There is a growing chorus among critics of AI & Machine Learning that the moment of reckoning for the AI community is here. While the specifics differ, the common theme in the criticism follows this argument: AI & ML models rely on the past for predicting pretty much everything – demand forecast, customer segmentation, price optimization and supply chain & logistic optimization. Now that the Corona pandemic has affected both the supply and the demand sides of the market, these AI & ML models have become irrelevant. Some go even further to suggest these approaches are now as good as useless.
While the concern itself is relevant, the conclusions inferred are at best from a place of cynicism or worst from naive understanding of AI & ML models. First a disclaimer: Any individual who claims AI & ML models would be able to predict with any reasonable accuracy the impact of such once-in-a-century event on supply & demand is doing a disservice to AI & ML community.
Now to address the criticism. Yes, these models rely on historical precedents to infer impact that are statistically sound. For example, the impact of promotions and annual events like Easter, Memorial Day, Christmas, Diwali, Ramzan etc are based on historical data and the current prevailing market conditions. However, to claim the models are now irrelevant is incorrect. There are evolving markets conditions where these models would still be useful. Three immediate use cases are highlighted
Summary – Careful use of historical data, especially the few weeks before the onset of the COVID pandemic, can help businesses adapt better to the situation
Published on April 24, 2020