As we continue to move into an era of Artificial Intelligence (AI) that can predict what will occur in the future, our methods of predicting these events have changed significantly. Whether you want to predict tomorrow’s weather, estimate your energy use for the next day, follow the way a disease spreads, or better understand how customers behave. AI-based predictive models are now being run each and every day. The fact that AI can take in very large datasets at incredible speeds allows it to identify patterns and make quick predictions. And so, another issue arises: “Is AI enabling us to predict things faster but ultimately just as well?”

For years, a number of different methodologies have been used to forecast, including forecasts based on historical data and the opinions of subject matter experts. The use of past data combined with the application of statistical models was one of the earliest ways in which forecasting was completed. Today, most companies utilize some form of forecasting methodology in their organization. In addition to using the opinions of subject-matter experts, there is also a wide variety of other methodologies that can be used, including the use of computerized software programs that are capable of utilizing large amounts of past data to make an estimate of what will occur in the future. While these methodologies remain widely popular today, they may not work as well in today’s environment, where large volumes of new information become available to decision-makers on a daily basis.

The way has opened for AI-based forecasting methods, which can use the characteristics of the data as well as relationships that were unknown prior to the application of AI-based methods. Traditional forecasting models will always need manual input to update on a continuous basis, while AI-based forecasting models are capable of continuing to learn and update from other types of information available. As such, there are many applications where AI has improved the overall quality of forecasting across industries.

In the area of health care, AI is being used to track disease outbreaks and provide hospitals with predictive resource needs. In the area of transportation, AI is also being utilized to optimize traffic flow and create the most efficient routes for traveling. The areas of energy generation and distribution are another example where AI is improving the forecast for future energy demand and allowing utilities to integrate renewable energy resources into existing grid operations. Finally, agencies providing aid during natural disasters are increasingly using AI to improve early warnings and preparedness for emergencies.

While speed can certainly provide a competitive advantage in forecasting, being faster does not guarantee you are going to be accurate. As stated earlier, regardless of whether your forecast was produced by a highly skilled forecaster using traditional methods or through the use of an AI tool, if the input data to that method or model is inaccurate or incomplete, then the output will also be inaccurate. This example illustrates one of the major tenets of data science, known as GIGO, which stands for “Garbage In Garbage Out”. For this reason, collecting and managing quality data is equally important as improving the sophistication of the algorithms used in creating forecasts.

Even when AI models may be trained on large amounts of complete and correct data, there are many reasons why forecasting will always remain difficult. This is due to the inherent unpredictability of the real world. Some events have occurred in recent history (such as a pandemic or an economic disruption) that have no precedent, in which we cannot say with certainty what could happen next. Additionally, even though an AI model can find patterns and generate predictions about probable outcomes, it has no knowledge of the larger context of these events and certainly does not possess human judgment. Therefore, it will continue to rely on humans who have specific expertise in certain areas for their interpretation of the AI-generated forecast.

This is why AI should be viewed as a powerful decision-support tool rather than a replacement for human expertise. AI excels at analyzing massive amounts of information, identifying complex patterns, and generating multiple forecasting scenarios within a short period. Human experts, however, contribute qualities that AI cannot easily replicate, including critical thinking, ethical judgment, creativity, and contextual awareness. When both AI and human intelligence complement one another, they create the conditions for forecasts to occur not just quickly but also meaningfully, reliably, and actionably. In a nutshell, “The true strength of AI for forecasting does not reside in the machine itself, but in the knowledge and wisdom of the individuals who use it.”

Prepared by: Dr. Mas Omar Mas Rosemal Hakim, Senior Lecturer, Faculty of Artificial Intelligence (FAI), Universiti Teknologi Malaysia (UTM)