Four key ingredients are needed before automating AI systems.

Deborah Yao, Editor

April 6, 2022

3 Min Read

Four key ingredients are needed before automating AI systems.

Consumer products giant Procter & Gamble is no stranger to innovation, from inventing the first foaming soap bar, Ivory, in 1879 to launching an AI-powered electric toothbrush in 2019 that tracks how people brush.

CIO Vittorio Crettella said the innovation continues. “AI is embedded in everything we do and also embedded increasingly in all our products,” he said at the ScaleUp:AI conference in New York.

He said AI is being used to optimize operations and has saved the company millions of dollars. For example, when P&G wants to phase out a product and introduce a new one in its place, it has to manage the inventory levels to strike the balance between not being out of stock and not having too much inventory.

This AI model was used for more than 65% of such product changes at P&G and is saving the company $60 million a year. Crettella said there are “hundreds” of such AI application examples at the company – in marketing, audience-building, programmatic ads, automated bidding, organizing activities in stores, and others.

Scaling AI

To unlock AI’s true impact on ROI, it must scale across the company. To do that at a big Fortune 50 company like P&G, AI must be automated. But first, Crettella said four key ingredients must be in place: data, talent, platforms and trust.

1. Make sure the data is usable.

Many large companies have legacy data architectures that are built around functions and typically siloed. Crettella said that P&G created a data lake using aligned taxonomies to make sure data is usable no matter where it comes from.

2. Skilled talent is extremely important.

P&G has more than 200 internal data scientists and hundreds of data and machine learning engineers, as well as business analysts and business players that use this data to drive decisions and make predictions using machine learning.

The company uses a mix of 50% internal talent and 50% external experts from partner companies. It tap external talent to manage spikes in demand.

3. Develop platforms that can be deployed in different use cases.

Develop algorithms that can be deployed in a broad array of businesses. P&G has an initiative called Neighborhood Analytics that helps the company understand on a granular, local level how consumers shop and what they need. It can be used for an array of use cases, whether sales or marketing, supply chain and the like.

4. Develop trust in the AI’s decision-making

Key business players, from middle management to top executives, must learn to trust the machine. It is a big culture change since people with years of experience are used to making decisions based on what they know. Now they have to trust an algorithm’s decisions, which is an easy mindset shift.

Adopting explainable AI by itself is not enough. It has to be complemented with educating people in the company about how algorithms work. Also, there has to be support from the top. There should be room for experimentation as well to adopt new ways of working, but watch out for possible bias and algorithm drift.

“You can have the best model, the best data, but in the end, you want … your decision makers to trust the algorithms,” Crettella said.

About the Author(s)

Deborah Yao

Editor

Deborah Yao runs the day-to-day operations of AI Business. She is a Stanford grad who has worked at Amazon, Wharton School and Associated Press.

Keep up with the ever-evolving AI landscape
Unlock exclusive AI content by subscribing to our newsletter!!

You May Also Like