Artificial intelligence is finding a foothold in multiple businesses, and the supply chain is no exception. But logistics providers have yet to fully realize its potential. Questions remain: Which processes can be optimized by the technology? For what kind of decisions is A.I. best suited? Will humans continue to play a role, or will automated “expert systems” make them redundant? In this conversation, excerpted from an episode of The SupplyChainBrain Podcast, we speak with Spencer Askew, founder and chief executive officer of Teknowlogi, which pioneered the Logistics Expert System. He discusses the real potential of A.I. and machine learning for carriers, logistics service providers, and other key industry players.

Q: What is your definition of artificial intelligence?

Askew: We think about A.I. for transportation and logistics in terms of the expertise that exists across our industry. It’s having the ability to package that knowledge and consulting intelligence into a machine learning-based platform. We can take advantage of the same capabilities that have been successfully deployed in other verticals. We can analyze decisions to execute change throughout the organization.

Q: On one hand, you’re describing A.I. as a so-called “expert system,” that has at its fingertips all possible information. It appears to be mirroring the old-line human expert in its ability to come up with the right conclusion at the right time. At the same time, you seem to be suggesting that once you have that capability residing within the machine, you can conduct analyses and make conclusions that surpass mere human expertise.

Askew: That’s correct. Once you’ve gone beyond the analytical component, there’s an advisory layer that’s taking into consideration the consulting intelligence of an organization’s experts. It’s making decisions that fall into line with best practices across the logistics industry. It can execute on the most statistically correct decision, based on current analysis as well as historical data. The system becomes more and more relevant, and the execution layer becomes increasingly correct. As a result, an organization can operate with the most perfect version of itself at all times.

Q: What got you interested in the subject of A.I. for logistics?

Askew: We’ve been contemplating A.I. and machine learning for many years. My background is in consulting for both asset-based and non-asset carriers, as well as technology providers and shippers. We’ve seen every layer of the industry – its challenges, frustrations and opportunities. One thing that becomes clear is that there’s a tremendous amount of expertise that exists within each layer. We came to realize that A.I. platforms could become an extension of an organization and its team members. The industry has evolved to the point where it is no longer humanly possible to deal with big data and analyze possible scenarios on an ongoing basis. The challenge becomes how the organization can focus on all items at all times, so that it can become great at everything, and not be limited by human capabilities.

Q: You talk about how A.I. can address the problem of an aging workforce in the logistics sector. Can you explain what you mean?

Askew: The logistics industry today is facing the challenge of succession planning due to an aging workforce. It needs to capture the resident expertise and tribal knowledge that goes along with managing shipment lifecycles, routing freight and optimizing lanes. There’s a tremendous amount of information that industry experts have in their heads. Unfortunately, when they retire or take a position with another company, that knowledge and expertise walks out the door. By deploying A.I., and infusing legacy technology with expert systems, all of that knowledge gets captured and retained. The A.I. component can help to enable the success of whoever is taking over that position – someone who might not have the 35 or 40 years of experience of their predecessor. In most cases, those are really big shoes to fill.

Q: You say “taking over” the previous position, but given the growth of A.I. in this sector, you have to wonder whether that position is the same job anymore. In bringing new and younger people into the organization, don’t you want them to be doing different things, and possess different skills than their predecessors?

Askew: Initially, we see the deployment of A.I. as gaining the ability to enable current staff members to focus on more important, bigger-ticket items. Statistically speaking, e-mail chews up more than 30 percent of everyone’s day. If you have A.I. built into your platform, it can take over some of those mundane tasks that take precious time out of the day. Your more experienced resources can focus on bigger, board-level topics. We see the adoption of A.I. as resulting in capacity enhancement, not capacity replacement. It’s about getting smart people to focus on more complex problems, and solve them more quickly.

Q: So you don’t think A.I. will necessarily lead to a net reduction in the number of human beings in the organization?

Askew: That’s not something we can predict at this stage. We definitely see resource capacity and productivity improvements. For an organization with a staff of 1,000, we see a world where it could potentially double the business without having to add headcount. As companies adopt new technological capabilities, either they’re repositioning staff to focus on more important items, or have the capability to grow without being concerned about resource constraints, a tight job market, or an aging workforce. We see the future of the A.I.-enabled logistics industry as super bright, allowing organizations to do way more with equal to or less than they currently have in place.

Q: But at the same time you’re moving humans into higher-level tasks, it would seem that you’d be looking for people with a different set of skills. You need data scientists, analytics specialists, coders. They’re not the same people who were on the assembly line, or processing packages in a warehouse. Are there challenges in finding these higher-trained people who can support your A.I. systems when you bring them in?

Askew: It’s a matter of working with the right partner. It’s a build-or-lease methodology. We don’t really see the industry experiencing challenges from having to look for a different skill set. Most organizations continue to rely on skills and support from the software vendors that they’re doing business with. We’re not minimizing the potential need down the road for businesses to think about having certain experts on their staffs, such as data scientists. But there will also be system capabilities that will do the data-science work, without companies having to staff those specific positions.

Q: Can you be more specific about how you think A.I. applies to logistics? What particular functions might A.I. help you to manage?

Askew: You would need to break that down by department level, topic, and goals and objectives. Depending on where you sit within the vertical of the logistics industry – if you’re an asset or non-asset carrier, for example – a big topic for you might be cash flow or accounting. Relate that back to accounts receivable – organizations have A.R. staff members who collect money from shippers. Thinking about the application of A.I. and machine learning in that area, one of the things that we recognize is there are a lot of factors contributing to whether or not customers pay within terms. The reason could be as simple as the method and frequency by which the A.R. team is reaching out to the customer to collect money. Customers paying on time might be receiving three phones calls and three e-mails on a weekly basis. Whereas those who are not paying within terms might only be getting one e-mail or phone call. The application would look across thousands of customers, and uncover the root causes that affect timely or untimely payment.

There are tens of thousands of possible scenarios under that one business topic, with A.I. able to statistically analyze each one. Such a capability doesn’t tackle the entire supply chain, but as you start addressing each business department and topic, you can move beyond that to achieve density maximization and improve operating ratios for the carrier. Many carriers have different divisions and systems that are not even talking with one another. So margin management is obviously a big topic, where the deployment of A.I. can be of benefit.

Q: Where are we today in the application of A.I. to logistics? Does A.I. dictate to a company decisions on density, route optimization, carrier selection and packaging, or does it advise on them, and leave it up to a human to make the ultimate conclusion?

Askew: From our perspective, the design has to do both. We absolutely look at A.I. as needing to build trust within any organization that has deployed it. When you hire an expert human resource, the resume may say all the right things. But you still have to go through a learning and trust curve. The same goes for A.I. As the new technology analyzes and advises, the ultimate decision initially remains up to the organization. At the same time, the platform is monitoring all of those decisions, and tracking whether or not they were correct.

A.I. can’t just jump into our industry and start executing on all the major decision points. Eventually, though, we see organizations hitting the “easy” button after realizing that A.I. has been correct on the last 100 recommendations. At that point, if you bring me another 100, I’m going to give you permission to execute on all of them. You’ll still report back to me on how we did, but there’s going to be that trust curve.

Q: Do you believe the message of A.I. is getting through to logistics providers?

Askew: Our industry isn’t talking enough about A.I. We feel that it’s our responsibility to introduce it to the logistics business in a big way.

Click here to download the podcast

Resource Link:
Teknowlogi