having an allergic reaction or something like that. So it sort of seems like a lot of the people that are already on this train of not wanting to have their kids vaccinated are pretty entrenched. in their perspective. And so I think it's going to take some time to see what affect this new module has.

Yeah, you know, that reflects the national trend. We see that in other places. Right. Thank you. Eric. Eric Newman is a reporter in KUER's Mountain West News Bureau, and he was joining us today. We're going to take a break on when we come back. Hurricane Florence has arrived to the Carolinas, and it's making its mark with a huge storm,

surge, heavy wind, and maybe the most important part of it is the rain, how climate change may be contributing to this double whammy of flooding. You got the surge in front of you. You got the rain behind you. What's that doing to the land and the farming and acreage and all kinds of stuff? We'll talk about it after the break. Stay with us. After long anticipation, Hurricane Florence hit the Carolina Coast and downgraded from

a category one to a category four to a category of one storm. It nonetheless comes with predictions of 90 mile per hour winds, potentially as many as 40 inches of rain in some places and a storm surge that has already reached 10 feet above normal sea levels in some places, which is why more than a million people were urged to evacuate before the storm. My next guest is here to talk about all that water. What happens when high seas meet unrelenting rain? Well, it's a unique kind of flooding, he says, that we can expect to see more of as the globe warms and sea levels rise. Amaragakouchek is Associate Professor of Civil Engineering University of California at Irvine, and he joins us by Skype.

Welcome to Science Friday. Good to be with you. Nice to have you. Let's just looking at Hurricane Florence. We're hearing already about having more than a foot of rain, having fallen, the storm surge, topping 10 feet in some places. What are the flooding dynamics at play right now? Right. Hurricane Florence is a relatively slow-moving hurricane.

Just to give you an idea, the average forward speed. of ranges from 10 to 35 miles per hour. But the average speed of Florence has been less than 10 miles per hour, I think around 5 miles or so. Miles per hour or so. So this means that there is higher chance of substantial rain over one specific city or basin. So this is one interesting and unique aspect of this hurricane. You know, we're hearing news also about really bad flooding in towns way inland. Like there's a town of New Bern in North Carolina that's not directly on the coast.

I think it's 20 miles upstream, so to speak. And it's at the junction of two rivers, and the town's getting squeezed in two directions. Right. So obviously, with hurricanes come surge and extreme rain, and both are drivers of coastal flooding. when multiple drivers like surge and rainfall in Iraq can cause flooding, we call it compound coastal flooding. And these events can go way inland, of course, depending on physiographic features.

So why is this important? The amount of water that the river can discharge into the ocean is a function of the gradient between river mouth and the ocean water level. So the higher the ocean water level, the less water can drain into the ocean. And in fact, the surge and high waters can cause the river to go backwards, backward inland, and cause even more significant impacts. So this is a kind of compound coastal flooding event.

We talked about how slowly this storm is moving. Is this partially true because of climate change, creating stronger storms, wetter storms, other ways that changing climate could alter the floods we see in hurricanes? Right. So in the past couple of decades, rainfall associated with tropical cyclones and hurricanes have increased. And many studies have linked the observed increase in rainfall to the increasing sea surface temperatures due to anthropogenic climate change. Sea surface temperatures can impact the size and intensity of storms because hurricanes

and tropical cyclones gain their heat, moisture, and also energy from the surrounding sea surface temperatures. So our satellite observations and direct measurements already show that sea surface temperatures have been warming in most oceans around the world. Can we expect this kind of compound flooding to be commonplace in future storms or even during regular high tides as global climate change kicks in? Absolutely. The problem we have is that sea levels have been rising and they are projected to rise

in the future. means the baseline ocean water will be higher in the future. So it means in future hurricanes, and the surge will be added to a higher baseline. And obviously, higher water level can cause more impacts, more coastal flooding impacts. What kind of infrastructure then do we need to cope with this? Well, this question is very local. So in some areas, you know, are projected to rise more in some places less. So there is no single solution for everywhere. This requires a lot of investigations and studies modeling,

understanding our exposure and how future is going to change. But the kind of strategies include relocation to building seawalls and levees and things like that. Thank you very much for taking time to be here with us today, Amir. Absolutely. Amir Agakukukukes is a very much. Kuchek is Associate Professor of Civil Engineering at the University of California, Irvine. Last month, California became the first state to end cash bail. Instead of waiting in jail or putting down a cash deposit to go home or wait for a trial,

defendants are now being just released after the pleadings. There is a catch, though. Not every defendant gets this treatment. The judge who decides who should and should not be released, He bases his or her decision on the advice of an algorithm. Algorithms have also passed judgment on which inmates get released on parole. Algorithms are used in the car industry, the medical world, and, of course, in determining your social media feed. Have algorithms made a decision for you or someone you know?

We want to hear about it. Give us a call 844-724-8255. That's 844-Sci Talk. You can also tweet us at SciFRI. Here to talk about how algorithms have found their way into our everyday lives is Hannah Fry. She's a mathematician at the University of College London, author of the new book, Hello World, being human in the age of algorithms. We have an excerpt at Science Friday.com slash helloworld. She joins us from the BBC. Welcome to Science Friday.

Hello. Thank you very much for having me. You're welcome. And now algorithms have been around for a long time, so why write about them now? Well, I think that things have changed in the last few years. You're absolutely right. Even the examples of algorithms being used to decide whether or not someone should get bail. I mean, this has a very long history. It dates back to the 1920s, 1930s, the very simplest kinds of algorithms. But I do think that something has changed in the last five or ten years anyway.

I think what's changed is the amount of data that is collected on us and how that data is analysed and then used to predict our behaviour. year. But I also think that, you know, with the advent of artificial intelligence, the amount of power that algorithms are being given and the number of situations in which they're being deployed, really, to make decisions about our lives is only increasing. And I thought that was something that was really sort of, well, quite timely, really. I think it was important to really put that all in one place. You write in your book that the only way to objectively judge whether an algorithm is trustworthy is by getting to the bottom of how it works.

So you write the air a lot like magical illusions. It's true. It's true. I think on the surface a lot of this stuff, especially artificial intelligence, it looks like it's actual magic. It looks like it's wizardry. But very often when you dig behind the surface and look at how the trick is done, there is often something, you know, incredibly simple lying behind the scenes. and often actually, well, at least occasionally, there's things that are quite worryingly reckless there too. Yeah, well, and you say that's really because your book is about humans, right? They're the people who write the algorithms.

Yeah, I don't think that you can really separate the two. I don't think that you can look at algorithms in isolation. I think that you have to accept that when they're out there in the world, they're being used by people about people. and all of us have these really inherent flaws in us. You know, we have all kinds of subconscious biases. You know, we have issues where we overtrust what a machine tells us. And then at the other end of the spectrum, we're very good at dismissing any machine

that makes any kind of mistake whatsoever and thinking that we know better. And, you know, that's happening within the people who are creating these algorithms too. And I think that we have to kind of think of this as humans and machines together, not just how good is the artificial intelligence? on its own. So it's not a question of when you trust a machine over your own judgment or not.

Well, you know, I think that it's different in different cases, really. I think that, you know, there are some situations in which you all you want is the best prediction that you possibly can. You just want the most accurate prediction. And in those cases, I think, you know, if an algorithm can prove, and it's quite a big if there, if an algorithm can prove that it can make a better prediction than a human can, then I think that that's sort of the situation in which you want to hand over some level of control. An example of that might be in some of the cancer diagnosis algorithms that are existing now that are screening biopsy slides and looking for tumours. Now they have their problems and I think that you have to work carefully in the way that you design them to work around those.

But if it can, if those algorithms are more sensitive than a human pathologist looking at hundreds of these slides every single day, if the algorithm can pick up on really, really tiny clues hiding amongst your cells as to what your future holds in store for you, then I think in that situation, actually, you should give up some control and trust the algorithm perhaps over just a human on their own. But I think there are other situations where, you know, particularly in the criminal justice system, where I think we have to be really, really careful

and think very hard, very long and hard about how much control we hand over and the ways that we do that. Let me expand on that. I want to bring in another guest, Sir Reg van Kota Subramanian, who is a professor in the school of computing at the University of Utah here in Salt Lake. He's a board member of the ACLU in Utah, and he's here at KUER in Salt Lake. Welcome back. Thanks for having me. You've looked into these issues about parole algorithms, right?

Are they good? What's the plus and minuses about them? Well, first of all, I just want to commend Hannah. the book is awesome. And I think the kind of nuance that you bring to it and that you just described is exactly more of what we need in this discussions rather than the kind of binaries

you've been talking about. So thank you for the book. So I think, as Hannah points out, right, some of the challenges, and I think California's discussion sort of brings us up into sharp relief is that there are often in these situations, laudable goals,

the idea of reducing pretrial incarceration, the idea of eliminating money, the idea of just not punishing people because they're poor. These are laudable goals, and I think it's worthwhile to see whether machine learning AI can help us achieve some of these goals. The problem is these issues are remarkably, impressively subtle, as Hannah points out, right, you know, just because an algorithm makes a recommendation, it does not mean a judge is required to or will take the recommendation. And it's not clear exactly, as you mentioned, how the algorithms do make their predictions. and it's not clear whether the data being fed in to train the algorithms,

and this is something we don't often talk about, has the right signals in it to capture exactly what you're trying to capture. There are lots of, so for example, shifting slightly away from pretrial going to parole, one of the goals of parole sort of modeling is to understand whether someone will reoffend after being released, and that would be considered a bad thing. But if you measure, for example, how people are re-arrested, that's a subtly different thing you're predicting. You're not predicting whether they will be.

recommit a crime, you're predicting where there'll be re-arrested. And if that data's being used to model reoffense rates, then you get a completely different system to what you expected to get. And so there are many, many subtleties that require extensive domain knowledge. And so just taking a black box algorithm and putting it in, is really not going to help you. And, Annie, you write a lot about that in your book also. Yeah, I do. I mean, I think that to sort of add to that point there, I think that within these systems, if you are using the whole history of all the arrests that have happened in the past and using that to kind of project forwards into the future, then inevitably within all of that

data, you are going to be, you know, encoding into your algorithm, well, centuries of bias and unfairness, really. I mean, the analogy that I, that I like to give is that, you know, if you do a Google image search for, you know, maths professor or math professor, as you might expect, an awful lot of the top 20 images are going to be white men. And actually, you know, the statistics of what's reflected back to us are pretty accurate. They do reflect what happens in university around the world. You know, the vast majority of maths professors are indeed white men.

But I think that there's a really strong argument that sometimes you don't want technology to be a mirror for society. You don't want it to reflect the kind of, you know, the history that we have that led us to this point. You want it to help us move towards a better society and move, sort of nudge us in the right direction. what that right direction is and how you should go about it. I mean, that's a whole other question. But that's one that exists completely outside of the algorithm itself. I'm Ira Flater. This is Science Friday from WNYC Studios. Talking about algorithms with Sir Ashvinkgo-Sherba Mungian and also with Hannah Fry, author of the new book, Hello World.

Lots of phone calls. A lot of people want to get in on the conversation. I think we're going to go to them now. So let's go to San Antonio. Let's go. Simeon in San Antonio. Hi. Welcome to Science Friday. Hi.

Thanks for taking me. I have a commercial driver's license, and about 10 years ago I left a job on good terms, and I left the vehicle. It had several scratches on it, and I had no idea, but they had reported it as an unreported accident, which is a, you know, it's a, It looks like, you know, a serious accident happened, and I wasn't even aware of it. And years later, when I was looking for a job, I couldn't get anybody to hire me because I had no explanation of why I was unable to get hired. But I looked into it many years after that, and I saw that there was an unreported accident on my record.

And it was just real hard to resolve. So to this day, it's hard to get on with large companies that have algorithm hiring. Let me get a comment from it. Hannah? Yeah, I mean, you know, it's an appalling story, and it's something that happens just depressingly often, that, you know, as soon as your information, your data, and Mark has been against your name,

once it's put into an algorithm into a computer, then suddenly it takes on this air of authority that makes it almost impossible to argue against. And I really think that, you know, we shouldn't necessarily just be thinking about how perfect can we get artificial intelligence, how perfect can we get algorithms to be, we should also be thinking about how can we design them for, you know,

redress, how can we design them to be appealable? Because stuff like this really shouldn't happen. Saraj. I think this brings up another important issue that I think has not been fully appreciated, I think also by the tech community and the larger world. When you put algorithmic decision making, you're putting it into a system. It is not existing in a vacuum.

And so it is not a necessary. enough to merely evaluate how the algorithmic system works. You have to evaluate how it affects the parts around it as well. In this example, you know, you're talking about algorithmic hiring or you're talking about the fact that one erroneous data point made an effect. First of all, we know that a lot of algorithms are very sensitive to small changes in data. So one small mistake can make a huge difference. And when you recognize that they're part of a larger pipeline,

then you think about checks and balances. You think about humans in the loop. You think about a larger system of decision-making, of which algorithms should be one part. And we don't design our systems that way. We sell them as black boxes that can replace humans, and that's really the wrong way to think about this process. We have a lot more to talk about, talking about algorithms.

We welcome your participation. You can also tweet us at SciFRI, S-C-I-F-R-I. We're going to take a break and come back and talk more with Hannah and Suresh. Stay with us. We'll be right back after this break. This is Science Friday. I'm I Refleto. We're talking this hour about how algorithms influence our lives and how we need to be careful to design them

when we sit down to design them with fairness in mind. My guest, Hannah Frye, author of the new book, Hello World. It's a great book. I've read, again, it was just example after example. It's a terrific book, Hannah. And she's Associate Professor of Mathematics of Cities at, no, University College London. It rolls off the tongue.

It does. You're the second professor I've had from that university. I make the same mistake all the time. I wait to try, like, yeah. And also, Sarajvankuro Subramanian, who is a professor in School of Computing at the University of Utah and Salt Lake City, both here. Our number, well, it's so full up, not going to give our number. I'll wait for a spot few to join in. Let me go right to the phones.

Let's see, where we're going to go. Let's go to, let's go. Okay, let's go to Fern. Is it Fern in Alexandria? F-E-N. F-N. My eyes aren't working critically today. Yes.

Thank you so much having me on, Ira. Go ahead. I'm actually a patent examiner for the U.S. Patent Office in artificial intelligence, and we see algorithms all the time and in such a variety of manners. And your guest, we're talking about crime systems. And actually, I'm actually examining a patent that's based on crimes, and it's really interesting. And I just wanted to make that comment that, you know,

there's algorithms that do everything. how important it's to the patent examining process. How, just as a patent, are you still there, Fan? Yeah, I'm still there. You can hear me? Yeah, as an examiner, how school do you need to be? How up do you have to be on, you know, AI technology and design to judge the patent? Sure, so one of the best things about patent examining is that you're learning all of the time. every patent that we examine is new, and the algorithms that they're doing are all new.

But, you know, I actually have a degree in electrical engineering from Drexel University in Philadelphia, and it's very helpful to have a very solid background knowledge, because you're expected to know what these algorithms do so that you can examine the novelty of these patents. I got it. Thank you for taking time to be there. Thanks for that call. So what do you think? Saray, she says algorithms are very important for getting a patent.

She's looking at all of these. I'm not surprised that she's seeing all these patent applications. I mean, I think, to some extent, there is a lot of hype. I think anything, there's a joke, at least in the research community, right? Anytime you take some data and put an Excel spreadsheet, someone's going to market it as AI now. So there's a whole spectrum of things where they really are using sophisticated methods to really are just kind of aggregating things in a box. and that's not AI. You know, that's interesting, because, Hannie, you say in your book that we need something like the Food and Drug Administration, not the FDA,

but a sort of mechanism like it to be able to judge how good an algorithm is. Yeah, I mean, I just find it extraordinary that there is this process that exists for testing the novelty of an algorithm so you can protect the intellectual property, which is, I think, exactly the right thing to do. You need to have that process. But there's no other system that tests whether... the benefit that it offers to society outweighs, you know, the cost to society. It used to be the case that you could just chuck any old coloured liquid in a glass bottle and sell it as medicine and make a fortune from it.

But you're not allowed to do that because it harms people and, you know, it's just not a morally good thing to do. And I think that, you know, we're sort of at this stage where we've been living really in the world west of data and algorithms where people are essentially allowed to use anything that they've created on, you know, members of the public. And I'd really like to see that FDA-style regulation come in where you have a group of experts behind closed doors, protecting intellectual property, but really kind of assessing the benefits

that these algorithms offer to society. Let's move out to something you touched on earlier, and that's algorithms in medicine and in diagnostic medicine. You mentioned how good algorithms are for sorting through data, like slides, picking out possibly cancerous slides versus non-cancerous slides. But so far, even IBM's Watson hasn't been good at sitting down with a patient who walks in and says, my stomach hurts, what is it?

Yeah, it's true. It's a very tough thing to do. Why is that so tough for an algorithm? So, I mean, there are some claims that there are systems that are as good now as human doctors. There's one here in the UK called Babylon, which has kind of been making a lot of news. recently making a lot of headlines recently. That work hasn't yet been peer reviewed, so I'm sort of, you know, holding back the, you know,

celebrating it until I think that process happens. But that's so much harder than just diagnosing or spotting tumours in an image because it's really open-ended. So, you know, if you're training a machine on looking at biopsy slides and finding tumors, you can send it, you know, hundreds of thousands of examples, get it to work through them itself, and tell it when it gets things right. or wrong. But when it comes to diagnosis, I mean, there could be anything wrong with you, right? You

could walk in with any possible number of conditions and describe it in any possible number of different ways. And the knowledge graph that's required to kind of fit all of that information together that's held in the head of just a general practitioner is really, really, really difficult challenge. In fact, I mean, one of the oldest sort of applications or at least proposed application of general AI was in expert systems. And one of the applications was that the expert system could do diagnosis for you. And I remember as a child sort of looking at some basic expert systems written in LISP and seeing how, and their claims to sort of be able to do diagnosis. So there's a long history of AI and medicine. But I think even this case, right? So I think

Hannah's very right in saying that, you know, the more well-defined and very precise the task is, the more likely it is that an automated system could help, like in the tumor diagnosis. But even there, I think, as you mentioned, the book also, even there, there's the issue of, well, does it work equally well for dark-skinned people versus light-skinned people? If you're looking for skin sort of blemishes and trying to figure out that's a sign of melanoma, there are all these issues that come up

even in those kind of settings where it seems like it might be simpler. And I think the larger issue is that a lot of work in AI right now, especially in deep learning, is centered around this idea of how do we represent information. And if only we could find the right representation of our data, then the inference would be easy. And so the hard work is in doing the representation.

The lot of tasks, the representation of information is a very, very complicated thing. It's not as simple as we'll just line up a bunch of numbers in a vector space and do some machine learning on them. It's way more complicated than that. It's kind of interesting. Let's move on. There are so many applications for this.

Let's see how many we can get through. One that was really fascinating, and we have talked about many times on the program, is our algorithms for used in driverless cars. right? Everybody is researching driverless cars. And you point out, Hannah, in your book, and I'll quote it, would people buy a driverless car in which the drivers knew that the car might decide to murder them rather than the pedestrians? It's true. It's true. You know, I think across the board really here, there's a slight difference in our attitudes if, you know, we're the one behind the wheel or, you know, we're the one standing in the dock versus if you're thinking about how the system should be, everyone overall. I mean, you're referencing there the very, you know, famous trolley problem where

a car has to decide who to kill in a certain position. And actually, in the book, I spoke to lots of people who works in with driverless cars. And they tend to kind of roll their eyes actually a little bit when you, when you ask them about this trolley problem, what would it do when presented to do this situation? So in the book, I'm kind of, you know, I try and caveat it heavily based on what they told me. Basically, they say, it will never happen. This is an unlike situation. But then the exact trolley problem happened to my husband about six weeks ago. So I've kind of gone full circle. Now I think that actually we do need to discuss these problems. So, Rachel, shaking your head up and down. Yeah, you're agreeing with it. So the funny thing is

when the German government put out guidelines for the development of technology for driverless cars back in June, I think of last year, they actually have a clause there. Please do not frame this in terms of trolley problems. But I think one thing I always want to give a shout out to is Corey Doctoro's short story called The Car Wars, where he discusses, I think, an issue that I think is very relevant to this in the sense that his argument about driverless cars is not at all about the trolley problem or about the efficacy of the automation. It's about control and governance. Who gets to control the car? What happens if you hack your car? Suppose you put in a new patch that does something that you like.

Are you still going to be allowed to drive your car? I think these issues of governance when you bring in algorithms are something that are not discussed enough. Well, that's what I was leading to. But I wanted to talk about the new European Union's general data protection regulation, the GDPR. How is they, they're aware of this, Siraj, and they're looking into trying, what is this, what is this regulation for? So, so Hannah's in the thick of it back in, back in the EU. I'll get to her after. Yeah.

But I think, I think we don't, the truth is sitting here in the U.S., we don't. really know how this is going to play out yet. I think we're beginning to see signs, for example, that, for example, the right to explanation that seems to, depending on who you ask, come along with the GDPR, the idea that you are given a right to ask the algorithm why it made the decision regarding you. Oh, you are. Depending on who you ask, that may or may not be the interpretation of what the guideline says. As a technology problem, we don't know what that means. What does it mean to provide this explanation? What constitutes a valid explanation? What constitutes a valid explanation? What

constitutes a complete explanation. Is it enough to dump 50,000 pages source code? Probably not. How does this going to play out? It's actually a sort of a fascinating time for researchers in this area because the law has now provided us with an opportunity to sort of think through our research and how we ask these questions, how we solve them. I have got 30 seconds for you to answer that question. Yeah, I mean, GDPR is supposed to put the power slightly back in the hands of the individual. But, you know, at the moment, it seems like you can hide a lot of stuff in terms and conditions. and in Europe we are essentially being drowned by terms and conditions in the last few months. Anna Frye, Associate Professor of Mathematics of Cities in the Center for Advanced Spatial Analysis University College London,

author of the book, Hello World, Being Human in the Age of Algorithms, and you can get a sneak peek of her book at Science Friday.com slash hello world, and Suresh Vramunian is professor in the School of Computing here, University of Utah and Salt Lake City, and a board member of the ACLU. Welcome. Thank you both for taking time to be with us today. Thank you. Next up, our science documentary podcast Undiscovered is back with its second season. Boy, we're glad it is.

And on its next episode, it covers one of my favorite topics, the mystery of what killed the dinosaurs. Well, it used to be a mystery, but, you know, 40 years ago, some scientists came up with an answer, which became the one theory to rule them all, as you can hear in this survey of people outside the American Natural History Museum in New York. Meteorite. Meteorite. An asteroid? A meteorite. A meteorite. A meteorite. A rock. A big rock.

From the sky. A lot of people think this is the case closed, right? The meteorite. An asteroid from the sky killed the dinosaurs. Except one scientist who isn't buying it. The new episode of Undiscovered tells her story and our host, Ella Federer and Annie Minoff are here to talk about it. Welcome back. Hey, thanks for having us.

Thank you. So meteorite theory, I thought that was pretty settled. Yeah, a lot of people did. Oh, this is Ella, by the way. A lot of people, you know, have treated this as scientific gospel since the 80s. And just to spell out what we're talking about, it's the idea that 66 million years ago, big rock from space, slams into the earth, kicks up a lot of dust, which blocks out the light, causes all kinds of environmental changes.

And then very suddenly, a lot of species die out, including the dinosaurs. So that is the concept. But your episode is about a scientist who says that idea that is dead wrong. Right? Correct. And we're talking about Gerta Keller. She's a geologist and paleontologist at Princeton University. And she, for the last 30 years, has been an extremely loud holdout on the so-called impact

hypothesis. Like, as long as this theory has been around almost, she's been opposing it. She says, you know, she's been ostracized by her colleagues, shouted down at conferences. And just speaking for myself, like, I don't know how long I could keep up my opposition under that kind of peer pressure. Yeah, so GERDA is not like most people. She is very tough. Actually, when I spoke to her, she would often refer to people just casually as her enemies.

It's pretty hardcore. She's a particular kind of person. And just to give you a sense of what she's like, we want to play you an excerpt from the episode where Gerta describes a brush with death that she had at 22. So this happens. She's at the hospital, and she has a wound that everyone's. thinks is probably fatal, and there's a priest there trying to read her her last rites. Here's what happened.

I got the last rights, and the priest told me I had to confess. And all I could tell him no. You told him no? This was funny. After this, Gerta passes out. She comes to, and there is that priest again. Last chance, you're going to die. Time to confess.

And I say, no. And I pass out again. And then they evidently they removed him after that. Amira Plato, this is Science Friday from WNYC Studios. So that's pretty much classic Gerda right there. So if she doesn't believe the dinosaurs were killed by a meteorite, what does she believe happen? Well, so GERDA, first of all, she's actually, I think more than anything, she's in the never meteorite camp. More than any espousing a theory, it's an anti-theory.

Her position is opposite. And the reason she thinks that the impact hypothesis is wrong is because according to this hypothesis, right, this rock from space hits the earth, wipes out a ton of species. Geologically speaking, this happens very suddenly kind of all at once. And it's not just dinosaurs, it's plants, it's other animals, microscopic things. And so if you look across the fossil record, you should see a very abrupt mass extinction. But Gerta, when she looks at the fossil record, she doesn't see a sudden mass extinction. She sees these species dying out gradually. And she says that they start dying well before the point where this meteorite is supposed to hit. So her conclusion is that obviously there was something killing a lot of species at the end of the Cretaceous.

That's not in dispute. But she says it couldn't have been the meteorite because it just came on the scene too late. It's like saying someone got shot before the gun went off, kind of. Yeah. So one of the interesting things about working on this episode and really frustrating things is that paleontology, I mean, there's a lot of room for interpretation, not as much room as maybe, well, some people tell you, within reason. And the reason for that is that paleontologists are dealing with, you know, very imperfect fossil records from long ago. And if something is missing in that fossil record, if a species isn't there, it doesn't necessarily mean it's extinct. it could mean that it's just not preserved.

All right. Well, if you want to hear what really happened and what Gerta really thinks, you're going to have to tune into our podcast. That's Ella Fetter and Annie Minnoff, co-host and produce our science documentary podcast, Undiscovered. You can check out that episode next Tuesday and more stories from Undiscovered Season 2. Here's the address, Undiscoveredpodcast.org or wherever you get your podcast. Thanks, guys. Hey, thanks for having us.

One last thing before we go. I got a pop quiz for all our West Coast listeners. What has science beer and yours truly making bad science puns all night? Well, it's the Science Friday trivia. Yeah, we're bringing the party to Portland on October 9th at the Oregon Museum of Science and Industry. That's Portland, October 9th, and you join me for a laugh-filled night, a geeky science trivia as you compete for the title of Geekery Grand Master Champion. How would you like that title? More info at ScienceFriday.com slash Portland.

That's ScienceFriiday.com slash Portland. We had technical engineering help from Mitch Kim and Sarah Fishman and here at KUERR from Michael Havy, Lewis Downey, and Tim Slover. I'm I Refledo in Salt Lake City.