managemnet company strategy managemanet Can AI and Machine Learning Help Park Rangers Prevent Poaching?

Can AI and Machine Learning Help Park Rangers Prevent Poaching?

Can AI and Machine Learning Help Park Rangers Prevent Poaching? post thumbnail image

BRIAN KENNY: Artificial intelligence or AI for short is certainly creating a lot of buzz these days. And although it may seem like this amorphous thing that’s somewhere off in our future, it’s already very much in our midst. Navigation apps have turned printed maps into relics. Alexa, knows what you need from the grocery store before you do. Google Nest has the house at just the right temperature before you roll out from under the covers. And this is all great, but now you have to wonder if this intro is written by me or chat GPT. Which raises an important question. At what point does AI go from being helpful to being harmful? And will we recognize that line before we cross it? Today, on Cold Call, we’ve invited Professor Brian Trelstad to discuss the case entitled, Smart AI and Machine Learning for Wildlife Conservation. I’m your host Brian Kenny, and you’re listening to Cold Call on the HBR Podcast Network. Brian Trelstad’s research focuses on social entrepreneurship, systems change, impact investing, and the role of business in society. He is also a partner and board member at Bridge’s Fund Management and he is a repeat customer here on Cold Call. Brian, thanks for joining us.

BRIAN TRELSTAD: Thanks for having me, Brian.

BRIAN KENNY: I really enjoyed this case because I’m one of those people who thinks that AI is this amorphous thing in the future and it really is right here and it’s being used in some really interesting ways. And I thought this was a pretty remarkable application of AI with the Conservancy, particularly in Cambodia. So, I think people will be interested in hearing how this is being applied and also maybe what some of the pros and cons of it are. So, thanks for writing it. Thanks for coming here to talk about it.

BRIAN TRELSTAD: Thanks for having me.

BRIAN KENNY: I’m going to start like we always do, which is just to say, can you tell us what the central issue is in the case, and what your cold call is when you start the discussion in class?

BRIAN TRELSTAD: So, the central issue in the case is for students to grapple with some of the operational implications and the ethical considerations of AI and machine learning. In this particular context there is a coalition of international conservation organizations called SMART, which is a spatial monitoring and reporting tool that was developed to help park rangers in remote rural areas around the world grapple with the challenge of poaching. And so the cold call in the case is I ask whether the students, as the protagonist, Jonathan Palmer, who works for the Wildlife Conservation Society, otherwise known as the Bronx Zoo, and one of the principle architects of SMART, Would you add a new AI ML tool on top of a pretty decent data analytics tool that has helped park rangers manage and map their roots for patrol in conservation areas? So, do you add this new widget? Does it help? What are some of the challenges of getting it into the field? What are some of the ethical questions that you might have about increasing the chance of catching or confronting poachers if you were a ranger?

BRIAN KENNY: And we’re going to talk about poaching in a little bit, which is a huge problem globally, so again, people might be interested in hearing about that part of it. Why was this important to you to write the case? What are the things that you think about as a scholar, and how did you hear about this particular instance?

BRIAN TRELSTAD: So, I teach a course on social entrepreneurship and systems change that focuses more on organizations and doesn’t have a huge amount of data. I got connected through our India Research Center with one of our colleagues across the street at the School of Engineering and Applied Sciences (SEAS), Milind Tambe, and Milind, had been doing a ton of research on AI and ML, teaches a class on AI and ML for good.

BRIAN KENNY: ML being machine learning in case somebody’s not familiar with that.

BRIAN TRELSTAD: And I thought it would be fascinating to have a social change case that really anchored on the technical questions, not so technical that I or my students in an MBA class can’t wrestle with them, but sufficiently technical to get them towards the cutting edge of what is happening in artificial intelligence and artificial intelligence that intends to do good.

BRIAN KENNY: Okay. Let’s talk about poaching and the prevalence of it. The question there is how prevalent is it? What does the global landscape for poaching look like?

BRIAN TRELSTAD: It remains a vexing and pervasive problem around the world. On the one end there is trade in international body parts of animals, the rhino horn, that is leading to population declines, threats of extinction. On the other hand, you have vulnerable populations that are displaced by conflict or climate that poach for sustenance. And so you’ve got a range of criminal networks that are using this to fund their operations by trading in highly valuable endangered species or endangered exotic pets. And at the other end you’ve got folks who are just trying to feed their families and their communities.

BRIAN KENNY: What are the economics of it look like? It must be pretty lucrative if people are going to the lengths that they are to do this.

BRIAN TRELSTAD: It’s an 18-billion-dollar global market.


BRIAN TRELSTAD: The fourth largest illegal trade around the world. And the economics, again, are everything from feeding your family to helping to fund extremist gangs and terrorist organizations by engaging in the global trade.

BRIAN KENNY: Okay. So, let’s talk about CITES because they factor prominently into the case. That’s C-I-T-E-S. What’s their role, and what are the kind of things that they’re tracking?

BRIAN TRELSTAD: So, CITES is the Convention on International Trade in Endangered Species of Wild Fauna and Flora. It is an acronym for the first five of those letters. It was adopted in 1963 and is now managed by the United Nations Environment Program. It is a global convention where countries, I think 160 some have signed on, and they agree to prevent the illegal trade of wildlife across borders. It is necessary but not sufficient to stop that trade. So, it establishes this as a high priority. It gets governments to commit to combat the illegal trade of wildlife and poaching, but it’s severely underfunded. It is not operational to any extent, and so it leaves to every member government the responsibility to fight poaching on their own. The SMART Coalition is a group of nine international conservation organizations that have stepped into the breach and have recognized that across the world there’s far too few rangers trying to prevent poaching from happening. And so SMART was formed as a way to use better geospatial mapping tools to enable more efficient deployment of rangers using historical data. What has happened in the past? Where have we found poachers? Where have the animals been? So, what PAWS, the Protection Assistant for Wildlife Security, is trying to do is use artificial intelligence to predict where poachers are going to be using topography, weather, road networks and historical incidents. So it’s trying to look forward. The insight of this, in fact, came from Professor Tambe’s work with port security where he has worked with the Coast Guard on identifying threats to security of ports.


BRIAN TRELSTAD: The number of incidents that have happened in ports are relatively few, and if you only look backwards at those that had happened, then you’d focus on the areas where things had happened. He has used AI and machine learning to identify future risks in port security based on traffic of ships, based on security of the parts of ports. And that analysis was his “Aha!” moment when he spent some time in a conservation park in Uganda, recognizing that there were ways of not just relying on historical data, but taking in other relevant forms of information and using AI and ML to define hotspots where you might be better off patrolling a park or a port for security or anti-poaching.

BRIAN KENNY: Okay. So, it becomes a predictive tool in that regard. And we’ve actually heard about some other applications of this. If you just think about retail stores that are looking at weather patterns to determine how to stock their shelves, there’s a big storm coming, we need more shovels, that kind of thing. I don’t want to oversimplify it, but that logic really works in this instance. They’re looking at maybe migration patterns of animals in different parts of the world, so they can tell where poachers might be at a given time. Is that the idea?

BRIAN TRELSTAD: That’s right.

BRIAN KENNY: So, the case really focuses in on Cambodia in particular, and that’s where they’re piloting this use of the new technology to see if it actually works. What makes Cambodia a good place to do this?

BRIAN TRELSTAD: Well, there was a willing park management that was open to experimenting with. Look, Cambodia emerging from the war is a politically fragile state. It just recognized the importance of the Mekong River Delta and Srepok as a conservation area was only created in 2016. It’s an area the size of Rhode Island that has roughly 100 paid rangers or guards and another 500 or so community volunteers who are helping to patrol the area. The number of paid rangers per square kilometer is about half to a tenth of what is best practice around the world. And so you have an under-resourced but significant biologically conservation area that’s relatively new that also borders Vietnam, which creates inter-jurisdictional challenges to patrol and protect. And it became an interesting opportunity to say they’re using this SMART tool part of the WCS Coalition of organizations and does PAWS on top of SMART improve their ability to protect and prevent poaching.

BRIAN KENNY: What’s the profile of the rangers in a place like Cambodia? What’s the education level? Are these just local folks that are hired to do the job or are they coming from elsewhere?

BRIAN TRELSTAD: One of the sad parts of this particular case is that given that it was written during COVID, I didn’t have a chance to go to Srepok and that would be on my list of things to do. I spoke to folks who were there. Some of them work for international conservation organizations like the World Wildlife Fund, but many of them are local community members. Some of them are semi-literate. And so the challenge is you’ve got a range of deep expertise, deep familiarity with the place and training in conservation and others who are making $150 a month to do what is right for their community and are learning on the job.

BRIAN KENNY: And for the historical data in SMART that makes that work, I’m just wondering where do they capture that? Is that coming from some of these local folks who know the area who are patrolling it on a regular basis? Are they feeding that into the system in some way?

BRIAN TRELSTAD: So, SMART developed an application that extends into handhelds and GPS devices and asks for, with a pretty rigorous five-day training, to get park rangers to make note of where traps were found, where poached animals were found, where they spotted activities of poaching and log that in. And so using a handheld in the field, SMART in a fairly light touch way enables for that data to be collected as part of the routine nature of park patrolling. That data is then managed by somebody at a central place who’s been trained to look at the data and then say, “Okay, next week here’s where we think we should be patrolling, given what we’ve seen historically.” And it’s being used in over 1000 sites in 70 countries right now and has been demonstrated to reduce incidents of poaching in the sites it’s been used. It’s still, however, only in less than 3% of all protected areas around the world. So it’s an interesting new innovation. And that’s part of the dilemma facing Jonathan Palmer, as our protagonist. Do you add new bells and whistles that might help you go forward, or do you focus on scaling an existing solution that seems to be working in the 3% of places where it’s being used?

BRIAN KENNY: And one of the things that you wonder about with this, any kind of implementation of technology is the business process change that has to happen. People have to do things differently in order to make the technology work. And I would imagine that’s a big challenge in a situation like this.

BRIAN TRELSTAD: That’s absolutely right. And they have focused heavily on the training of rangers. They’ve made it icon driven. There’s a number of languages in which the tool is used, and the training has been translated into many languages. So they’ve done a very good job of grounding it in the experience and the education levels of the rangers to foster deployment and encourage the proper adoption.

BRIAN KENNY: So, we’ve got SMART and that tool seems to be working and it’s being deployed, maybe not at the scale that you’d want it to be, but the more skilled they get at it, the easier it is to scale. And now they’re looking at PAWS, which is this predictive technology that’s based on a whole different set of data. How does that interface with SMART? What does that look like?

BRIAN TRELSTAD: So, what they’re looking to do is create a shared application that takes topographical, or weather, or other economic data from the region, so Srepok in Cambodia, and feed that into computational software that will do analysis of all those different data points and provide the rangers with amplified assessment of the park, not for specific route recommendations like SMART does, but for, “Where are the hotspots? Where do we think that poaching is most likely to happen?” That translates for the park rangers into a new map of the park, which they can then decide, “Should we in fact patrol in these new hotspots.” One of the limits is that the PAWS tool is agnostic to where are the trails and the paths, and where can you get to on a motorbike? A lot of the parks are patrolled with rangers on motorcycles or off-road bikes. And the rangers will often go out and be on patrol for three or four days at a time. The challenge operationally seems doable. There’s a fair bit of data that can be aggregated centrally. There’s a machine learning process that requires some computational power that’s achievable. But the question is whether the end format of these hotspots are incrementally that much more useful for the park rangers than the patrol map recommendations that SMART comes up with. So there’s incremental cost, there’s incremental time, there’s incremental complexity. Does it lead to measurably incremental value? And part of what the team at SEAS, Professor Tambe, and his graduate student Lilly Xu, were trying to do was understand whether or not PAWS tool would deliver this sort of value. And so the data is still being collected on how this translates into incremental value for the rangers.

BRIAN KENNY: And this is obviously the kind of decision that a lot of business leaders are going to have to grapple with as AI becomes more and more common. Which is going back to my introduction, it’s how much information is helpful?

BRIAN TRELSTAD: That’s right. And there have been lots of new technologies, bells and whistles that have approached conservation organizations. And the fact that SMART has gained the kind of acceptance in the parks where it’s being used is a question of should we stick with the thing that’s actually working that, or do we have to take the flavor of the day? And so it’s a really rich debate, particularly as it was taught at in Seas where students are constantly trying to push towards the cutting edge, recognizing when is good enough and when do we need to be at the cutting edge.

BRIAN KENNY: This is the engineering mindset versus the management mindset maybe?


BRIAN KENNY: So, you also mentioned that there are some ethical considerations that come up here in the case. Can you maybe talk a little bit about what some of those are like?

BRIAN TRELSTAD: Yes. So, the ethical questions that come up really center around both the experience of the rangers and the poachers. And so on the one hand, the rangers want to try and deter poachers and want to try and prevent poaching, but they’re not that interested in direct conflict with the poachers themselves. There have been a handful of examples of rangers being shot and killed around the world by poachers. And so the question is whether or not the route mapping of SMART is a good enough way to make sure that you’re getting the right number of traps out of the park as quickly as you can with limited manpower. The fear with the PAWS tool, if you’re going to be going into hotspots, are you increasing the chance of potential conflict, potential armed conflict? And the PAWS team doesn’t know the answer to that yet, but it is something that the SMART coalition has raised questions about and is something that the PAWS team is aware is a risk. The other end of the spectrum, if you’re in a place where subsistence poaching is happening, not that it’s acceptable, but where there’s an attempt by the government to deter poaching, if you capture somebody who’s trying to feed their family and the consequences for that are prison or worse, is that something that PAWS wants to be in the middle of helping to adjudicate? And so, the PAWS team is aware of those consequences as well. And so how can you reduce the chance of conflict while increasing the likelihood that you contain and minimize the negative effects and harms to wildlife of poaching? And those are the central questions that at the end we have students debating the pros and cons.

BRIAN KENNY: And I guess this is what Palmer’s wrestling with too as he goes through this decision-making process.

BRIAN TRELSTAD: That’s right. And he and his colleagues at the other eight conservation organizations, many of whom are in the regions where they do the work, working with the park rangers and staff, get to know their colleagues quite well and are wrestling with, “How do we best equip our colleagues who are doing really important work and not set them up for an increased personal risk?”

BRIAN KENNY: So, in your experience at looking at different applications of AI, is this a really unique situation or is this the kind of thing that could be applied to other businesses or industries in a relatively straightforward way?

BRIAN TRELSTAD: Well, I will confess that one of the great privileges of being here at HBS is to look into things that we’re not as familiar with. And so this is my first chance to really understand AI and ML in a kind of social change context. I think that there’s a lot of techno optimism that we see, and the social sector, in my experience, is often slower to adopt technology. But once they are adopted, they can have profoundly positive effects on the management of an organization or the delivery of social environmental programs. And so part of what I was looking to do is understand how do you approach the development of a new methodology that is still bleeding edge, but do so in a way that is very conscious, again, of the operational and ethical considerations. And I think that’s a capacity where many nonprofit organizations are actually quite resource efficient. But they don’t have lots of money to invest in IT or new capital that’s not going to translate into mission. And this was a really helpful case where again, the SMART coalition is bootstrapped, each organization contributes 30, $40,000 a year. So it’s not like this is a super well resourced, this is operating in between a number of incredibly well-intentioned and highly skilled professionals, who are almost doing this nights and weekend as this thing that doesn’t really have any organizational coherence. So, if they’re going to make investments in time, training, new technology development, it really has to drive mission further. And I think that, to me, is the cutting edge of the kind of cases that I’d like to study more how AI and ML can enhance rather than distract from delivery of mission.

BRIAN KENNY: And it does seem like this tension between those who want to really push the technology boundaries and those who want to take a more measured approach is probably going to be inherent in almost any AI discussion that’s happening in organizations. Because we do tow that line about what’s good and then where does it become not so good?

BRIAN TRELSTAD: That’s right. I think that’s right. And I think this is representative of some of those debates.

BRIAN KENNY: This has been a great conversation, Brian. I have one more question for you before we let you go, which is just if there’s one thing you want our listeners to remember about this case, what would it be?

BRIAN TRELSTAD: I think it would be to center on the experience of the ranger and how in limited resource environments, they are doing a tough and important job and there’s incredible power for technology to make them more effective at what they’re doing. And there’s incredible power of ideas that are invented in labs in Allston that might be a distraction. And I was really pleased at how aware everyone from our colleagues at Harvard, to the conservationists in DC, to the park rangers and staff in Srepok, were aware of that trade off and just to center on the use case and the user and how it will make their lives better.

BRIAN KENNY: Brian, thanks for joining me on Cold Call.

BRIAN TRELSTAD: Thanks for having me.

BRIAN KENNY: If you enjoy Cold Call, you might like our other podcasts, After Hours, Climate Rising, Deep Purpose, IdeaCast, Managing the Future of Work, Skydeck, and Women at Work. Find them on Apple, Spotify, or wherever you listen. And if you could take a minute to rate and review us, we’d be grateful. If you have any suggestions or just want to say hello, we want to hear from you. Email us at [email protected]. Thanks again for joining us. I’m your host, Brian Kenny, and you’ve been listening to Cold Call, an official podcast of Harvard Business School and part of the HBR podcast network.

Leave a Reply

Your email address will not be published. Required fields are marked *

Related Post