Year-End Thoughts About the Exhilarating World of AI and Optimization

Thursday, December 15, 2022

This post is adapted from Irv’s appearance on the Alldus “AI in Action” podcast series.

These are truly exciting times for advanced analytics practitioners, whether you are launching your career or continuing to develop and implement solutions.

Reaping the Benefits of Improved Performance

In the 1980’s when I was a grad student, I used a minicomputer to solve a difficult linear programming problem and it would take 24 hours to solve. Today, we can solve the same problem on a laptop in less than a second. While computational hardware has certainly improved over the past 30+ years, the underlying algorithmic technology in optimization, which I was part of as an early member of the CPLEX team, has improved by magnitudes more. There is also much more data available. As a result, we are helping executives make much better decisions in more real time scenarios.

Traditionally, optimization has been used to support long-term strategic decisions, as when companies decide where they want to build distribution centers for shipping their products to their customers. That kind of network optimization problem isn’t used to make a decision about building a new distribution center each day, as it can take a year or more to actually build a new facility.

Today, optimization has emphatically moved from mainly the strategic arena into the tactical and operational arenas. For example, E-commerce companies need to fulfill customer orders that are coming in 24/7. We worked with one leading E-commerce retailer where, every few minutes, 500-1,000 orders are directed to distribution centers to minimize overall shipping costs while balancing work across the distribution centers. The supporting operational optimization solution is necessarily fast and robust, requiring no human intervention.

More and more executives will recognize that there are applications of optimization technology to make better decisions about allocating limited resources, with more decisions happening in real time. We are seeing optimization applications for complex workforce scheduling in healthcare, government, and in businesses.

We are currently helping a regulatory organization optimize staff scheduling, an activity that for decades has been done by a team using spreadsheets. Historically, it takes about a month to create the master annual schedule. As things change during the year, the schedule must be modified, with the processing managed by a dedicated team. A single event, such as a worker leaving the organization, creates ripple effects that requires a new schedule to be created, with the property that the new schedule contains small modifications to the original planned schedule. We are working to automate and optimize the scheduling process with an application that allows executives to handle these modifications in real time. I believe we will see more real‑time applications for scheduling and for allocating resources, so that executives are increasingly making decisions leveraged by optimization technology.

Beyond optimization, in other AI solution areas, rapid response is also required. Many executives ask, “How fast can I make critical decisions? How fast can I get information?” We all use our mobile phones to quickly get information about anything, and we then expect that performance in business environments. The technology and the applications are evolving to support more real‑time decision‑making because the users simply won’t wait hours to make a business decision. If the nature of the problem doesn’t allow for a provable optimal solution to be immediately delivered, recommending a good solution, computed quickly, that determines the next best thing to do now with limited resources, can in many cases be highly valuable.

The Joy of Consulting

I regularly discuss new levels of performance, the need for speed, and the red-hot popularity of AI and Data Science with fellow practitioners, leaders of commercial optimization solver companies, and the INFORMS community of academics and industry executives. In our various roles, we are each dealing with opportunities and challenges in a world that is increasingly demanding our expertise, solutions, and technologies.

Working as a consultant, I continually get exposed to different industries and different kinds of problems, which I find stimulating and enjoyable. There is constant learning, not just about the technology, but also about the kinds of applications of the technologies.

As a practitioner, you meet interesting people in different walks of life. Their reactions vary, and you can develop very human relationships. When we work with our clients to develop a vision of a solution to their business problem, we build social skills and relate better to people.

Travel is picking up post-pandemic and our consultants are increasingly visiting our clients onsite. We see how things differ from industry to industry, how executives make decisions, and how they move their businesses forward. Experiencing a variety of organizations and business problems helps practitioners improve both their “hard” and “soft” skills. They can grow and become leaders because they are learning how to interact and work with people, which is necessary to develop successful applications and make a positive impact.

Through the INFORMS Edelman Competition, I am exposed on an annual basis to the spectacular results that AI and optimization yield. The ever-improving technology is there for practitioners around the world to thrive and to benefit businesses and society. Here’s to more exciting work and success in 2023!