How AI Planning Helps Organizations Make Better Decisions

If only we had a crystal ball to foresee what lies ahead and the impact on business. Despite multiple warnings of a global pandemic, humankind was ill-prepared (no pun intended) for the outbreak that was to come, and many businesses were scrambling to meet demand while others were struggling to stay afloat.

Planning is bringing the future into the present so that you can do something about it now.

– Alan Lakein

However, hindsight is 20/20, and hopefully, we learned, and continue learning, from our preparation shortcomings to prevent history from repeating itself. If nothing else, the pandemic illuminated the need for a better way to analyze patterns, assess the conditions and respond swiftly to problems. To that end, effective AI planning can give an organization critical vision that humans might otherwise miss.

What is AI Planning?

Simply stated, AI planning is about using computer algorithms to plan ahead. The process involves identifying what events are forthcoming and the actions that need to be taken to counterbalance the predicted event or problem and achieve a specific goal. Planning takes a business from its current state to the goal state.

The benefits of artificial intelligence planning are many. For example, Artificial Intelligence can help business leaders make better decisions, product recommendations, improve process accuracy, and highlight anomalies. And when combined with human know-how, artificial intelligence can make companies, schools, government, hospitals, and humans in general better.

Getting a machine to understand real-life problems is an ambitious challenge but essential to weaving AI into our daily lives. Imagine a personal assistant preparing your weekly shopping list or planning your annual vacation. The tasks would require a machine to understand your usage patterns, personal preferences, and help you plan for the future, such as booking travel and a hotel. The same is true in business applications— AI planning is about enabling modeling outputs to “see” into the future and increase situational awareness so that humans can respond more strategically under new conditions or in a crisis scenario. The fight against COVID-19 is an excellent example.

Practical Applications of AI Planning Tools

Remember when COVID-19 first hit your community and retailers were out of basic necessities, like toilet paper, hand sanitizer, and even chicken wings? As you walked through your local grocer, it was common to see shelves empty and everyday staples out of stock. Understandable — existing retail ordering tools proved worthless during the pandemic, and inventory management was a challenge.

During this time, consumer usage patterns also changed. Restaurant closures forced people to cook from home. Sporting event cancellations led people to develop other creative options, like increased finger food and adult beverage consumption to stay entertained during the temporary lockdowns. In short, businesses had to relearn fundamental activities, like which locations to open, how to staff, and what to put into their ordering systems to respond to the evolving problems. It was a period of blind navigation, trial, and error.

AI planning could have helped retailers see into the future and prevent the resulting shortages. For example, many researchers, academic institutions, and government agencies used artificial intelligence and LUCD’s infectious disease modeling tool, Avicenna, to generate actionable insights into the predicted spread of COVID-19. The model enabled organizations to predict where outbreaks would occur and make adjustments, including social distancing guidelines, school closures, and work-from-home policies to prevent further infection. With as high as a 95% accuracy rate, Avicenna has also helped commercial leaders better prepare for local and regional impacts, such as usage demands, supply chain disruptions, and workforce planning.

We previously relied on only what we could see in business — the equivalent of driving while looking only in your rearview mirror, your front window blocked, and no headlights. AI Planning and Avicenna transformed the old way of driving into a significantly safer model, with enhanced driver displays, LED lights, and sensors, equipping drivers to see what’s ahead and respond more strategically.

How AI Planning Works

As humans, our actions are generally based on achieving a specific goal — we have an objective in mind and take the necessary steps to reach it. Therefore, planning is the reasoning side of acting. The same is true in planning. AI planning is all about deciding the actions that will be performed by the artificial intelligence system and the functioning of the system on its own in domain-independent situations.

As an example, the goal with Avicenna was to prevent the spread of COVID-19. The model was trained with years of complex datasets like population, individual interaction research, and prior patterns of infectious diseases to create a mathematical representation of disease transmission. With data-driven insights, government leaders have been able to adjust actions and implement proactive procedures to slow COVID’s spread. The model can also be adjusted for different scenarios, as the underlying engine is nearly limitless.

AI planning starts with a domain description, an action specification, and a goal description. Thus, the plan is a sequence of actions based on a set of preconditions resulting in effects that can be either positive or negative. Planning can either be classical or non-classical.

Classical Planning

Classical planning is the simplest form of AI planning, used when the environment is fully observable, deterministic, and static. Three tasks are performed in classical planning:

  • Planning based on an identified problem
  • Acting based on the decided actions that should be taken
  • Learning from new data

Additionally, Planning Domain Definition Language (PDDL) is used to describe the four basic things needed in a search problem: initial state, actions, result, and goal. Air cargo transport is an example of classical planning and PDDL. The problem is based on loading and unloading cargo, flying it from one place to another, and the environment is observable and static.

Non-Classical Planning

Non-classical planning is when the environment is only partially observable and non-deterministic, meaning the current state and chosen action cannot completely determine the next state of the environment. Non-deterministic actions signify more than one outcome is possible, as we do not have total certainty about what will happen in real life once we start executing the actions of a given plan. As such, non-deterministic actions enable us to be prepared for contingencies.

Conclusion

COVID-19 accelerated how organizations use artificial intelligence. According to IBM’s Global AI Adoption Index 2021, nearly three quarters of companies are now using or exploring the use of AI. The increased adoption is largely driven by the repercussions of the pandemic, general business needs, and as AI tools become more accessible. Similarly, Gartner research reports 50% of supply chain organizations will invest in AI planning and advanced analytics in an effort to make better and more informed decisions faster.

COVID-19 was a difficult time for our country and world economy. But with AI planning, executives can now use this data to understand the course of events, the business impact of the pandemic, and more importantly, what is likely to happen in the future.



Subscribe to the Latest Insight

By clicking "Get the Updates" you are agreeing to the Terms of Use and Privacy Policy.

Achieve AI at Scale

Read our whitepaper to discover how BOSS.AI can help your organization's AI initiatives succeed quickly and easily, at scale

Popular Posts

>
Success message!
Warning message!
Error message!