Visit Our Sponsors
Spencer Askew, chief executive officer with Teknowlogi, offers his definition of the term "software intelligence," and discusses how it enables new developments in artificial intelligence and machine learning.
SCB: How would you define “software intelligence”?
Askew: As we look across the supply chain, we recognize that being able to harness decades of knowledge into the DNA of a software platform, to help organizations move forward with the next-generation supply chain, has become a critical path for organizations.
SCB: How do you go from software as a series of codes to this whole concept of intelligence? At what point does the system actually become intelligent?
Askew: There are tools around machine learning, but we also need to be able to transfer head knowledge from an industry expert into the DNA of the software, to help the company make better and more proactive business decisions.
SCB: Is the term “head knowledge” synonymous with expert systems, which have been a key element in the development of artificial intelligence over the years?
Askew: Absolutely. The application of intelligence is contextualizing information that resides within organizations, and converting it to tasks and recommendations that benefit the organization's profitability. We call it a logistics expert system, or LES. We want everybody in the industry to think beyond a WMS [warehouse management system] or TMS [transportation management system], with bits and bytes of functionality.
SCB: At what point does the system start making decisions, instead of just recommendations for humans to execute?
Askew: It's like climbing a mountain. Part of the issue in A.I. and machine learning is the trust factor. A lot of times organizations don't want to dive in head first. You want to combine user intelligence with machine intelligence – we use the term applied intelligence – to move from predictive to prescriptive capabilities.
SCB: It’s interesting you used the word trust. Does the human user understand how the system came to a particular conclusion? Or is there just a “black box” in the middle?
Askew: The user doesn't necessarily know how the machine came to the prescriptive recommendation, but the user cares about the results. It’s about trust initially, but what you're really evaluating is whether the machine produced better business outcomes. A user can understand and comprehend that. It's kind of like digging a ditch. Would you rather do it with a shovel, or would you like a machine to help you? You don't necessarily know how all the components work, but if you can see it working and you can see the results, you’ll come to trust the machine to make better business decisions on the organization's behalf.
SCB: Does this lead us to assume that in the future, the percentage of decisions made by the machine versus those by humans is going to increase?
Askew: We see that evolution. It’s a mindset shift to change the trajectory of how organizations across the supply chain evaluate their systems. They need to ask themselves, Is my software more intelligent than my users? We see a future where the machine, the software and the platforms are able to make better decisions. When you can pool all of the data from a WMS, TMS and ERP [enterprise resource planning system], and have an intelligent application analyze that data, advise the user, and help execute decisions, then that's a world where I think we're going to see some gained efficiencies.
Enjoy curated articles directly to your inbox.