Tech Trends 2018: AI is Ready for Prime Time

Read the full "Tech Trends 2018" report

Q: Artificial intelligence has been identified as “critical” to future success. What are the primary benefits it’s bringing to consumer goods manufacturers? 
REILLY: It’s all about shifting time from “causation” to “ideation.” Virtually every leader we talk with in the consumer goods world struggles with this. They have barely enough time to determine what happened in the last reporting period and not enough time to figure out what to do about it. They need to get creative with retail partners, brand strategies, media mix — the list goes on. It takes much too long to understand the changing landscape, which hamstrings efforts to do what humans do really well: get creative and drive actions to deliver growth. Fortunately, our industry benefits from the trifecta of AI skills: huge volumes of data that change over time, with many unknown underlying factors and complex effects of every decision.

Q: In our Tech Trends survey, insights/analytics was identified as a priority area for implementation. What should be the key objectives here?
REILLY: To focus on the desired business result, which is often bottom-line growth. Like a newspaper that signals items worthy of attention, an insights “feed” should relieve a brand manager from ever having to browse data directly. Reams of KPI reports don’t have to be reviewed when AI is already scouring all of that information in real time for the important events that need attention. 

Machines can voraciously take in the volume of data produced by CG businesses and point out important events that a business expert should review. Over time, the machine learns to point out an insight and what action to recommend for it.

Q: Where else in the enterprise should companies be placing an emphasis right now? 
REILLY: I’d emphasize the parts of the business that drive revenue: sales, category management, brand management, etc. Modeling all of the moving parts of a CG business involves vast amounts of data and computing power at a scale that is only now coming into focus. Today’s market mix models will give way to demand simulators tracking and responding in real time.

Q: Despite the recognition of AI’s potential, a lot of companies have been slow at implementation. How do companies move beyond the current obstacles? 
REILLY: Another thing Ng talks about is the idea that “anything a human can do in one second, a machine can either do now, or will soon be able to do.” He suggests looking for things that humans can do in one second as potential targets for automation.

One example he gives is the task of evaluating a security video. A human can evaluate the current scene in security footage in one second. How many people, where are they headed? Are they carrying anything? Analyzing the health of a brand is similar in many ways: Is share increasing or decreasing? Penetrations? Value? Volume? Which brand has the highest increase in share? Companies should look for these opportunities and start experimenting.