Partnerships in AI Drive Conservation Efforts: WWF’s Dave Thau

Wildlife conservation efforts might not be the very first thing that involves thoughts when one thinks about alternatives to make use of synthetic intelligence and machine studying. However Dave Thau, knowledge and know-how lead scientist on the World Wildlife Fund (WWF), can share myriad examples of how these applied sciences are serving to our planet.

On this episode of the Me, Myself, and AI podcast, Dave joins Sam Ransbotham and Shervin Khodabandeh to debate WWF’s many makes use of of AI and machine studying. Amongst them are functions that predict deforestation, analyze photographs from motion-sensitive cameras to establish species, optimize wildlife patrols to catch poachers, and cut back the unlawful wildlife commerce on-line. These conservation efforts will not be solely supported by nonprofit companions with shared objectives however by tech-company companions which are sharing superior AI applied sciences.

Learn extra about our present and comply with together with the sequence at https://sloanreview.mit.edu/aipodcast.

Subscribe to Me, Myself, and AI on Apple Podcasts, Spotify, or Google Podcasts.

Give your feedback on this two-question survey.

Transcript

Sam Ransbotham: From satellite tv for pc imaging to marine acoustics, wildlife conservationists can use synthetic intelligence to advance their important work. Discover out extra on right this moment’s episode.

Dave Thau: I’m Dave Thau from the World Wildlife Fund, and also you’re listening to Me, Myself, and AI.

Sam Ransbotham: Welcome to Me, Myself, and AI, a podcast on synthetic intelligence in enterprise. Every episode, we introduce you to somebody innovating with AI. I’m Sam Ransbotham, professor of analytics at Boston Faculty. I’m additionally the AI and enterprise technique visitor editor at MIT Sloan Administration Evaluation.

Shervin Khodabandeh: And I’m Shervin Khodabandeh, senior accomplice with BCG and one of many leaders of our AI enterprise. Collectively, MIT SMR and BCG have been researching and publishing on AI since 2017, interviewing lots of of practitioners and surveying hundreds of firms on what it takes to construct and to deploy and scale AI capabilities and actually remodel the best way organizations function.

Sam Ransbotham: Welcome. Immediately, Shervin and I are excited to be joined by Dave Thau, international knowledge and know-how lead scientist on the World Wildlife Fund. Dave, thanks for taking the time to speak with us. Welcome.

Dave Thau: My pleasure. Thanks for having me.

Sam Ransbotham: Let’s begin with the World Wildlife Fund. My first blush is that’s just a little little bit of an uncommon group that you simply wouldn’t usually pair with synthetic intelligence. How are you utilizing synthetic intelligence in your job?

Dave Thau: WWF may be very attention-grabbing. It’s a federation. We’re lively in about 100 totally different international locations. A lot of these workplaces are run utterly independently, however slicing throughout all of them, we’ve got a community of analysts. The group I’m on known as the World Science group, and we span the community. Now we have scientists on our group who deal with forest, meals, [and] local weather, and so they work with the individuals all through the community who deal with these areas.

I handle the Information and Know-how group, and I’m working with these scientists. We work throughout the group, serving to with tasks which are beginning up in all of the native workplaces. Now we have our personal set of labor inside the World Science group. Loads of that’s targeted on affect monitoring. After which I’m out on the earth speaking to different conservation organizations which are doing knowledge administration and synthetic intelligence and coordinating with them.

Conservation organizations have been utilizing AI for a very long time. One of many first functions of synthetic intelligence has been in land-cover monitoring. So there are satellites surrounding the Earth, monitoring the setting, however the indicators from the satellites are very noisy, and so synthetic intelligence has lengthy been used to do issues like establish “Am I taking a look at a forest? Am I taking a look at a grassland? What kind of land cowl am I taking a look at?” In order that’s one of many preliminary functions of machine studying and synthetic intelligence in conservation.

Sam Ransbotham: So that you mentioned “preliminary.” How far again is preliminary? Is preliminary final week, final month, final yr, final decade? When did you begin doing all this?

Dave Thau: The usage of machine studying to investigate satellite tv for pc knowledge typically dates again to the ’70s. I’m unsure when WWF began utilizing it for conservation functions, but it surely’s been fairly some time. I’ve been at WWF for 4 years, however the software of machine studying on satellite tv for pc knowledge preceded that. It actually broke by although form of round 2008, when NASA made the Landsat satellite tv for pc knowledge archives freely accessible.

It is a sequence of satellites which have been gathering Earth knowledge because the early ’70s, however up till 2008, you had to purchase the imagery, and so that you had been restricted to what sorts of analyses you might do. Round 2008, the U.S. authorities determined to make these publicly accessible and free, and so you then noticed an actual explosion of using that form of data.

Shervin Khodabandeh: Yeah. That’s fairly fascinating. We do a good quantity of satellite tv for pc imagery work at BCG. It’s not my space of experience, however I’ve to think about that with each this proliferation of information, as you’re speaking about, in addition to the upper and better decision that’s turning into accessible, in addition to the huge soar in computing and sophisticated neural nets and machine studying fashions, that the cutting-edge has modified lots. So possibly in case you can distinction, what’s the leading edge of these items right this moment versus possibly what it was a decade in the past or twenty years in the past?

Dave Thau: Yeah. The adjustments in machine studying particularly over the previous 5 years even have been huge. And it goes hand in hand with entry to computational assets and knowledge. Prior to now, you might do a Ph.D. on one Landsat scene, which is about … 100 kilometers by 100 kilometers — that was, you understand, leading edge.

Now, persons are usually doing international evaluation on these knowledge, accessing hundreds of thousands of those photographs, and that’s as a result of they’re accessible and likewise the pc energy is obtainable. What’s occurring now could be, the pace at which you are able to do the analyses is growing and the pace at which the info are collected can be growing, and that’s all been enabled by this explosion of information and computational energy and breakthroughs in machine studying.

Shervin Khodabandeh: Perhaps for our listeners, it could be additionally useful to begin explaining, what are a few of the outcomes or use instances that this sort of functionality permits you to predict or preempt or forestall?

Dave Thau: Yeah. Loads of examples. On the deforestation entrance, at WWF we work on a mission referred to as Forest Foresight. That is principally with WWF Netherlands, and it’s a forest loss-prediction algorithm and power, so it tries to foretell forest loss six months out. And utilizing Forest Foresight, they’ve been in a position to predict the place deforestation is more likely to happen and act on that earlier. In order that’s one instance.

We do lots with motion-sensitive cameras, additionally referred to as digicam traps. The substitute intelligence there may be figuring out species from camera-trap photographs. And it is a mission involving many NGOs [nongovernmental organizations], in partnership with Google who … they’re doing the AI a part of it. And there, the outcomes are massively growing the pace at which scientists can analyze the info.

Generally in these knowledge units, 90% of the pictures are attributable to leaves triggering the digicam. There’s nothing there that’s of curiosity to the scientist, so that they’re spending 90% of their time saying, “Nothing, nothing, nothing.” So with these synthetic intelligence fashions, they’ll in a short time simply do away with all of these and likewise establish the species. And utilizing that form of method, they’re in a position to goal totally different interventions, so in the event that they discover out that there’s an invasive species in an space that’s impinging on the endemic species, they’ll do one thing about it.

One other instance is utilizing synthetic intelligence to assist optimize patrols for people who find themselves in search of snares — like wildlife snares … to attempt to lure elephants. There are regular patrols that occur in conservation areas, and making an attempt to determine the simplest manner of doing these patrols — there’s a system that we’re utilizing referred to as PAWS [Protection Assistant for Wildlife Security], which was developed by Milind Tambe in his lab, that helps optimize these patrols. And that additionally makes use of synthetic intelligence.

Sam Ransbotham: All these are enjoyable examples as a result of, Shervin, so typically you and I are speaking about fraud detection or optimizing clicks or enhancing these algorithms and proposals for merchandise, which, you understand, once more, [are] crucially vital, however it is a enjoyable and refreshing different.

Shervin Khodabandeh: Yeah, and for good — additionally, for good.

Sam Ransbotham: Precisely.

Shervin Khodabandeh: And, Dave, you mentioned “species,” and I couldn’t assist myself possibly digress a bit. You’re the one knowledge scientist that has an ant named after them, so possibly you inform us about that story. And I’m assuming you didn’t use satellite tv for pc imagery to find this new ant species, however I couldn’t resist the joke. However we might like to know just a little bit extra about your background.

Dave Thau: Certain. In order that ant is Plectroctena thaui. It’s bought to have the “i” on the finish since you want it to sound Greek within the species names. It was Brian Fisher, who’s a researcher and curator on the California Academy of Sciences, who named the ant that. And that got here out of labor I used to be doing with the ant taxonomist neighborhood; again in, like, 2002, I began engaged on this factor. It’s a web site, nonetheless up, referred to as AntWeb.org, and it’s like social media for ant specialists. So all of the ant specialists use it to share their knowledge on ant taxonomy. They’re a really social group of individuals, and the platform may be very nicely utilized by that neighborhood, and form of as recognition of the work I did setting that up, Brian named that ant after me — which I really like.

Shervin Khodabandeh: That’s fairly poetic, too, that you simply go from floor stage and, like, subterranean all the best way to hundreds and hundreds of ft above floor stage to satellite tv for pc imagery, and an incredible instance of being on the 40,000-foot stage and likewise just a few ft beneath floor as a scientist.

Dave Thau: Yeah. Managing this ant taxonomy knowledge and tons of photographs of ants, that are wonderful — you need to go to AntWeb.org and take a look at a few of these ants. They’re unbelievable. There are about 14,000 totally different sorts of ants.

Sam Ransbotham: I believe we simply misplaced Shervin, he’s off on … I can inform he’s off on the web.

Shervin Khodabandeh: I’m going to have a look at your ant.

Dave Thau: And it’s imagery too. AntWeb has a lot of photographs of ants, and so it’s managing this imagery knowledge, and it’s the identical with the satellite tv for pc knowledge. It’s like, “How do you handle large quantities of satellite tv for pc knowledge in a manner that you may analyze it?”

Sam Ransbotham: That was a segue a bit to your background. I imply, I can see you’re a pc scientist and also you’ve been doing this for fairly a while. Take us alongside your path from being initially fascinated about these types of subjects to the way you’re on this place at World Wildlife Fund. What’s the lengthy and twisted story?

Dave Thau: Ha! It’s lengthy and … it’s too lengthy and too twisted. I’ll inform a considerably shorter model of it.

As a child, I used to be at all times fascinated about biodiversity. All of the books that I rescued from my dad and mom’ home once they moved had been about animals and vegetation and the bizarre issues that they do. And I grew up in New York, and we went to the American Museum of Pure Historical past regularly, and it was like heaven for me. I simply cherished it. After which we moved to California, and one of many first issues we did was [go] to Yosemite. And I used to be mountaineering across the redwoods in Yosemite, and it was wonderful. And in order a child, I used to be at all times fascinated about biodiversity.

What I ended up finding out initially was nothing to do with biodiversity. Properly, just a little bit; I used to be finding out cognitive science, and I used to be fascinated about how individuals categorize. One of many attention-grabbing questions on that’s, do individuals categorize pure issues, like vegetation [and] animals, in a different way than they categorize instruments like, you understand, forks and knives?

In order that was one of many questions. However what I used to be doing was a whole lot of modeling, and as a form of fallout of all that work, I picked up a pc science grasp’s diploma, simply because I used to be taking so many laptop science lessons in my graduate program.

Then, quickly after that, I went into trade. I labored for a monetary agency doing choices and derivatives coaching on the Chicago Mercantile Change. I labored for Wired Digital, which was the net a part of Wired journal. I used to be one of many individuals who began a web based advertising firm, which was one of many crazier issues for me. So I did many various issues that every one use my laptop science expertise.

Then the dot-com bust occurred, and I and plenty of of my buddies had been out of labor. And so I took that point to do some soul looking out and suppose, “What do I wish to do with my life? I’ve completed all these items since I bought my grasp’s diploma … however they’re not including as much as something.” That’s once I mirrored again on my love of biodiversity and nature. After which I simply appeared for a company that I might be part of that will let me do laptop science in that space.

And so I went to conferences, workshops, meetups, every kind of issues, and I heard about this mission referred to as the All Species Basis, which grew out of the Lengthy Now Basis, which is a bunch right here that’s targeted on long-term pondering. They usually had been making an attempt to speed up the invention and identification of latest species, and I believed, “Oh, that is the right job for me.” So I pestered them incessantly till they employed me, and that’s what kicked me off on what I’m doing now.

And since then … all I’ve been specializing in is biodiversity, conservation, and ecology utilizing the pc science expertise that I’ve.

Earlier than WWF, I labored at Google. For more often than not I used to be there, I used to be working with companions creating options utilizing the instruments that we had developed at Google. Particularly, I labored on this factor referred to as Google Earth Engine, which is that satellite tv for pc knowledge evaluation platform.

I labored lots with companions creating platforms on Earth Engine, and one of many issues that introduced me to WWF was making an attempt to know higher how these programs had been really getting used to drive change. My focus inside WWF is on the science aspect and on the conservation program aspect. Many of the issues that we work on are in partnership with different organizations, which is nice.

Sam Ransbotham: One of many issues I believed was attention-grabbing once we met a couple of yr in the past was a few of the work on smuggling. Are you able to clarify just a little bit about a few of these actions or endangered species and the way that’s working? I believed that was an enchanting use as nicely.

Dave Thau: One of many issues that we’re utilizing AI for, and that is in partnership with large tech companies, is making an attempt to cut back the quantity of unlawful wildlife commerce that occurs on-line. So there’s a whole lot of unlawful wildlife commerce taking place on-line by numerous platforms. And thru one thing referred to as the Coalition to Finish Wildlife Trafficking [Online], a whole lot of the massive tech organizations which have social media platforms and search engines like google have gotten collectively to attempt to restrict it.

Synthetic intelligence is a method that persons are utilizing to establish the place commerce is happening. It’s very difficult as a result of the language individuals use to do that is consistently shifting. I imply, it’s unlawful, so that they’re making an attempt to not be caught. It’s a really, very attention-grabbing problem.

Shervin Khodabandeh: That’s simply fascinating.

Dave Thau: Then there’s … nicely, an incredible instance of machine studying is World Fishing Watch, which is now its personal factor. It grew out of a partnership of SkyTruth, Oceana, and Google. They observe vessels above a sure dimension, that are mandated to emit a location sign each 15 seconds, and so they use that sign to find out whether or not a ship is illegally fishing in a conservation space, and likewise they’ve been in a position to observe when one ship will go offload their catch onto one other.

Sam Ransbotham: What I believed was fascinating about these tales is, naively, my tackle what your group [did] earlier than I talked to you was … I simply had no thought of the size of those types of issues. And every thing you’re saying to me feels like scale.

Dave Thau: Yeah. One space that basically pursuits me is the temporal scale. Usually, we’ll do some conservation work and we’ll monitor to make sure that we’re having the affect we would like on the time. However generally you’ll do conservation efforts which may take 10 years to succeed in affect simply because we’re coping with issues like “How briskly can a tree develop? How briskly can a inhabitants improve?” These are long-term efforts, and so there’s this form of temporal scale that’s actually attention-grabbing too; like, can we develop programs that can have the ability to observe issues over the long run in methods which are environment friendly, particularly on condition that, typically, funding for a mission would possibly finish earlier than the impacts of the mission are felt? So how do you develop tasks that may leverage AI and know-how to do this long-term monitoring? That’s one of many issues I’m spending a whole lot of time on.

Sam Ransbotham: It looks like a lot of this comes all the way down to a whole lot of measurement that these methods mean you can do, that beforehand you wouldn’t have [had] earlier than.

Shervin Khodabandeh: That’s an excellent level.

Dave Thau: Mm-hmm. We name it monitoring analysis and studying, and it’s the training half: You want to have the ability to measure the impacts with a view to study what labored and what didn’t.

Shervin Khodabandeh: I actually love the place you’re going with this, as a result of the factor concerning the temporal scale and whether or not it’s a 10-year scale or whether or not it’s some main indicators of one thing good or unhealthy that’s about to occur or has occurred is actually fascinating and is an enormous a part of what Sam and I’ve been speaking about round … you want suggestions loops.

Sam Ransbotham: We’re so enthusiastic about measuring what we’ve got knowledge on, and we’re thrilled in any respect the brand new knowledge that’s popping out. However so typically, we’re lacking out on the absence of information itself as a sign. And I believe that’s one of many issues that you simply had been pertaining to: that there’s an entire world of alternative within the knowledge that we don’t have, and the absence of information tells us lots.

So Dave, we’ve got a phase the place we ask you a sequence of rapid-fire questions, so we simply need you to reply the very first thing that involves your thoughts. What are you proudest of up to now with synthetic intelligence? What have you ever completed that you simply’re happy with?

Shervin Khodabandeh: Every thing.

Dave Thau: Yeah, I imply, a lot. Simply the flexibility to use AI to those conservation challenges in so many various contexts: the satellite tv for pc knowledge, the digicam traps, bioacoustics, eDNA [environmental DNA], the number of functions, the pure language processing. AI … there are various varieties, and what I’m actually happy with is that we’re leveraging as many variations of AI as doable.

I didn’t even speak about work we’re doing with the Basque Centre for Local weather Change on driving financing to nature-positive companies, utilizing AI to measure potential impacts of these companies. And that’s utilizing not customary statistical machine studying however extra symbolic AI, which is one more form of AI. So I believe I’m principally happy with the breadth and depth of the functions that I’ve been in a position to leverage.

Sam Ransbotham: As nicely you ought to be. So possibly, maybe except for bias that we’ve heard lots about, what worries you about synthetic intelligence?

Dave Thau: Proper. OK. So bias clearly is a large one, and there’s been a whole lot of dialogue on bias. One of many issues is the info sovereignty points. AI functions are so knowledge hungry; there’s a strain to have them eat as a lot knowledge as doable, and there are various contexts in conservation the place that’s simply not acceptable. One of many challenges, which is, I believe, an thrilling problem, is how do you progress ahead with these applied sciences [while] respecting the sovereignty of people that simply don’t wish to share the info for superb causes?

Sam Ransbotham: What’s your favourite exercise that doesn’t contain know-how in any respect? What do you do this’s not AI — apart from ants?

Dave Thau: Oh, let’s see. I’ve been studying lots about time recently, making an attempt to know how time works, and I’ve been studying a bunch of books on that, in order that’s been form of good. And I’ve been beginning to attempt to make music. It’s utilizing know-how, as a result of it’s electronica, however I’m making an attempt to make electronica by trying on the display as occasionally as doable.

Sam Ransbotham: What’s the primary profession you needed?

Dave Thau: Properly, initially I used to be going into neuropsychology. That was the very first thing. I used to be at all times actually fascinated about how individuals study, and so I used to be going to enter neuropsych to review how individuals study utilizing neuropsych.

Sam Ransbotham: What’s your best want for AI sooner or later? What are you hoping that we will get from all this?

Dave Thau: So nature and local weather are twin challenges, and so they go hand in hand. However local weather change is inflicting injury that we see now, and we have to handle it, and nature loss is one other disaster that we’re within the midst of.

In keeping with Residing Planet Index, there’s been a lower of 69% in species populations since 1970, proper? Utilizing AI to handle each of these challenges is what I actually hope for, however I would like it to each have international options and native options.

So I wish to, as a planet, have the ability to handle the local weather and nature-loss challenges utilizing AI on a planetary scale but in addition make it domestically related in order that the individuals within the locations the place land use is altering or who’re making an attempt to protect the character that they’ve can do this as nicely. So I’m hoping that AI can be utilized each form of globally and likewise domestically and that the native actors are as empowered to make use of these methods as the worldwide actors are.

Shervin Khodabandeh: Dave, it’s been such a pleasure speaking with you, from the very minute to the very large, and floor stage to satellite tv for pc stage, and sound waves to electromagnetic waves. It’s all been fairly fascinating. Thanks.

Sam Ransbotham: Thanks for taking the time to speak with us. Thanks.

Dave Thau: Oh, it’s been a pleasure. Thanks a lot for having me on. I’ve actually loved the dialog.

Sam Ransbotham: Thanks for becoming a member of us right this moment. On our subsequent episode, we’ll speak with Stephanie Moyerman, director of information science and wellness at Instagram.

Allison Ryder: Thanks for listening to Me, Myself, and AI. We consider, such as you, that the dialog about AI implementation doesn’t begin and cease with this podcast. That’s why we’ve created a bunch on LinkedIn particularly for listeners such as you. It’s referred to as AI for Leaders, and in case you be part of us, you’ll be able to chat with present creators and hosts, ask your individual questions, share your insights, and acquire entry to worthwhile assets about AI implementation from MIT SMR and BCG. You’ll be able to entry it by visiting mitsmr.com/AIforLeaders. We’ll put that hyperlink within the present notes, and we hope to see you there.