Artificial intelligence (AI) solutions are built using lots and lots of data, most of it historical. When hundreds of pictures of cats are shown to an AI system, it learns what a cat is and what isnt a cat. AI systems in retail learn from historical data what are, for instance, the buying patterns in one store, how it changes over months/seasons, how it is different in another store in another location. That enables it to make predictions.

When something like Covid-19 happens and changes many historical patterns dramatically, AI systems are as unprepared as humans. And they require fresh training, rewriting of the algorithms.

Bengaluru-based AI solutions provider Manthan works with many global retailers. One of the companys fashion retail customers in the US saw almost all its 700 stores shutting down at the end of March. The companys chief product officer Sameer Narula says they had to rework their algorithms to handle the short-term and longer-term impact of the event. For the short-term, we had to quickly build new models, within weeks, which could use e-commerce data to identify changing patterns and trends and mash it with geo-spatial data to predict the demand in stores when they reopened, he says.

The pandemic event-specific data is still limited and evolving to be useful in calibrating the longer-term impact of it

Sameer Narula, chief product officer, Manthan

For the longer term, its more difficult. Narula says the pandemic event-specific data is still limited and evolving to be useful in calibrating the longer-term impact of it. Weve been working on incorporating several new micro and macro economic signals like consumer sentiment index, economic optimism index into our demand forecasting models to bring in more permanent and generic adaptability to deal with the impact of such kind of exceptional events in future, he says.

We used AI to check for compliance of mask usage by delivery partners, and to stop delivery of non-essential items

Dale Vaz, head of engineering and data science, Swiggy

Read more:

Not just humans, AI too is getting used to the new normal - Gadgets Now