Artificial Intelligence Gets Real for Big Firms
Siemens, Mastercard, and John Deere are using AI as one more tool for competitive advantage
By Beth Stackpole, MIT Sloan
As artificial intelligence continues to move into the mainstream, companies are combining AI and big data to build and design better products, react faster to changing market conditions, and protect consumers from fraud.
According to experts at EmTech Digital, MIT Technology Review’s annual event on artificial intelligence, big data plus AI creates a foundation for more intelligent products and services — ones that initiate maintenance procedures before something breaks, perform more precise operations, or automatically recalibrate resources to meet changing demand and usage patterns.
While AI and big data pave the way for such evolutionary use cases, the pair do not constitute a business strategy on their own accord. “The question is how do you use AI right or use it wisely,” said panelist Ed McLaughlin, president of operations and technology for Mastercard.
“The biggest lesson learned is how to take these powerful tools and start backward from the problem,” McLaughlin said. “What are the things you’re trying to solve for, and how can you apply these new tools and techniques to solve it better?”
In various EmTech conference tracks, experts outlined use cases where firms have effectively embedded AI into complex processes and scenarios to solve real-world business and social problems.
Here are three examples from Siemens, Mastercard, and John Deere:
Design, development, and manufacturing
While it’s not yet possible to get Alexa or any other AI-powered digital assistant to pump out the perfect drone design and queue it up for cost-efficient manufacturing, that’s ultimately the direction as the technologies mature over the next decade, said Stefan Jockusch, vice president of strategy for Siemens Digital Industries Software.
Industry players like Siemens have already taken steps to make this vision a reality. Consider AI-infused generative design features now common in some engineering software: Engineers can specify critical design and cost parameters such as weight or performance characteristics, and the software automatically explores the design space, quickly coming up with a range of options that a typical human couldn’t ideate on their own.
By automating design and engineering tasks, Siemens customers, among others, are already seeing notable results, including greatly reduced manufacturing costs and improved product performance, Jockusch said.
The ability to churn out better designs faster is especially important in markets where discerning buyers want more customized choices, yet don’t necessarily want to pay more for the privilege. There’s also volatility in the economy, exacerbated by COVID-19 and the subsequent supply chain breakdowns, that require companies to be able to shift gears quickly.
“What we have seen over the last year or so is that the winners are usually the ones that are very fast at adapting to new situations, including quickly delivering products that are urgently needed and adjusting their supply chains,” Jockusch said.
Looking forward, Jockusch sees AI and data coming together to generate self-organizing and automated processes for creating a product like a drone. Consumers could input specific requirements — for example, an autonomous drone that can carry a 1.2 lb. camera and fly for three hours, but not cost more than $250 — and AI-driven software will go off and analyze a knowledgebase of designs until it finds something that fits the bill. From there, the software would automatically connect to an intelligent marketplace where it would start sourcing components, identify suitable manufacturers, and handle the bidding and contract process.
“The basic technologies for this vision might be 10 or more years into the future, but the technologies are already helping to facilitate increasingly complex design jobs in many of our applications in a much faster and more reliable way,” Jockusch said.
Fighting fraud
Most people understand the utility of the iconic plastic Mastercard in their wallet, but are less familiar with the underlying network of merchants, institutions, government agencies, and technology companies associated with the billions of transactions generating data on an unprecedented scale.
That data gives Mastercard an opportunity to leverage AI to come up with services and offerings that improve the customer experience, McLaughlin said.
One of the most visible ways Mastercard is channeling those resources is to fight fraud. While the company had historically tackled fraud detection through rules-based technologies, those systems are more likely to trend towards false positives — most consumers know full well the frustration of a credit card being shut down while traveling because a purchase is initiated from an unknown location. “We took that as our purpose — how do we get as many good transactions as possible through?” he explained.
To accomplish its goals, Mastercard built a decision-management platform on top of a massive in-memory grid in its network that holds over 2 billion card profiles with 200 analytical vectors. The system, which is embedded in all of Mastercard’s transaction flows, leverages 13 AI technologies along with some rules-based tools for optimization, a help given that decisions on fraud have to be made in less than 50 milliseconds.
“We were able to have a three-time reduction in fraud and a six-time reduction in false positives using AI with that graded dataset,” he explained.
Precision agriculture
In a perfect world, a farmer would tend to a single crop all season, staying razor-focused on soil consistency, nutrient counts, and the perfect time for harvesting. No one can make a living on such one-to-one treatment, but AI is helping farmers achieve that kind of plant-by-plant-level management at scale, said Julian Sanchez, director of emerging technology for John Deere.
John Deere has integrated a modern AI and computer-vision platform into industrial machines like sprayers and combines. Equipped with intelligent systems, these machines can detect in real time what crop is on the field and initiate decisions while also shuttling back data to the cloud to drive insights for others in the greater farming operation.
For example, the robotics-enhanced sprayer uses computer vision to recognize plants, ensuring it sprays herbicide on weeds and fertilizer on crops. The result is less herbicide used, which has both economic and environmental implications for farmers and the greater population.
“We can leverage AI, machine learning, and machine vision to go through a field at a high level of productivity while still helping farmers farm more profitably and sustainably,” Sanchez said. “We are managing every inch of the field, every plant, with the highest level of specificity possible. That’s the aim of precision agriculture.”
Originally published at https://mitsloan.mit.edu on June 14, 2021.
