The Decisive | Season 3 Ep.19 - How AI Is Influencing Shipping and Supply Chains

View Full Transcript

Kristen Hallam:

You're listening to the Decisive Podcast, insights and analysis to empower confident decision making. Welcome to the Decisive Podcast brought to you by S&P Global Market Intelligence. I'm your host, Kristen Hallam. In this episode, we're discussing the use of artificial intelligence, hereafter called AI, in shipping. And joining us for this discussion are Eric Oak, Senior Supply Chain Analyst at Market Intelligence, and Eric Johnson, Senior Editor, Technology, at the Journal of Commerce. Welcome to the podcast. It's great to have both of you on the Decisive, although I'm feeling a little left out as the only non-Eric here.

Eric Oak:

Thank you. Great to be here. Well, we'll induct you to the club, honorarily.

Eric Johnson:

It's great to be here. And, also, I had no idea before we started working on this, Erics were allowed to discuss AI, but now we know.

Kristen Hallam:

Now you know. Now you know. So let's start our discussion at a high level and talk about the difference between AI and automation.

Eric Oak:

I kind of have three different ways of thinking about these technologies across the spectrum. So when you think about AI, a lot of times, especially right now, that term is being used to describe generative neural networks. So kind of Gen AI, which is essentially these super cool, super deep neural networks that are generating new content from, you know, seeds. Or just if you let one sit there and do what it wanted, it would actually just start generating things randomly. The kind of the mid version and what I would still consider AI personally, but I'm a little bit old school in my approach is things like, you know, regression, Bayes, you know, support vector machines, stuff like that, which is machine learning really in its core, but it is essentially a form of artificial intelligence. And then finally, my definition of automation would be just anything that uses essentially if/else statements, essentially a deterministic pathway through a set of decisions. These were also called expert systems back in the nineties. And so that's what I would call automation.

And automation can be very complex and have lots of different branching paths. And, you know, you can maybe as a human, you might not know what is going to come out of an automation before you start it. But, at its core, automation is deterministic. You do know for any given set of inputs what will come out of an automation. With AI, you sometimes get what you expect, but sometimes you get stuff you don't expect. I think that's really one of the core strengths and weaknesses of generative AI in general.

Eric Johnson:

Yeah. Kristen, I think that's an extremely enlightened cerebral understanding of those two things. Part of what I try to do is, like, explain to people in the logistics shipping world who are not trained in this stuff, how to think about it. And the distinction I make is I sort of think of automation as something where there's a task or process that there's a limited number of pathways that result from that process, and you let a program or a system sort of do those things instead of a human doing. That's, like, very big picture how I think about automation. And then AI is this huge tree of different types of tools with different branches of artificial intelligence that have been around for eighty years or more. And those, depending on which branch of the tree you're doing, they're essentially all trying to mimic human intelligence or human cognition in some way. Right? So if you think about it again at the sort of outcome level, AI is trying to affect the decision-making that a human might take, and automation is just replicating something that a human might do, but in a more efficient, theoretically efficient and faster way.

Kristen Hallam:

Got it. Got it. That makes sense. And with those definitions in mind, how is AI being used in shipping today? What kinds of challenges or pain points are companies trying to address with AI?

Eric Johnson:

Well, so I think the first thing to understand and what I've sort of had to understand over the last couple years and what I think others should internalize is AI is sort of already very much part of the tool set for anybody building any type of software these days. Right? It's not like ketchup on French fries or ketchup on a burger that you may come across and you may not—like, you have to just assume that most software companies are using some type of AI in the formulation of their solution. So it's sort of everywhere, and yet it's, most of the time, it's sort of abstracted from the view of the user of the software, which, you know, for decades and decades, that's sort of how software worked. No one really cared how it was made or what type of foundational architecture it had. They just cared whether it was easy to use and it worked. Over the last decade or so, we've come to be very intrigued by how things work, not just whether they work. So there's been a lot more focus around that. So it's sort of everywhere, but I think where we're starting to see people sort of understand where it is, is in functional processes where a program again is doing some sort of automation and AI is a part of the tool or mostly the tool that's used to achieve that automation.

Now, there are tools that help you figure out what you're looking for without having, like, the compass or the map to figure out, to know what it is that you need to look for. So that's just a basic example. And I'm certainly seeing it sort of used in very discreet ways, generating customs entries out of an email, out of a flat file in an email, taking a huge amount of data and figuring out what the right pricing for a freight forwarder is at any given moment. Right? So, I mean, I could go on. We could spend the whole time just talking about the individual use cases. But the reality is it is here. It's sort of already behind the scenes of almost everything that we're using anyway. And it's really like understanding maybe new cases where it's doing something different than kind of older software that people are used to using isn't using the AI. So.

Eric Oak:

100% correct. I think one trend that I've kind of noticed is that AI seems to just be a continuation of the digitization trends that we've seen with shipping kind of for a while now. So trying to put all the legacy bills of lading and stuff like that into a searchable database. AI is an extension of that path. And that's not every AI platform, of course, but just that seems to be a trend within the shipping industry is you could use some of these new tools to kind of extend that process and even sometimes make that process easier to move into digitization.

Kristen Hallam:

And I think you both have touched on this a little bit, but just getting out the crystal ball for a second, what does the future of AI and shipping look like?

Eric Oak:

Yeah. It's a great question. I think, generally, what we'll see is—and this is, again, my own opinion, my own prediction—generally, what we'll see is the increase of its usage in the back office. And then at some point, I would expect one or two kind of game-breaking, you know, extraordinary use cases to surface and obviously a company to—or several companies to put those together out there. I don't know what that necessarily could be, but I would imagine that's something around risk management, understanding your goods, where your goods are flowing, etcetera. My market is more of a supply chain perspective. So I know, I apologize.

I'm not thinking of this as much as a carrier, but as a BCO, understanding where your shipments are and then having an agent that sits there and receives, let's say, an alert from a carrier that the box is delayed and the box is routed, takes that alert, you know, automatically knows, okay, is this, you know, an input widget or a pen for the office? Is this important or not? What do I need to do? Do I need to escalate the notification? Do I need to send that up to our ERP system to change our manufacturing plans? Things like that, where you get this instant automatic alerting to notifications or events or information that's coming into your business. I think that's going to be a really core component at some point, whether it's, again, different flavors of AI or automation or whatnot, but having that kind of quick decision-making that then humans can look at and audit later is going to be, I think, a big part of this. The other side of it when an invoice comes in, and I think some systems are already doing this now, I've heard talk of this, would just be, you know, any invoice that comes through the door, whether it be shipping or, again, just general POs or procurement. Audit that invoice against your ERP, against what you have in your inventory files, and just immediately make a decision. Is this a good invoice or a bad invoice? Don't even involve a human that could potentially, you know, be fooled by a phishing attack or something like that. Just immediately look at it and then obviously kick up anything that looks weird to your accounts payable team. So I think there's a lot of places where it augments the human experience or the human work. There's always—I firmly think that there always will be a human in the loop somewhere, whether that's one human to a hundred bots or one human to a million bots. We don't know at this point. But generally, I would say that these systems are going to increasingly kind of automate the routine aspects or the routine kind of interfaces between different companies as we move into more and more capable systems.

Eric Johnson:

And, I mean, what Eric just described is very much happening right now. Over the last few years, you know, deploying AI-enabled tools to figure out whether an invoice is accurate and not having that take days and weeks and a human to do it is very much in play. So AI has been sort of evolving, as I had mentioned, for decades. Right? And yeah, it does feel like there's this interesting time that we're in right now where there's just an acceleration in our knowledge of what it is and what it's capable of. But in the meantime, like, if you overlay that with, we went from a period, let's say post the dot-com boom through 2010 in the logistics industry. That period was mostly about getting people comfortable with doing things in a multi-tenant cloud-based environment, like essentially using our browser and paying monthly for something. And then that sort of continued straight into a decade where we went from big data.

Remember, everyone was like, think of all the things when we have all the data that we never had before. And that quickly transitioned to too much data. We have all this data coming in. We can't make any sense of it. Well, if you think about that sort of evolution and then you layer AI underneath that, in a perfect world, let's just assume for a second that all that big data is good data. AI is the only way to actually make sense of the amount of data that we are now generating and trying to consume at any given moment across some of the systems Eric just said and even more broadly. Right? You have a bunch of systems generating data within your company, and then you have an infinite number of systems and data coming from outside your company that you could or probably can't use because you can't make sense of it. So that, I think, is the real interesting, like, big picture kind of where this—you just asked me to take the crystal ball out.

Like, there is finally maybe a path where all of the data that everyone knows exists, but we have no way to, like, even think about how to make sense of it could be made sense of. Right? Because now we have the horsepower to actually, like, sort of get it under control in some way.

Eric Oak:

100% I mean, there's—you can think of some wild examples. The one I always think about is, you know, if you're a convenience store, can you tell whether your coffee machine is broken by the traffic outside your store? They're completely separate data sets. They would come from different sources, but theoretically, they could be highly correlated and they could tell you stuff that maybe you don't know about your own business. So, yeah, that's great. Okay. Weird. That's a hard question, Eric, but do you think that there's going to be another AI winter or do you think we're pretty much there?

Eric Johnson:

It's sort of natural that there are these ebbs and flows in investment too, because like everybody jumps on something that's hot. It maybe doesn't work as quickly as everyone hoped. It's hard to find use cases that are unique and can drive scale. And then, you know, they find the next shiny penny on the sidewalk to go after. And then maybe come back once, you know, the glare is a little bit off and you can actually, you know, find real use cases that can scale. So, yeah, I think there will probably be—there will definitely be in the tech periodicals, there will definitely be, you know, AI's winter, AI's at the bottom of the roller coaster or whatever. And then it will eventually come back because there's no putting this horse in the barn. I don't think.

Eric Oak:

Yeah. I think that's the important point is that I think we found enough use cases for it that it's never going away at this point. But will we be at the singularity next month? I don't know. But I think, you know, maybe we'll call it the AI autumn. I mean, it's just cooled down a little bit.

Eric Johnson:

I don't know what your lives are like, but I can't even coordinate a meeting like a dinner out with, like, six friends. So, like, the idea of getting the singularity next month seems a little fanciful. So, yeah.

Kristen Hallam:

Yeah. Well, this leads in really nicely to my next question, which is about investment. So I mean, how much investment is flowing into these kinds of AI applications for shipping? You know? And is it an upward trend? Is it a flat line? Are we starting to see the beginning of the autumn, as Eric says, and it's falling a little bit?

Eric Johnson:

Yeah. I can definitely take the lead on this one. So for sure, AI, if you have AI in your name or your pitch deck or it's a critical part of your solution, you are automatically getting more attention and you're automatically—in some cases, it's almost like if that's not part of it, you're sort of excluding yourself from a certain set of investors. So certainly, investors are all over this. I think the one challenge is there are so many companies coming up with AI as sort of their secret sauce or part of their secret sauce that there's not, like, 10 companies to choose from. There's a thousand companies to choose from in every single industry and sometimes dozens of duplicates in the same kind of, like, solution space. And so, you know, it makes it harder for the investors to know what's right. The other thing is the very thing that is making these companies so attractive is also making the environment change more quickly.

There's a company building an AI-based tool to solve some problem in logistics or shipping. They are undoubtedly using a range of different tools that are also AI-based or enabled, which enables them to go super fast and develop super fast to prototype really fast. So, A, the pace of change is happening so fast and B, like companies are going to get success or fail faster because the cycles are getting faster because AI is making those cycles faster. From everything that I read and when I'm talking to investors, it's challenging to kind of know which horse to bet on in that environment and way more challenging than it's been in the past.

Eric Oak:

I think there's probably a business cycle component to investment up or down, but I think overall, the trend's going to be upwards as people find new use cases for these tools. You know, it's going to be an increasing learning process and people are going to learn what works and what doesn't work. But I would assume that there's going to be use cases ten years from now that we haven't even thought of. So I would assume investment will trend upwards.

Eric Johnson:

Someone at TPM said it's going to be sort of silly to even reference AI at some point. It's going to be like saying your company uses computers, you know, or, you know, like, it's gonna be a given.

Kristen Hallam:

Cool. Alright. So let's play a round of agree or disagree. You need to have clean data for AI to be beneficial. Agree or disagree, Eric Johnson?

Eric Johnson:

Yes. For sure. I don't know if I don't know anyone who could make the argument that at this point, maybe there's some future state where the AI can take the worst data and make it amazing before it then ingests it into the model that it needs to use. But, like, I think at this point, the big challenge is you have something with huge horsepower and you're feeding it, like, you know, sludge instead of high-performance fuel, right? Like, it has to be the right information to base, you know, to feed into a model, to train a model. So it gives you the output that you actually want. But I, I mentioned Eric’s point of view of this because you're the one who's actually, like, living in data every day, much more than I am. So you would know, like, where this falls down a lot more quickly and accurately than I would.

Eric Oak:

Yeah. Well, I 100% agree as well. I did some time as a data engineer, so I'm a little bit biased in this regard, but data is everything. In fact, the AI part probably doesn't matter. Really at this point it's commoditized. You can go and get a model of a Hugging Face and it'll probably do what you want it to do, depending on how much horsepower you want to throw at it. But the data, the data is everything. There is no replacing data.

You are essentially, you know, shooting yourself in the foot if you don't clean and process and kind of transform your data into a form that makes sense for whatever problem you're trying to solve or whatever solution you're using. I think it's, not to use horrible analogies, but, you know, making an AI product or even prompting AI, I like to think of it as kind of like growing a bonsai tree, where you need to make sure that you shape the path that that tree grows, and you need to understand kind of where those branches are going. And then you snip off leaves that you don't want. So if you are making an AI application, there is a potential that your data could send that algorithm into a random tangent that doesn't help you at all. Some of those are very easy to find and some of those are very hard to find depending on what your solution is. But you need to be shaping that solution from, you know, A to Z and really A at that point is data, is data collection, honestly.

Even before the data gets into your enterprise systems, into your warehouse, you need to be understanding where it comes from, why it was created, and how it impacts your model biases and kind of how that model behaves and tries to infer against the world. So, yes, 110%, data is everything.

Kristen Hallam:

I think that's the first Zen reference we've ever had on the Decisive podcast, Eric. The bonsai tree shaping. I like that.

Eric Johnson:

I love that visual. I'm totally going to think of that anytime I think about this topic now. That's great.

Kristen Hallam:

Yeah. We're all gonna steal that from you. Alright.

Eric Oak:

Please do.

Kristen Hallam:

Okay. So agree or disagree, venture capital is poised to ramp up investment into AI-focused startups that are zeroing in on discrete forwarding functions? Eric Johnson, I'll come to you again for that one.

Eric Johnson:

Yeah. This is probably my territory for sure. Yeah. No question about it. I think if you combine there being a lot of capital still looking for a home to find and you take sort of the learnings that the venture capital world has accumulated over the past now coming up on fifteen years of exploring what works and what doesn't in this space. I think, you know, I would broaden it out even from your question. Right? Like, they're going to look for use cases across any type of user, whether it's a port, a shipping line, air freight, shippers, you know, certainly. You mentioned forwarders and forwarders, I think, are probably the biggest target and rightly so because they have to cater to shippers as technology providers, but they also have to run really efficiently and be good partners to the carriers that they're booking space on.

So they have to sort of play three different roles, all of which are tightly wound around technology requirements. So they are a very easy target, I would say, from a startup perspective and rightly so. And, you know, I think a lot of investors see that as an area to really back and hone in on making them as efficient as possible. So, yeah, I do not expect that to slow down anytime soon until, of course, we get to the overhyped AI winter that we discussed earlier.

Kristen Hallam:

There you go.

Eric Johnson:

If you want big, huge three comma or more returns as an investor, you need to be backing companies that completely transform an industry or transform a player in an industry. But the reality is most large enterprises that are involved in international trade or logistics or shipping, the idea of big overhauling and transforming things is a nightmare and something to be avoided at all costs. So you have this sort of friction that the investors want big, expansive companies that touch a whole bunch of different things at these enterprises, and enterprises only want to fix one thing at a time. And so AI is interesting because what we talked about before, if you are a startup and you can quickly scale something up that fixes one specific problem, that sort of helps meet those two problems in the middle. Right? That you can—you're not waiting ten years on a company to fix one very small problem with a relatively small market size. They can fix that problem, get really big in that problem in a couple years, and then figure out what that next step is, whether it's to sell or whether they're big enough on their own to IPO or whatever it is the investors want out of it. So that's, I think, a very interesting element of the AI experiment we're all in right now.

Eric Oak:

Yeah. It's that cold hard truth with all that stuff, especially in shipping, is that goods need to move from A to B. And it's a goods-moving business. You have to move boxes. And that's the only thing that really matters in the end. Either you move boxes or you don't. And really, I would say most BCOs probably don't care about a lot of the internal mechanisms of how that happens, if it's using AI or not. They care about whether the delivery times are good, whether the prices are fair, whether the boxes arrive where they need to arrive.

So I think it's kind of a tough problem to go in and transform an industry that really can't be transformed. It's very hard to transform real aspects of the economy, much harder than virtual aspects. I think that's definitely a shift that needs to be addressed in some of these investment decisions.

Kristen Hallam:

Alright. Well, last one in this agree or disagree round. Agree or disagree, companies investing in AI will be in a better position to navigate labor shortages due to demographic trends? In other words, the aging workforce or people who are looking at different careers other than in the shipping industry. Eric Oak, I'm going to come to you first for this one since you didn't get the other two.

Eric Oak:

No. No worries. I think, I'll punish you for that because I'm going to be very indecisive and potentially tear the question apart. So I think, generally, investment in AI is going to be obviously good in the short term for potentially replacing certain types of work. And in the long term as well, it will replace certain types of work. I think what we're kind of forgetting from the whole AI workforce replacement thing, especially when we think about demographics, is that when we get to the point where there's, you know, let's say a manager and a hundred robots, how do you get that manager? How do you, where do you hire the manager from? What if your process is very specific, you have a very specific manufacturing function, and that manager has to know how those machines work or has to know how your process functions for it to be efficient and to debug those robots? So there definitely needs to be some changes to how companies think about labor and pipelines and things like that. Like, I don't know if that means, you know, you have an intern along with your robots that you can hire managers up into that function.

There definitely will be a benefit to having AI workers, AI essentially bots running certain discrete tasks. Again, you can scale them up. They work all the time. They don't—you don't have to worry about anything other than kind of keeping them well-fed with data. But I think you're going to start to see a different workforce problem emerge. I don't think it's gonna solve the workforce problem. I think it's just going to create a different workforce problem to be, to be annoying about it. There will always be issues with this. It's a fundamental disagreement between, you know, human nature and capital. There's really no getting away from it. There will be new challenges. There will be new things for companies to address. And then, of course, as the way the world goes is that those challenges will then create opportunities for new companies to use AI to fix those challenges. So, hopefully, as we continue into the future, we'll see different things and different aspects of this emerge. But, again, I'd say I agree and disagree at the same time.

Kristen Hallam:

Alright.

Eric Johnson:

This is not the indecisive podcast.

Eric Oak:

Yeah. Oh, that's a good point. Oh, no. I've ruined the whole theme.

Eric Johnson:

Well, I will try. I want to think about this in two ways. So I think the first part is there's sort of parallels in, you know, every transportation mode is essentially about capacity versus demand. Right? I mean, that's the crux of it. If you say that there are labor shortages in certain types of jobs in the shipping, logistics, trade, however you define it, supply chain, broader supply chain worlds. The more AI starts to permeate those worlds, theoretically, less labor will be required to do a certain number of jobs, which releases that labor to theoretically be part of a bigger labor pool. You augment the capacity side because you have less demand in certain areas. And, again, to Eric's point about needing new skills, you've maybe attracted pools of workers that never were even thinking about logistics because it's all like technology now. It's not like a stack of papers that needs to be stamped and put in an envelope and trucked somewhere.

Right? It sounds a little more like a tech job than it does a manual labor job. So, I think the quote-unquote labor shortage or skilled labor shortage that the industry has sort of like been wrestling with for a couple decades now sort of solves itself because a lot of people will not be doing jobs that they do right now. And that frees them up if they like logistics to do something different potentially, if they're able to upskill themselves. Or it taps into new pools that never even thought about logistics before. The other element that I think about is, so there's lots of talk and Eric's talked about it, about this idea of like, you use some sort of AI agent or program or bot as an assistant, to augment your capacity or your team's capacity. So what happens when you have a set of agents or bots and the person who's running that team retires or gets a job at another company? One of the things that is absolutely true about the logistics industry is people have learned from their colleagues and their bosses how things work.

For better or for worse, right? Like, they may have learned some bad habits along the way, but they learned how the industry works. You know, you got to talk to Jim over in Chicago because he's the best guy in trucking. You got to talk to Helen who's in Shanghai because she knows how to get through customs over there. You know, like, those are the things that you learn by having conversations and being around someone. If we sort of move to this, like, really agentic world where it's like teams of bots, there's not going to be that information transfer in the same way to whoever the next group of people that is that are interacting with those bots. So we're going to have to learn how to do the jobs in a different way as well. Not just learn how to interact with the bots, but we're going to have to learn how the crux of logistics, the physical part of it also works. So we don't have a game plan or a blueprint for this. We're just going to have to make it up as we go, I think.

Eric Oak:

I would foresee in, you know, at least ten years from now, or as soon as ten years from now, you know, a whole class of jobs that kind of revolve around this AI mechanic type of thought. You know, why do my trucks keep missing the left turn on Route 5 or whatever?

Eric Johnson:

Yeah.

Eric Oak:

You know, it's like, Okay, you have to go in there, figure out what's happening with this AI. And I don't think it's necessarily, especially as we move forward and, like, actually building models becomes less and less important. I don't think that's a data science role. I think it's like a human—just a human that can very quickly put some things together, tweak some things within the program, and kind of get things moving on.

Eric Johnson:

Knows how to communicate with the program, you know?

Eric Oak:

Yeah. Exactly. This is a communication factor. And I think as much as some of these LLMs are able to talk to people at this point, and I'm sure they'll even get better as the future goes, at a certain point, there's a little bit of X factor that humans provide. So, and I think you don't necessarily need that for every job all the time. And I think we very much have seen that with AI deployments, but at some point, you're going to need that. And so the question is, how do you provide that? How do you train for that? Is it just one guy who you call up when everything goes wrong? How do you compensate that person? How do you train that person? How do you replace that person? How do you find that person? You know? I think there's a thousand different things to think about.

I think we just don't know kind of how that looks, and I'm excited to see kind of the next generation of startups try to solve that problem once they find a problem that they can solve.

Eric Johnson:

Is it like that old, the old, like, seventies apartment superintendent who fixes anything in every apartment? It's like a specialist in nothing, but knew how everything worked, or is it going to be like a team of specialists that, oh, I know how to do this and this and this and this, but not this and this. So, yeah. Very interesting.

Kristen Hallam:

How much of what we've been talking about today is about incremental improvements versus fundamental change?

Eric Johnson:

I'll go first on this one. I think we're right now, we're in a phase of AI sort of replacing things that already exist and maybe making them faster and more efficient or more accurate. Certainly more, you know, like twenty-four hours, as Eric mentioned before, like they don't need a lunch break. It doesn't get bored. It's not on LinkedIn looking for its next job. It's—well, maybe it is. I don't know. If it's programmed that way, it could be, I suppose.

Eric Oak:

Get advanced enough, yeah.

Eric Johnson:

Yeah. So right now, it's definitely about the manner in which the physical side of things is undertaken. Hasn't changed and probably doesn't change all that dramatically because it's limited by physics. But we also haven't seen the sort of accompanying processes that run parallel to the physical supply chain. They haven't really changed all that much. So it's possible that some AI pieced together starts to transform what, like, the job of coordinating and managing freight movement is. But we're not there yet. And it could be one of those situations where it's slow, slow, slow, slow, slow, and then, like, this rapid sort of change without us even realizing how quickly it's happening.

Eric Oak:

Yeah. We're incremental till we're not, and then we look back and go, holy cow, that was actually a crazy change. I think that the crazy idea that popped in my head when you said the AI bot looking for its next job at LinkedIn would be, you know, if I'm a company, I create and deploy a bunch of generalist bots. Let's say I have 10,000 generalist bots that run on my hardware. I then have an auction for their labor every day. I bid them out to different companies. So if a company wants, you know, 10,000 bots that day, they have to bid and buy all my bots.

Maybe a company just wants two bots. You do this, like, dynamic labor auction of AI-based labor. Yeah, that can be every day or even, like, every minute, you know. Hey, you know, my company comes in, bids, it's sent more for that time. It gets deallocated from the lowest person who then has to scramble, and I can imagine, like, these crazy sci-fi scenarios.

Eric Johnson:

Permutations are crazy. I mean, I've joked with a couple of people that, you know, if we're, like, essentially moving to a situation where it's just agents talking to agents, then I think that starts to take place really fast. Because then there's, like, no human emotion or culture in that sort of interaction anymore. And I was sort of joking, well, like, I hope these agents still are programmed to, like, buy stuff on Amazon every so often, so there's still physical goods to move. But, you know, who knows?

Eric Oak:

Yeah. No, absolutely. I think that just comes back to—I think it's incremental until it's not. I think there is the potential, obviously, for groundbreaking fundamental change. But personally, and I'm a bit of a curmudgeon, I haven't seen that in anything so far—an actual, like, fundamentally groundbreaking change to anything from AI. I think it's really cool that you can make a chatbot that you can have the chatbot do all sorts of cool things. You have these AI systems to decide things, but it's all enhancing the human workforce at this point.

Kristen Hallam:

Well, this has been a really great conversation, and I could keep it going, but it's time for us to wrap up. Any final thoughts for our listeners?

Eric Oak:

Sure. Yeah. I think, generally, as we think about how AI is changing the workforce, the world, business, logistics, shipping, etcetera, I think we have to just make sure we keep in mind what AI is good at and what humans are good at. So AI is good at, again, being up all the time, you know, finding patterns that are in immense swaths of data. Humans are good at adapting, at fixing problems, you know, at interacting with anything. So just kind of keep that in mind, like, there is—I'm gonna be very—maybe humanist is not the right word, but there will be a human in the loop all the time. So how do you prepare yourself for that eventuality where humans have to do different things, different types of jobs in this kind of futuristic AI world? And kind of how do you prepare yourself for that is, I think, the biggest question of probably the next decade of workforce training and stuff like that.

But again, humans have a role. Humans are the ones that can adapt instantly, provide that kind of X factor, and get around things. Where there is a world where AI bots are talking to AI bots, there are going to be humans that are trying to break those patterns. So there will always be hackers. There will always be people on the fringes that are trying to break those patterns, which means there will always need to be humans that are anticipating those other humans' actions, you know, and then just multiply that across, again, a web of cyberspace and potentially in this sci-fi world. So just keep that in mind, like, there will always be humans. Like, this is not a doomerism or kind of a human negative idea. AI will change and change things potentially very drastically, but I think humans will always have a place in shipping and in work on this planet to get really, really up in the air about it.

Eric Johnson:

Yeah. I think I would just say everything in moderation. Right? Like, you can't get too carried away with where things are going. You also can't bury your head in the sand and pretend like nothing is happening. So you have to be really open-minded about what's going on without getting distracted by things that are interesting, but ultimately not all that important for you to understand. Right? Like, I think big picture, you have to be sort of efficient at understanding what this stuff is about, but you don't need to be a genius or a wizard about it to understand where your place is. Like, I mean, I think every day I'm hearing more and more about tools that sort of put power in the hands of people like me who could not code a single line. So I think some of this stuff gets leapfrogged.

Like, some of the understanding that we might have ordinarily needed to do if we really needed or wanted to understand how AI works is going to get leapfrogged by the technology that sort of abstracts that from us anyway. So, yeah, just try to stay measured about this and not get too paranoid or too delusional about it. Because generally speaking, either side—if you're on either side, you're in a bad place. So try to stay in the middle of the road, I would say.

Eric Oak:

Well said. Yeah. Absolutely.

Kristen Hallam:

Don't panic. That's very Hitchhiker's Guide to the Galaxy right there.

Eric Johnson:

Yeah. And a good Coldplay song at that. So...

Kristen Hallam:

There you go. Well, thank you so much, Eric and Eric, for taking the time to share your insights with us. That wraps up this episode of the Decisive Podcast. Until next time. Thank you for listening to the Decisive Podcast from S&P Global. Please subscribe and join us for next week's episode. Until then, stay curious and stay informed.