Smart Talks with IBM: Human Slavery Still Exists. Can AI Help Curb This Scourge?

human trafficking is a major one. When you consider the impact of the issue not just on those who are the direct victims, but also their families and communities, as well as the various companies that are profiting off the proliferation of human trafficking, it can quickly become overwhelming. That's why I was excited to speak with Neil and John, as they helped me get a better understanding of the issue and how technology is playing an intrinsic and at

times non intuitive part in combating human trafficking. First, let me thank both of you for being on the show, and before we get into the topic at hand, I thought it would be nice to get to know the two of you and to learn more about your background and what brought you to your current positions. So John, could we start with you. Could you tell us a little bit about yourself and what it is you do and how you got there. Sure. So, my name is

John McGrath. I'm a senior solution architect for IBM in Ireland based out of the Dublin Lab in in Ireland. My background, Jonathan, is fourteen years working in lab services for IBM, which involves the dealing with clients on a daily basis. But about two and a half three years ago I got involved in the Traffic Analysis Hub initiative and from that I managed to form a team called a Tech for Good Team in Dublin and uh and

that's what I do on a daily basis. Now I work with the Tech for Good Team excellent and Neil, can you tell us a bit about yourself and your position? Surely, Jonathan M. So, my name is Neil Jonles. I'm currently CEO of a new reform not for profit called Traffic Analysis Hub. My journey here is a torturous one. I spent thirty six years in law enforcement in the United Kingdom, concluding that time as Deputy Director of our National Agency.

I'm an organized crime intelligence expert UM and while I was serving with our National Agency, I came across a small not for profit called Stop the Traffic, who were specializing in preventing human trafficking. They began their work with the cocoa industry in West Africa that was using thousands of child slaves to pick cocoa for our chocolate UM and I was disappointed to learn that they knew more about trafficking than the intelligence systems in my national agency.

UM so began formula relationship with them too, to grow that understanding in the agency and and to begin to build unusual partnerships with NGOs and other subject matter experts. And when I left law enforcement nine years ago, I began working with Stop the Traffic more routinely, realizing that we needed a richer picture of trafficking if we were going to be effective as societies to begin to make

a history. We haven't done that yet, but we've begun to create that richer picture through the work that we've been doing. And Neil, I think you've hit on something that I really wanted to focus on in the early part of our conversation, the fact that even in your role in intelligence, that there was a lack of real knowledge about human trafficking. I think that certainly can apply

to the general population. I know that for myself, it's something that I am aware happens, and typically I don't really even think about it until I'm going through an airport and I see a poster that's bringing it to your attention directly, and otherwise I'm kind of in the dark. Can you give us sort of a an outline of how big a problem this is? Give us the scope and the impact of human trafficking. Human trafficking is pretty

well defined as a global phenomenon now. The the academic estimates, which are reasonable, suggests that something like fourteen million people globally are in circumstances that we would be comfortable to describe as trafficking and exploitation. Um, that's an enormous number of people. Even in in the UK, the best estimates suggest that something like d thirty five thousand people are

in circumstances of exploitation having been trafficked. So you could fill the biggest sports stadium that we've got twice with those people. And I think the best way of describing it to people is that it's it's an errant economy

in its own right. Traffic, trafficking and exploitation splits into two chunks thirty five percent estimate of those people and exploitation tend to be in some aspect of commercial sexual exploitation are in labor markets, particularly those labor markets that rely on seasonal workers contract workers, so agriculture, food processing, and manufacturing, construction, big fishing, sea fleets, logistics are very popular destinations for traffic labor where very criminal recruitment gangs

infiltrate them into the workforce. Most people's journeys into exploitation beginners journeys of hope. They're tricked into taking a journey on the basis that there's a great new future for them and their family, and then when they get to that destination, it turns the dust and becomes a creeping debt bondage situation. And it's worth something like three quarters

of a trillion dollars a year. We estimate. There's a new official estimate out this year or sorry, early next year that will define it slightly differently, probably m but but that's our best guests. I hope that that that gives you a sense of of of how the thing operates. It needs to recruit something like of its workforce newly every year, so somewhere somewhere up to eight million people

a year as a recruitment requirement. It's about money, and most of that money goes through financial institutions, And it's about creating a market, creating demand and maintaining demand. And it can't be solved just by the justice process, and it can't be solved just by humanitarian activity rescuing and rehabilitating Neil. That also brings me to a follow up question. Traditionally, what measures do various agencies and governments take in an

effort to prevent human trafficking? You had mentioned that this is beyond the scope of any one organization, but what are the sort of efforts that have been put forward? Uh, so far, we need traffickers to have a real sense of risk if they do this, that that they are likely to be discovered and held to account. And therefore there there is a significant role for investigators for the

justice process. But but more broadly, we need to think about the problem in an economic sense, um and and that's the aspect that I think has taken too long to develop. You know, in lots of parts of the world, the justice process doesn't work well, and of course trafficking is a global issue. In the more developed societies, the justice process does hold people to account, perhaps not in the numbers that we might like, but but it's a sanction that people fear um and and therefore it's a

very worthy element of of the program. Um And and encouraging other parts of the world where that doesn't work so well to get better at it is really important. But we have over relied on in my view, on on that outcome as the resolution to the problem. And of course, while there's money to make in good quantity and not enough fear of sanction, then traffickers will still

flourish and demand will still maintain or grow. Right, so without us having any you know, without addressing those root causes, what we're looking at really is dealing with the consequences, and that's just going to be a consistent issue without addressing those root causes. Obviously, this is an enormous issue that is going to require a lot of work across the globe in order to really tamp down on it. John,

I'm curious about how you come into the picture. We're about to start talking about using technology in a way to detect and then take measures to prevent things like human trafficking, how did you get involved with this particular challenge. Okay, So I think I mentioned earlier Jonathan that I was

working as a services person. So I was based in the Middle East working with some government agencies on behalf of IBM Security, and in my role, I had a give back opportunity and I was invited by IBM Corporate Social Responsibility to come to London to help facilitate a workshop for Stop the Traffic and that was the first exposure I had really to the issue of human trafficking

beyond what the casual lay person knew about it. But the thing that was interesting for me when I walked into the room to host the workshop was the attendees weren't just the people I expected. So I expected to see non government organizations and not for profits there, and I expected to see law enforcement agencies and some government agencies. What I didn't expect to see where financial institutions, and

there were a lot of financial institutions present. And it was really during that workshop that I kind of got the realization that this was across sectoral issue and the

solution had to come from multisectoral collaboration. So that was really the starting point for me and from that I worked with Neil and to stop the traffic team to learn more about the issue, and I spent many evenings and weekends in the hotel in the Middle Least building prototypes and sampling what could be done using various technologies. All are all based on this principle of how do

we get to data sharing collaboration around this issue. Can you talk a little bit more about those technologies, what form did they take? What was it that you were thinking, like, what metrics are you looking at and how are you analyzing them? Sure? The the starting point in the first workshop was there was kind of a division in the room depending on the agencies and the at the sort of core mission of each organization, but there was a

basic two requirements primary requirements that came out. The first was for this ability to do a global level analysis of the problem to see where the areas of intensity were, first particular types of trafficking, to be able to see how this is influenced by not just geography but by time. And then also there there was a requirement to be able to see the roots we're being used by the traffickers to move their their victims from point day to point b so, so that was kind of the one

half of the room. We're looking for this macro level view that would give them the global picture and and if you like, validate some of the high level figures, the estimated figures that Neil was talking about earlier. And then the other half of the room were more interested in, Okay, now that I know where the issue is, how do I pull that into a secure environment where I could start to investigate it and start to understand the network in more detail. Who are the people involved, how are

they moving people? What tools are they using? You know, what addresses, account numbers, all that kind of stuff. So we had this kind of a double requirement, so we started to look at what kind of technologies we used and used in the past which could help to satisfy

both of these requirements. While you were developing this in the early days, what were some of the lessons you learned, What were things that you know, were there pathways that you were taking early on that turned out to be less rutful than you hoped, or things that you discovered that surprised you while you were developing this early approach. Sure the well, one of the first things that hit us wasn't necessarily a surprise, Jonathan, but uh, the extent

of how it impacted us kind of surprised. This was the whole data privacy issue and the challenges around sharing data across jurisdictions. So so this became a reasonably high priority in our requirements, if you like. When we were trying to design the system. A lot of the basis of what we were trying to do is captured data from all over the world and make it available to

partners from all over the world. But we had to be very careful that we took out any sensitive information, any unique identifiers, and then we had to run the proposals true you know, various legal people to give us advice on whether or not we were following the right path. So not not so much a technical issue, although there are technologies that can help with this. It was more about,

you know, requirements issue. And then we started to look at things like, um, the largest amount of the data is contained in the narratives that the victims are, the narratives about the stories the victims, and to do that we we turned to natural language understanding and machine learning, and then we hit the challenges that everybody hits in this domain of making sure it's accurate, make sure it's unbiased, but also dealing with multilingual issues, so a lot of

the data is not necessarily in the primary languages. So that that was another one of the big challenges that we had to think about. Yes, this is an enormous challenge, justin in machine learning in general, is the natural language processing and being able to parse what someone means when they say something in particular way. And I imagine when you are trying to handle or analyze an enormous amount

of data, that problem becomes magnified enormously. What was it the a particular set of efforts that then led into the Traffic Analysis hub or did that come about in a different way. Yeah, the Traffic Analysis Hub came out of a kind of vision that stopped the traffic it had for a while. It became part of that workshop on the back in London. It was that macro level view that everybody could share and everybody got value from, and that became the primary target for the initial prototypes.

So when we were looking at that, we were trying to get a geospatial view, you know, a map based analysis of data. We were trying to figure out how to capture data, and then we realized that every different source that we accessed kind of classified their data uniquely, and now it's very difficult to do comparative analysis across these things. So then we hit the challenge of how do we make it consistent so that it makes sense to everybody. And then we we hit challenges like things

like locations. So there's lots of in the narratives of stories, there's lots of references to location. We needed to understand not just where location was referenced, but the context in which has been referenced. And then when we knew that, we had to go find a coordinates for it to

put it on the map. But we had to be careful that we were getting to correct coordinates for the correct location because there's lots of For instance, I think there's seventeen different Londons around the world, so we have to be clear about which London was actually been referenced in text. So so that was really kind of the

progression of the prototypes. Yeah, I think that for for a lot of people, myself included, we can sometimes fall into a trap where we're thinking about these very sophisticated systems pulling data as if it's magically all in a centralized, uniform database. I think the magical thing for a lot of folks who look into this is that we see how these systems are able to spot patterns, uh and trends in data sets that are so enormous that to us there's no signal, it's just noise. So seeing something

that can pick out the signal does seem a little magical. Well, as the t A hub is evolving and taking shape, have we already seen some impact in the real world? Is it being used right now to help identify and

prevent trafficking? Today? It's being used by over We have over a hundred organizations who are members of the Hub at this point, and all of them have their own secret missions or their own, uh, their own core missions of what they're they're trying to achieve with us, But we have anecdotal stories from various parties of where they've

got value from the data that's in the Hub. And sometimes the value, interestingly, is not just in the data, it's in the collaboration with their peer organizations and the other partners in the hub, which was part of what we tried to set out to do in the first place, was achieved as kind of safe collaborative environment where people could share their expertise as well as their knowledge for

the purpose of disrupting human trafficking. But we have got a lot of feedback from various partners where they've been able to validate data that they had seen in their internal systems when they were starting to investigate issues. They're able to validate some of that in the Hub by

looking at the data that we've been collecting. And then conversely, we've also had the same feedback from organizations who are investigators or say, we're able to identify new areas of investigation in the Hub that we weren't aware of because we've never looked there before, but once we started to look, we started to see patterns in our own data sets

in those locations. There are facilities for different audiences in the Hub, So you've got people like researchers and academia who come in and the facility we have in which in the hub, which allows them to navigate by concept through large news data sets, and that's a facility that they give us feedback on a lot that tells us it helps them to find information and to support their their research. We had one person who um Every month we have an analyst call in the community where the

community and the Hub come together. They look at the functionalities that we're building and the data sets that were gathering, and they give us direction of what they need and we feature a participant on that every month. So we have had a person who actually actually presented their thesis and part of their thesis was based on data that they pulled in from the Hub to to validate their own their ow own insights into human trafficking. That's phenomenal.

So not just building a system that's doing this very technical work, but also just building these relationships, forming relationships across various sectors and various countries that can all be you know, directed toward helping stop this problem. What other ways do you see the Traffic Analysis Hub impacting various industries? So we've well, first off, we've built a platform underpinning the Traffic Analysis Hub which allows allows us to reuse

the capabilities across different um issues. So we've also used it for things like food redistribution to avoid food waste, and we've also used it in the area of migration and population displacement and trying to create prediction models and stuff.

So the thing that kind of excites me about this is we're starting to bring in new sectors, but also not just industry sectors, but sectors within the n g O world who are focused on different parts of of of social issues and we're bringing together into one platform and one community and start to share information. So we've been approached by organizations who are who are focused on animal trafficking to see see if they can get access to the hub and start to share their data in

there as well. And we're all starting to see the reusability of some of the things that we've built. For instance, we've built a causality model in partnership with IBM Research, and where we were looking at the cause that the attributes that are most prevalent in causing things like population displacements, and these models are things that we can then reapply

from one use case to another. So we're trying trying now to move that model into human trafficking to see if we can determine, for instance, the the likely outcome

analysis for interventions in certain locations. To me, that's also inspiring because in that process you could be working on issues that are tangentially tied into trafficking, you know, some of those underlying root causes we were talking about, and being able to solve some of these social issues can also help remove some of those causes or at least diminish them somewhat, and thus have the sort of positive feedback loop of being able to solve these these traditionally

incredibly difficult problems, largely because it is hard for us to even get a grasp on all the data that plays into this. I sometimes liken this too, you know, making making the challenge of making a long hot term forecast for the weather. There's just so many variables that are out there, and they interact with each other in ways that we don't fully understand. It can be difficult to make anything, you know, uh, like a forecast that's

ten days out. On a similar front, we see this real world you know, unfolding of of trying to tackle these enormous social problems that also have all these different variables, many of which are at their heart human issues, and

humans are largely unpredictable creatures. So it's fascinating to see these systems that are starting to glean insights into the way these these large systems of people and and the way we work, how how they actually perform out in the real world, being able to draw conclusions about that,

predictions and perhaps solutions. Um, what would you say are some of the lessons you have learned in this, both just as seeing how the t A Hub and the related technologies have given insight into the human trafficking problem, and also lessons you've learned as as leaders in that space. Sure well, certainly from from my side, one of the big lessons I've learned is how super motivated the IBM

staff are to get involved in initiatives like this. It's been I was talking to somebody earlier today and I was saying, I could spend fifty of my time talking to volunteers within IBM who want to help, and they're all bringing individual skills and capabilities and experience here and offering to help us out with various pieces of the puzzle. So there's a huge potential here to apply technology to

some of these challenges. The other thing that's very interesting at the moment is a lot of these core major social issues, whether it's the pandemic, whether it's climate change, whether it's population displacement, whether it's trafficking. They're all intertwined and one is influencing the other. And the attributes that influence influence the prevalence of this of these events and different parts of the world, they're very often common attributes.

So we're trying to figure out can we build models that will help us to identify, you know, what are the attributes that are that are interesting and trying to lead a team through this, you know, keep them focused on stuff that we have to deliver, but also giving him the freedom and the ability to go and explore these new opportunities and new ideas. That's as a core

learning for me. Yeah. From my side, Jonathan, I think the first thing I discovered was that whilst we are absolutely data rich, we are terribly knowledge poor um and and the work that we've been doing together with IBM and the Tech for Good team, I think has begun to change that picture um and and then So the next key element in that chain of activity needs to be to ensure the widest possible appropriate audience can access that knowledge. Because no one's got enough resources to do

everything at once. It's it's it's the classic problem. You can only focus on so many things, so you need to use that knowledge like I would have used intelligence in an investigative way in law enforcement, to focus the resources that you've got at the hot spots and points where you can make a difference. And that that's how we get this thing on the run. And we need to we need to start undermining the economic pillars that

currently comfortably support trafficking in persons and exploitation. And I think that we've mind a decent stuff, And I like Neil how you brought that around to this challenge of being data rich and knowledge poor. To me, that was we're seeing that that that pivot now where the early days of big data seem to be an emphasis on look at how much data we have access to, and now we are kind of moving into a new era.

We're well into a new era really where it's how do we actually leverage this enormous fire hose of information. It's coming in from all directions, generated by more devices than ever before in the history of humanity. And we're actually starting to see systems like the the t A Hub, systems that are able to take that information and do something that's truly useful and impactful. How do you see the approach to trafficking changing over the course of the future.

What do you see as the evolution of addressing human trafficking. I think the big gains are in commerce and industry. I think that the ability forum for corporates to begin to understand where they need to focus their activities and what questions they need to ask of their suppliers, particularly and in difficult parts of the world UM And similarly for financial institutions, again it helps them because because every errant business has a banker and a banking facility UM,

and the clues are there. If the if the customer management process knew what those clues were and knew what questions to ask and and our view is that the more the more we grow access to the data that we've we have two businesses and financial institutions, the greater influence they'll have on opportunity or and reduce opportunity for

trafficking to flourish. Before I sign off in this episode, I just want to reiterate some other things we covered in this and that is these are non trivial problems. Both the real world problem of human trafficking, which is clearly non trivial, it is critical, and the actual computer problems that the teams are trying to solve in order to really take full advantage of artificial intelligence machine learning

and apply that to this incredibly difficult issue. Everything from natural language processing too, pulling in information from various sources and contextualizing it in a way that's useful. These are hard problems to solve, but as we've seen, it is worth it in the effort to stop human trafficking. I want to thank John and Neil again for joining the episode. It was an honor to talk with them about such

an important issue. I hope that you learned something in this episode, and I look forward to sharing more Smart Talks episodes with you in the near future. Take care, ye. Text Stuff is an I Heart Radio production. For more podcasts from my Heart Radio, visit the i Heart Radio app, Apple Podcasts, or wherever you listen to your favorite shows.