The Future is Now

20.01.25 The Future is Now

SM Sarah McConnell

AS Audio Sample

SJ Sylvester Johnson

AM Alex Monea

EH Eric Hallerman

SX Sara Maxwell

00:00:00

[music]

SM If you're paying attention to the tech world right now, you're probably hearing a lot of grandiose claims about artificial intelligence or A.I. Take this YouTube documentary, for example, narrated by A-list celebrity Robert Downey Junior.

AS Freewill. It's something we've been grappling with for thousands of years, from Aristotle to Descartes, and we'll continue to grapple with for a thousand more. Will we ever be able to make an A.I. that can think on its own? A second artificial version of me that is truly autonomous?

SM For some, A.I. is just the next item on a list of human technological triumphs.

AS Impossible? Well, when you consider what human cooperation has already accomplished; a man on the moon, decoding the human genome, discovering faraway galaxiesÖ I'd put my money on dreamer's any day.

SM Yeah, but these new technologies also bring new ethical questions and not just about whether the world really needs two Robert Downey Juniors. From Virginia Humanities, this is With Good Reason. I'm Sarah McConnell. Today, a closer look at how algorithms and artificial intelligence are shaping our society. Sylvester Johnson is executive director of Tech for Humanity Initiative at Virginia Tech. He shares his thoughts on how a humanistic approach can help us meet these challenges.

[music fades]

SM Sylvester, you started out as a scholar of religion. Do you remember when you were moved to begin thinking about things like artificial intelligence?

SJ I distinctly remember. I was doing a digitization project aiming to create a scholarly edition of a book published in the 1600s and went through four editions by a British cleric named Samuel Purkiss. And in order to do that, we partnered with an organization known as Early English Books Online that makes early English text available for study. So they actually paid a human to retype this - this text of a thousand pages. And they have 97 percent accuracy rate, which is very high, but which means that for every hundred words, three or more errors, if you average it out and if you extrapolate that over a thousand pages, there was a lot of correction that we needed to do to the digital file that we had. And so then we had to figure out how do we correct all these errors? Well, you can pay humans to do it, who would then introduce more errors, and it would never get corrected. Or you could use A.I. You could use a deep neural network algorithm. And so that's where it began. It was we need to do something that was conventionally, it would seem, to be very much the kind of work that I thought was in my wheelhouse. So, scholarly edition of a book about religion and race. And what I quickly realized is one I hadn't really I didn't know that machines could read first of all, you know, they had syntactical analysis. But also we - we trained it to recognize words that were misspelled and then from there, it got good enough to do the correction itself.

SM You have said that in the next decade, A.I. will be about 15 percent of GDP. Is that world or U.S. GDP?

SJ So it's - it's going to be 15 percent of the 100 trillion net increase in GDP that's happening in digital technology. So 100 trillion is just digital technology. And - and then out of that, 15 percent is going to be A.I. And these are obviously up approximations, so give or take a few trillion. But it's still - it's still makes the point.

SM And honestly, does that mean a lot of us will lose our jobs to machines and artificial intelligence?

SJ That's a great question. That's the concern of a lot of experts. We're facing an incredibly vast scale of growing inequality. Part of the concern is that it's easier to replace what humans do with machines. And so, no - no one actually knows the future, of course. But it - it would be foolish to ignore some very clear trends that are absolutely gonna create some deep challenges for us. And - and it is it is fair to say people don't agree on the year specifically 2040, 2050, 2030, they vary. But at some point we will see a rapid drop off in the degree to which people will be able to continue to participate in an economy with living wage jobs, given the ever accelerating capacity to replace their labor with machine labor.

SM Andrew Yang, a businessman who's one of the Democrats running for the presidential nomination, thinks the job loss will be tremendous, as you're saying, and that we need a human centered policy to anticipate it. He had an opinion piece in The New York Times that was headlined ëYes, Robots Are Stealing Your Jobí. And he says, for instance, self-driving trucks will put millions of truck drivers out of work in the not too distant future.

SJ He's right. So that was in response to a debate in which another person on that stage who is also running for president had suggested that it was inaccurate to say that machines are replacing jobs. And he's right to point out it's not inaccurate. This is - this is actually happening. People squabble about the details. What percentage of jobs? Will - will the new technology create more jobs that will have a net effect of replacing lost jobs? You know, no one - except for entertainment, no one drives a horse carriage to transport people anymore unless it's just for fun. But there are other jobs in automotive manufacturing, for example. And so some people have said, well, you know, A.I. is going to be the same way, so don't worry. And - and what's clear is that this is - this is different. And the difference is that we are the endgame of A.I. We are not at the endgame with automotive manufacturing, but we are the endgame with A.I. It's about being able to replace people. That's - that's not the deliberate point. So he is correct.

SM Can you give me an analogy that shows, this is not going to be like, yeah, we lost horses, but we gained cars and therefore mechanics. Give me a small parable of how, A.I. will actually not offer new opportunities for jobs enough to satisfy the population and the need?

SJ If you can think of a company like Instagram, for example, whose market capitalization was greater than that of the Kodak company at its height. But Kodak employed tens of thousands of people and Instagram was able to match that market capitalization with fewer than 100 employees. How is that possible? It's possible because corporations, this is - this is light clause, you know, ideally would have no employees. That's how you can be optimally profitable. And so they have to have some humans, but the fact that A.I., that algorithmic services can now extract value from people's data to make a company profitable, and the fact that the - the increase in the ability of algorithms to do things and this is the point of that initial observation that look like thinking and reasoning, people can say, well, it's not real knowledge. These machines don't really know anything. All right. Keep saying that. But just understand that decisions are being made by these machine systems. And if you want to believe that a machine system can make medical diagnoses, can make decisions on its own and in warfare about - about when and where to kill a person, can make decisions about transportation networks, can actually debate public policy by ingesting half a billion articles on a topic which is already happening. And if you think all of those things can happen without it actually knowing anything, then maybe knowing is overrated. But the point is that decisions are actually being made that are shaping the lives of billions of people. And increasingly, I'm not saying that there won't be any people involved, but increasingly more output, greater efficiency with fewer humans in it. So that's - that's the concern. And this debate is going to go on. But - but I don't think the debate should distract us from getting ready for a very different future.

SM It's confusing because we're told the jobs and money and careers, and our futures are in STEM. So we're pushing that in K-12 and college, and downgrading the value of the humanities. Are we digging her own grave? Are we marching inexorably to a system that will ultimately lay us all off?

SJ It's not destiny. So we - we can shape our future. In fact, we're going to. Whatever happens is gonna be a consequence of what people actually did. Whatever A.I. replace humans, we can't blame the A.I.s. It is people who are designing these systems and it's people who are leading our institutions. So what I try to emphasize, we have an initiative at Virginia Tech that is a university wide initiative called Tech for Humanity to focus on human centered approaches to technology, leveraging talent across our entire university. And - and one of the central aims, of course, is to actually shape a future that we want to live in. So what's I think is exciting about this technology, is that it's actually demonstrating that technology is not just a STEM issue. It is it is a technical issue, but it's also a political issue and cultural issue, itís a historical issue, it is one that is extensively economic and political and their - their intersectionality, the big questions around technology that we have to face and that are actually even harder than the technical side of technology are things like democracy, or what we've been talking about, the future of work. And the talent that we need to address these questions cannot simply be technical. If we're gonna have more wealth, so that 100 trillion dollars is going to come just in the next 10 years, approximately from digital technology needs to be put in context. The entire global economy in twenty seventeen was less than 100 trillion. So in the next decade, we're going to see this digital technology sector create more wealth than - than exists in the entire world in 2017. And that's gonna be in the hands of a very small number of corporations in a few places in the United States and China. And the rest of the world is going to share the rest of that sliver. So, meeting that challenge will not require us to know better math. And we should know better math, but that's not what's going to solve that challenge. What we need our humanist to solve that challenge.

SM So let's say just the one sector, that millions of truckers' in just a few years may be put out of work by A.I. and driverless trucks. How - how does your human centered approach to A.I. help them?

SJ It's a great question. We're not gonna suffer from a one of wealth. Okay. I'm just - that number about all the trillions that's gonna get created.

SM But that's just gonna be the one percent, of the one percent. Right?

SJ It could be, if we do it that way. Right. If we if we keep treating the infrastructure, the capital infrastructure and the A.I., I.P. intellectual property and the data equity, to be in the hands of a few people. So right now, we have what - what may be called data monarchy, where there are few entities that control the wealth creation.

SM That's a great phrase. A data monarchy.

SJ And what we need is data democracy, meaning that both the capital infrastructure. So think of server farms and the instrumentation as well as the software, the A.I. that is being used to leverage and monetize data. That those things if - if we continue to say that Sylvester Johnson, you know, name your tech company, already knows what Sylvester's gonna want to buy next month or where he's probably gonna want to travel or go. And they're monetizing that, you know, all of the data extraction that goes on. Sylvester Johnson has a zero equity right now in the monetization of Sylvester's data. I don't own i- even though it's my data, no one sends me a check for it. I don't get any money from that. And - and that's why that hundred trillion is going to be such a big number. It's the A.I. that is deriving insights from information from all of the data. While we're also able to collect more data than ever before. So if we say that only a handful of companies should continue to exercise their right to exclusively monetize everyone's data and take the lion's share of the profit, if we - if we continue to do that. And, you know, if we as people, I'll use humanities say, well, I'm not interested in that A.I. stuff and I don't think everybody should run out and do A.I., that's silly. People should do different things. But - but if none of us say, you know what, we're not going to create the A.I. so that we can become stakeholders in the digital future of humanity and in the future of capital, we're just gonna be resentful. And every time someone mentions technology, we're gonna say, oh, you know, I hate technology. They're trying to take away everything from us. If- if we continue to do that 10 years from now, 30 years from now, we're gonna be on the sidelines with our resentment, unable to participate as stakeholders in the future of capital and the digital future of humanity. But if we change the relationship between capital and A.I. and this project of humanity. And if we do that in a way so that there is shared prosperity instead of the concentration of wealth, then we don't have to have that radical inequality. But we - we have to do the work now for that future to be different.

[music]

SM Sylvester Johnson is director of the Virginia Tech Center for the Humanities. Coming up next, filtering out Internet pornography.

00:15:30

SM Many would agree that filtering out unwanted pornographic images from Google searches is a good thing. But our next guest says not so fast. Alex Monea argues that filtering out porn from Google searches catches a lot of other material that deserves to remain online and searchable.

[music fades]

SM Alex is an English and cultural studies professor at George Mason University. Alex, you're studying how Google uses its powerful search technology to decide when to censor X rated results. Has Google always filtered out X rated sites or is this new?

AM Google has been filtering pornography very soon after it began as a for-profit company. Around 1999, Google hired Matt Cuts and Matt Cutts, his first job at Google was to develop safe search porn filter. So his early days in the office were spent combing through Web sites for search signals that would indicate pornography, these were largely textual and he would build rudimentary filters and then asked people in the office to test them out, which there are funny stories about how uncomfortable that made people working inside Google early on. And this was largely because advertisers were already imagined as being the future of Google's revenue stream, and it was imagined that essential to these advertisers would be making sure that their advertisers never appeared alongside pornographic content.

SM So how strict were the early filters? Still, a lot of porn was getting through, right?

AM Tons of pornography was still getting through. Matt Cuttsí early filters. And that's because these filters were largely classifying pornography based on words or language. They couldn't look at the content of images or videos, and so what they would do is look on the websites that hosted these images or videos and try to figure out what sorts of words surrounding those images and videos on the site were likely to indicate that that site was featuring pornographic content.

SM So with that kind of graphic, are we talking, they'd look for names of body parts?

AM Certainly that would be part of it. But really, it didn't matter what the words were, as long as they were highly statistically correlated with the incidence of pornography. An example of an early and problematic one, is the term bisexual. So because the word bisexual oftentimes gets collapsed when we write LGBT, it doesn't occur as often online in reference to the actual identity type. It occurs much more frequently as a descriptor of pornography that includes a man and two women. And so there are periods in Google's history, as well as a couple other platforms histories, where bisexual was so statistically correlated with pornography that it was flagged as an indicator of pornography. And sites that use that word frequently were de-prioritized in search or completely censored.

SM You've said that it used to be that you and anybody else could put some word into an image search and eventually find sexual images, but it all changed at one point. Tell me about the early days and then what happened.

AM In the early days, these images could not be classified all that well. Google could not look at the content of an image and try to figure out what it was, especially not at Web scale, because, you know, there's billions of images that they would have to sort through. And so for many years, when you would search through Google photos or Google images, you would inevitably and randomly get pornographic or scantily clad female bodies in your search results for almost any word. And so I played a game when I was younger with friends, where we would try to input a random single word into Google image search and see if it would not lead you eventually to, if not pornography, sexually suggestive imagery. And it was almost impossible to win. You could search pizza and get to it, lawnmower and get to it. And this persisted all the way up until 2012, and then Google broke our game.

SM And how did it break the game?

AM Google learned how to automatically analyze the contents of the images themselves and classify whether or not there are pornography, based on what actually appeared in the image rather than language or words surrounding the image on the site that was hosting it.

SM Why was Google so intent on filtering out the material?

AM Advertisers are still notoriously conservative in the United States. And so there's been huge pressure to keep advertising dollars away from pornography and to punish companies that allow their networks to access pornography, that help individuals view pornography or that place advertisements in and around pornography.

SM Your idea is that most coders are white or Asian males who are screening out what they consider to be porn, and you think some of it actually isn't?

AM Yes, this is definitely true. So this has been a problem with any pornography filter, and the problem is that when you want to block pornography - if you want to block all of pornography, what that means is you're going to have a lot of false positives. And that means that if you try to block all pornography, there is going to be a ton of stuff that is certainly not pornography, but gets caught by the filter. If you want to have a medium strength filter that tries not to catch those things, things like art or sex education or LGBT discourse online. The problem is, is that some pornography will still slip through that filter. And so the question is always either do you want no pornography, but a lot of what's called ëover-blockingí of non-pornographic materials, or do you want a trickle of a little bit of pornography here and there, but no ëover-blockingí? And without fail, people always tend to lean towards over blocking and no pornography whatsoever. It always seems in these companiesí eyes to be much more important that no one ever see unwanted pornography than that lots of LGBT discourse and sex educational discourse gets blocked online. And so I've tried to track bias at all levels.

SM So talk with me about what the social problem is from all this. Who is really being hurt that Google and these other sites are over screening what they perceive as pornography or that they have very limited, undiversified perspective on what that is? Most people, as you pointed out, would say, oh, over screening is good.

AM Yeah. So number of communities that are hurt by this over-blocking, which only gets made much more rampant when it's automated. Right. So these people that have biases are - are just the beginning. They developed the algorithms and they train them on biased datasets that only amplify these sorts of problems. And it leads to a lot of negative impacts for a number of communities. And so you can trace this out in two directions. You can look at the way that this impacts people that are censored, like people that are doing sex work or people that are producing pornography. And there are tons of negative impacts there that are problematic at a social level. And then there are the people that are getting over-blocked or blocked accidentally. And these people could inspire more public sympathy. They have essentially done nothing wrong, but are caught in the overbroad filters that we're using on the Internet. And these people range from sex educators to community forums where LGBTIQ+ people are engaging in discourse among themselves or building community. There's also activists networks for LGBTQI+ people that are often inordinately blocked by these filters. So a really good project that was done by a transgender woman named Courtney Demone challenges these overbroad applications of the filters and also challenges the idea of binary gender that is embedded in these filters to begin with. So community standards across companies like Instagram and Facebook, as well as the automated filters themselves, are trained to think that there is an essential difference between male and female breasts. One is illegal and must be made invisible and the other is legal and can be laid bare without any problems. And this idea that there is a binary between the two has rendered really problematic in the case of a transitioning transwoman. And so Courtney Demone started her transition using hormones and began photographing her bare chest at the beginning of this transition and would post these photos on Instagram using the hashtag #DoIHaveBoobsNow? Begging the question of, at which point in this gender transition would Instagram and or Facebook, its parent company, determine that these breasts had become censorable on the site?

SM That's interesting, right? That's a - that's an interesting thought.

AM It's an incredibly interesting project, because when you're thinking about trying to specify in an algorithm what gets censored or not, you can easily see people on a coding team that doesn't have a transgender person. And that might not have a lot of interaction with the transgender community, thinking that the simplest route would be dividing bodies up into male and female and classifying breasts that way. But any trans person on a team could immediately tell you that gender binarism doesn't work that way, and that you need a more robust way to make that classification.

SM So are you advocating that we do something about this? That that we help Google and these other sites change their minds in response to people saying this is over censorship?

AM I definitely am advocating that we need to engage in a number of steps to try to alleviate some of these problems. One of the most important steps would be that we just don't have robust way of archiving or keeping track of instances where these filters are applied over-broadly. No one's keeping track of what is going on at web scale and what sorts of problems are emerging.

SM Right. What we're missing out that we don't know.

AM Right, and that's what makes it so hard to talk definitively about some of these issues, is you know, when I want to figure out how these filters are being applied over-broadly. I have to trawl through Twitter and Instagram and other Reddit forums where people are reporting having their content taken down. I have to screenshot all of that because it might be deleted later and try to keep an archive of all these instances that I'm finding of LGBT IQ+ discourse being over-broadly filtered. And so one of the first steps would be to advocate for research centers or non-profits to start keeping track of this and keep their eye on the ball, because this is a growing problem as these sites become increasingly conservative in their content moderation.

[music]

SM Alex Monea is an English and cultural studies professor at George Mason University. This is With Good Reason, we'll be right back.

00:28:00

SM Welcome back to With Good Reason, from Virginia Humanities. In 2015, the Food and Drug Administration approved the production and sale of genetically modified salmon. Some consumers were alarmed by the prospect of so-called Frankenfish. Eric Hellerman is a professor of fish and wildlife conservation at Virginia Tech, and he believes there's no reason to fear genetically modified animals.

[music fades]

SM Eric, when did GMOs, genetically modified organisms become a part of our diet?

EH That goes back to 1996 when the USDA approved the production of Roundup Ready Soybeans that were originally produced by Monsanto.

SM So have Americans for years been eating GMO soybean products?

EH Well, corn and soybeans, over 90 percent of each of those crops is genetically modified now, if youíve had processed food, you've eaten GMOs. If you had corn flakes for breakfast, it was GM. Most of the feeds consumed by the animals produced in America, they eat GM feeds. So it's pervasive in our food supply.

SM What about non-plants? What about animals?

EH Well, the United States government has approved only one GM animal for consumption by Americans, though it hasn't yet been marketed in the United States. That's the AquaBounty Atlantic salmon that's been modified by an additional growth hormone gene, such that it grows much faster than the usual Atlantic salmon. The regular, selectively bred salmon takes about three years to go from egg to harvest size. Now, the AquaBounty salmon, because it has greatly accelerated growth rate early in life, can be produced in about 18 months. So it's not surprising that a lot of growers are interested in producing the AquaBounty salmon.

SM I saw a picture of two salmon side-by-side. One was huge compared to the other.

EH They were full siblings. They were from the same parents. But one had the new gene and one did not.

SM Is there anything dangerous to us in the genetically modified zone?

EH Well, I was part of a National Research Council study that looked at a wide range of potential food products that might come from genetically modified animals. And we concluded that there weren't any real issues in this. In the case of the salmon, that growth hormone gene is not active in humans. It will not make your children grow faster. Allergenicity may be an issue, but in this case, they're taking a gene from one particular species of salmon and putting it in another species of salmon. If you're allergic to fish, you wouldn't buy the product anyway, so allergenicity is not an issue. And so you have to look at a case by case.

SM How afraid are you personally of the concept of unintended consequences? Sort of Jurassic Park style?

EH Well, I'm not particularly scared of this. There is a rigorous regulatory review process. It goes out for scientific peer review. I have more faith in FDA than I do in any other food safety agency on the planet. If they say it's safe to eat, I believe it. Now, in terms of the ecological sorts of risks. Well, that's what my colleagues and I like to consider. And so far, the sorts of organisms we're talking about are safe, to me. The one possible exception that we might be concerned about is this new effort to produce what we call gene drives, which are usually introduced into populations to try to drive them down. For example, you might have a mosquito that transmits a human disease. And we might want to put a gene drive in to the same species of mosquito to try and drive it to extinction. That's not fully understood. And so we need to do experimentation in isolated places to get a better understanding of that.

SM Tell me how a GMO is made. Are they born in a petri dish?

EH Well, sort of. I was a postdoc at the University of Minnesota many years ago and we produced genetically modified fish. There were two things about that, one was that that was very encouraging, that our technical approach worked. On the other hand, I was very alarmed that some people really wanted to stock those out in the waters of Minnesota very quickly, which in my mind posed risk and we needed to research that more. So this research was funded by the state of Minnesota. And as I presented a poster about it in the state house in St. Paul, a state congressman came up and asked me, young man, how quickly can we get these fish out in the waters of the state of Minnesota? Well, sir, we need regulatory approval. And he said, I don't care about that. I want to be able to advertise in the Chicago Tribune that really large walleye grow in the waters of the state of Minnesota so that people will come up and spend their tourist dollars here in the state. That scared me and I wrote up my qualms about that. And to my surprise, a leading journal published it. And this risk assessment research has been a part of what I've done now for 30 years.

SM You can understand the unbridled glee at the prospect of making so easily what seems like an improved product.

EH Well, indeed. And some of these genetically modified animals can produce phenotypes, in other words they can look different than anything we could breed by selective breeding. They can have traits that we can't get any other way. And so I understand the fear of that. I also see the promise of that.

SM Give me examples of the fear of it and examples of the promise of it.

EH Alright well - the fear of it, if we had walleye that grew twice as large, they could perhaps consume the entire prey population of the lakes in Minnesota. And so the ecological impacts are something that's foremost in our mind. Some of the benefits, though, that might be realized for agricultural animals are pretty reasonable. We have goats that might express an enzyme in their milk, that retard spoilage. We can control pest populations of insects. We can reduce the incidence of disease in animals. We can control the sex of chickens, perhaps, or for the layer chickens, we don't have to destroy all the males. We can produce cattle that don't have horns. So there's a lot of different ways in which GM technology can be applied to improve the lives of animals and of humans.

SM Is almost all the genetic modification of animals going toward controlling reproduction?

EH Actually no, a lot more of it has to do with agricultural production traits, the reduction of losses to disease, the production of only the sex that you want, the production of animal products that have qualities that you couldn't get any other way. A lot of it has to do with that. In insects, however, you have an awful lot of attempts to try and control pest populations. For example, there's been a lot of work done with mosquitoes to produce males that are sterile, they don't know that, they would go out and reproduce with others in their - their species. And because the young are doomed, that population would crash. So this has been applied, for example, to the mosquitoes that transmit Zika and dengue and chikungunya viruses to screw worms, which are pests of cattle production. And to Mediterranean fruit flies that are agricultural pests.

SM What is waiting in the wings in America now before the FDA and before the USDA?

EH There's genetically modified plums, they're called the Honey Sweet Line that are resistant to plum pox virus. There's the Innate Potato, it has less of a certain amino acid in it, asparagine, so than what you fried at high temperature, it doesn't produce acrylamide, which is not good for human health. The Arctic Apple, which if you cut it, the slices won't turn brown anytime soon.

SM And there's a group at UC Davis that is looking into goat's milk to prevent the goat milk from spoiling. You mentioned that earlier.

EH Sure. Lysozyme is an enzyme that lysis - that breaks the cell wall of bacteria. And in so doing, it retards spoilage. The notion here is that in large parts of the developing world, people milk goats and sell the milk, there's no cold chain of - of possession from the producer to the consumer. So the possibility that this milk might spoil in the half day or day until human children drink it, it would lessen the amount of E. coli poisoning or any other sorts of bacterial spoilage in milk. That would be a real benefit for those populations.

SM You mentioned the time when you were a postdoc and you had to stand up to this idea that we give early release into the wild from genetically modified organisms we hadn't really studied. At the time, did you get a lot of pushback from GMO companies who later came to appreciate your approach?

EH Yeah, what happened is a lot of the people that are proponents for animal biotechnology basically said, ëno, Eric, you're a turncoat, you're a traitor. You've turned upon usí. And what I said is, folks, unless we utilize these animals sustainably. The pushback will be much greater, later. It's well that we take control of the risk assessment right away and promote only those products that are sustainable. And in the end, we'll all be better off. Well, some of my colleagues in the field set out to do experiments to prove that my colleague and I were wrong. And lo and behold, we weren't entirely wrong and that many of the sorts of mechanisms we were hypothesizing were true. And so some of them turned around and said, well, Eric and Ann have a point and really we need to research this carefully. And so people that had been shunning me for five or even ten years suddenly started being friendly to me again. And that's great.

SM So what if we have the tsunami situation in Japan and we swamp the confinement? Or we have people who were unregulated elsewhere, who are releasing them into the wild?

EH Sure. Well, I mean, all of this depends on the regulation that's put on these different operations by the respective governments and by enforcement by them. And so we have to at some level, trust government to do it. Now if we have a tsunami the size of the one that struck Japan, the least of our worries is that some GM salmon might have gotten out.

SM What do you think is the best use for genetically modified foods?

EH I think perhaps the most promising area has to do with foods that are modified to present benefits to the consumer. For example, there's cattle and pigs that have been modified such that they'll have a heightened concentration of omega three fatty acids in their meat or milk. Many people have holdup heard of golden rice, that is a strain of rice that has a heightened amount of the vitamin A precursors in it. Thereís tomatoes and rice that have been modified to have heightened concentrations of vitamin B precursors in them. Thereís banana, cassava and sorghum that are fortified with other provitamins. So these sorts of things pose real benefits, especially for the developing world where vitamin supplements are not routinely taken, itís beyond the means of the people.

SM What about people who prefer natural organisms in their diet to modified? Is there any basis, in your opinion, to fears that people have that actually ingesting this food could be bad for us, let alone whether it escapes into the wild?

EH Well, these are the most tested foods that have ever been had on the planet. We know more about the composition of salmon today because of the AquaBounty Salmon than we did about salmon before. What I would tell the listeners is this food is safe.

SM Well, Eric Hellerman, thank you for sharing your insights on this on With Good Reason.

[music]

EH Thank you for having me.

SM Eric Hellerman is a professor of fish and wildlife conservation at Virginia Tech. Coming up next, what if we had Google Maps that showed whale, fish and turtle traffic in the ocean?

00:40:30

SM Fishing technology has advanced significantly over the years, but fishers still catch an enormous amount of what they call ëbycatchí.

[music fades]

SM Sarah Maxwell is helping reduce it. She's a professor of biological science at the University of Washington, Bothell, and specializes in finding ways to keep whales, turtles and other large marine animals out of the way of the fishers. Sarah, we understand if a turtle is caught up in a fishing net, how can a whale be accidentally caught?

SX It tends to be from nets. So, for example, a gillnet is set in the water, it's usually very long, in some instances it can be over a mile long. And so if a whale is swimming along and - and runs into that net very easily, it can get tangled. And depending on where the net is set, it might be too far below the surface. That whale can't come up to breathe and it potentially could die or be injured.

SM I've heard a staggering number of turtles, dolphins and other animals are killed in this way. Is that true?

SX Around the world, yes, it's a significant problem. It's a significant problem in the U.S. as well. However, the U.S. is arguably one of the best managed countries in terms of fisheries. Some of the numbers that I've seen globally are that around 50,000 turtles a year are caught in fishery bycatch or at least observed in fishery bycatch.

SM Do they live?

SX In some cases they do, and in some cases they don't. So some turtles will be released alive, some might be released with injuries and we don't know whether or not they'll survive in the long term, or some might be released dead.

SM How can sea birds, birds who fly be part of a accidental fishing catch?

SX Well, anyone who's ever been fishing knows that birds love fish. So often when you are on a pier somewhere and you're fishing, you'll see a lot of, for example, seagulls in the area. Seabirds in particular, are often attracted to fishing vessels for the same reason that they're attracted to fishing piers. So, for example, in long line fisheries, it's a very long line and there are hooks coming off of that long line, sometimes for miles on end as those hooks go over the side of the boat, they have some sort of bait on it and birds try and capture that bait before the line goes underwater. And so they can grab the bait as well as the hook and then be drowned or injured in that process.

SM How many birds do we estimate get killed in this way?

SX That's a tough number to come up with, but it's on the order of hundreds of thousands a year globally. Seabird bycatch has had some phenomenal successes. In one of the, quote, technological advances that has reduced bycatch in that fishery are known as streamer lines. And I say, quote, technological, because in reality, it's actually very simple. So, for example, what they do is as that line goes out, they basically just use these plastic streamers that come off the line and it's enough to scare the birds away from the hooks so that the hooks can get far enough underwater that the birds can't dive to get them. And that's reduced bycatch by over 90 percent in a lot of fisheries.

SM Some of the American fisheries have government employees embedded to report how many of certain kinds of animals they're catching in the bycatch. What's a good example of such a fishing industry in America?

SX Well, the fishery that's the most observed is the Hawaii Long - Long Line Fishery. And that fishery has observers on every single boat all the time because sea turtle bycatch has been a huge issue in that fishery. They've had that fishery ñ has been shut down as a result of catching too many turtles or an amount that has been deemed to potentially have an impact on the population.

SM You've been helping the swordfish industry work on this problem.

SX Yes. So, for example, in California, there's a fishery targeting swordfish using drift gillnets. It's put out in the water over a mile long and it tends to just kind of drift with the currents. Usually they're set in the evening and then brought back in, in the morning. The swordfish population in California is actually very healthy. Yet the bycatch in gillnets in general tends to be rather high. And so there've been movements to try and shut this fishery down. But one thing that is of concern is that in shutting this fishery down, we would end up importing swordfish from other countries where they're not as well managed and where the bycatch of some of these species, like whales and turtles, might actually be higher. So there's been movement to try and keep this fishery open. And we've been working with the fishery to try and find ways to reduce their bycatch. And one way we've been doing this is by looking at the movements of some of these animals, like whales and sea turtles. And if we can figure out what combination of - of environmental factors determine where they're likely to go, we can then predict into the future where they might be. And we can suggest to fishermen where not to go.

SM So you're using satellites to keep track of the large mammals in real time. So fisher people who are heading out over the ocean know which quadrants to avoid at a given time?

SX More or less, yes. So, we - in a lot of instances, what we've done is we've put satellite transmitter, which you can just imagine we more or less glued an iPhone to an animal. And then that tells us where in space and time that animal is. And then we can use imagery that comes from satellites, so this is a lot of this comes from NASA, and that gives us imagery of the oceanographic conditions in the area we can relate where the animals to the oceanographic conditions. And then we can basically predict in real time where those animals are likely to be in the future. I've been working with a master's student and some colleagues in the U.K. and also in Gabon. And what we've done there is we've attached transmitters to sea turtles to determine where they're likely to go because Gabon is a huge nesting beach for a number of different sea turtle species. So by determining what areas they're likely to use off the coast when they're hanging out in those waters coming up to lay their nest throughout the nesting season, we've been able to say, OK, these are the areas where we should keep fishing from being during the nesting season. And that's been really successful in helping to reduce bycatch in that area. And they've also just - the country of Gabon has created a system of national parks in the ocean, in part designed to protect sea turtles and using the information that we gathered using the satellite transmitters.

SM Is your satellite work with the tracking devices a new way of looking at this?

SX You know, it is. So, this has been a technology that's been advancing so rapidly in the last 20 years. It really has given us new insight into where animals are going and helped us to understand the threats that they're facing. And then also, more importantly, how to mitigate those threats in ways that we just - I mean, 30 years ago we could have never fathomed. So being able to know where animals are going, specifically what areas they're using, very specifically what habitats, weíre - they're using, has really made it possible to understand how they might overlap with humans such as fishermen.

SM It seems worrisome that the fisheries who have so much financial stake in this might also use the satellite tracking technique to just simply empty the oceans faster and maybe bycatch along with them. Is that a thought you have sometimes?

SX Not for the species that I work on. In most cases, these species are very highly protected. And so there's not the threat of fishermen using this data to try and catch, for example, turtles or whales. And while fishermen are definitely trying to make a profit, it's been my experience that no one appreciates the ocean as much as they do. And so their desire to reduce bycatch is usually as great, if not greater than, for example, mine, because they have a respect for the ocean, both because they spend so much time there and also because it is their livelihood, there - everything comes from that, for - for them.

SM Can you imagine a future where we're basically mapping the oceans the way we map our streets? Or is that just too hard, do you think?

SX I think we're - we're getting there. We already know so much about where animals are going. And as this technology, in particular, is getting smaller and cheaper, we're able to do more and more. So, for example, I put transmitters on a turn that weighs a little over a couple of pounds and the transmitter was one point four grams. This species had never been tracked before simply because this is a flying animal and you can't put something on it that's too heavy and might impede its - its ability to fly. And as these transmitters have gotten smaller and smaller and smaller, we've been able to see where animals are going that we couldn't even know where they were going, even just a year ago. So it's really exciting and I think that we're getting to an incredibly strong level of understanding of the ocean and how it works. And hopefully we can use that information to help continue to protect animals.

[music]

SM Well, Sara Maxwell, thank you for talking with me and With Good Reason.

SX Thank you so much for having me.

SM Sara Maxwell is a professor of biological science at the University of Washington-Bothell. Major support for With Good Reason is provided by the law firm of McGuire Woods. Support also comes from the University of Virginia Health System, connecting doctors and patients through telemedicine to deliver high quality care throughout Virginia, the U.S. and the world, UVAHealth.com. With Good Reason is produced in Charlottesville by Virginia Humanities, our production team is Allison Quantz, Cass Adair and Matt Darroch. Some of the music from today's episode is by Blue Dot Sessions. Special thanks to Bill Foy of Virginia Tech. For the podcast go to WithGoodReasonRadio.org. I'm Sarah McConnell, thanks for listening.