CIO.com
6 machine learning success stories: An inside look
By: Clint Boulton
Fewer technologies are hotter than artificial intelligence (AI) and machine learning (ML). Leading organizations are already harnessing the technology, which mimics the behavior of the human mind, to woo customers and bolster business operations. And the trend will only gain more traction in the years ahead, as AI and ML will be a top five investment priority for more than 30 percent of CIOs by 2020, according to Gartner.
Initial fears over AI and ML being used to displace jobs appears to be dissipating, with more than 67 percent of business executives surveyed by PwC saying that AI will help humans and machines work better together. Recognizing the opportunity to move the needle for their businesses, some CIOs are experimenting with, building and even patenting new AI and ML technologies. These IT leaders shared their ML use cases with CIO.com.
AI augments securities research
Putnam Investments, a provider of mutual funds, institutional investment strategies and retirement services, views AI and ML as essential for driving improved coverage of stocks by the financial services firm's research analysts, CIO Sumedh Mehta tells CIO.com.
The analysts work closely with Putnam data scientists to create theses that help glean insights from large amounts of data, Mehta says. Putnam is also working on algorithms that will recommend the most important sales prospects.
"It's a hugely disruptive and transformational power and the whole business driver for it is efficiency and productivity," says Mehta of AI and ML.
Mehta, who relies on a combination of software engineers, data scientists, analytics and vendors, has created a data science center of excellence, which is essentially ground zero for AI and ML efforts that support business stakeholders. He says his "enlightened" business partners have embraced these approaches to achieve better automation.
The AI and ML work is part of Putnam's broader digital transformation, which entails modernizing IT infrastructure with cloud computing and creating a single platform on which to run the business.
Key advice: Organizations should take their time and set expectations appropriately, understanding that the first few ideas will lead to new questions rather than answers. "There is no such thing as a eureka moment when it comes to AI," Mehta says. "It’s not the case that suddenly your algorithm will yield insight you didn’t already know about."
AI makes finances less taxing
Intuit is accelerating AI and ML efforts under Ashok Srivastava, who joined the financial software maker as chief data officer in October.
Intuit is using Amazon Web Services to help its QuickBooks Assistant chatbot better understand and process natural language, says Srivastava, who joined the company after building out Verizon's big data platform. A growing area of focus is shepherding users through the hundreds of categorizations that inform Quickbooks.
"We're dealing with over 1 billion transactions from QuickBooks and we can optimize the categorizations that occur with high accuracy," Srivastava adds.
The company’s TurboTax uses AI to help users get their maximum refund by guiding them through the itemized deduction process, potentially saving users up to 40 percent of tax prep time and efforts retrieving documents.
The company is using ML and cloud technology from AWS to scale more rapidly, Srivastava says.
Key advice: Cultivating sound algorithms requires attracting the right engineering talent to solve real business challenges. Srivastava, who also worked for NASA’s Ames Research Center, is currently hiring engineers who can work with ML and AI technologies to achieve the company's goals.
Historical data predicts future performance
Rich Hillebrecht has unique challenges as the CIO of Riverbed Technology, a provider of software designed to improve the performance of wide-area networks. Hillebrecht says he is testing how to use ML to ingest data from multiple sources across the company's supply chain to drive better business insights.
"We want to apply machine learning techniques to process way more data than we normally would have," Hillebrecht tells CIO.com
For example, Riverbed might combine order management and other ERP data with historical data about weather and other factors to find patterns that could predict future performance. "We want to be more predictive in terms of downstream risk in terms of capacity and our ability to fill orders to customers," Hillebrecht says.
Other Riverbed use cases could include using ML to automatically tune performance configurations and spot cybersecurity threats. Hillebrecht anticipates creating a single data lake from which business insights can be drawn.
Key advice: Sound strategy for AI and ML requires a cautious approach. Hillebrecht says he is carefully evaluating tools and technologies, including IBM Watson.
Banking on better customer insights
Like many large banks, U.S. Bank has collected a wealth of customer data. And like most banks, U.S. Bank has struggled to derive actionable insights from this data. Bill Hoffman, chief analytics officer of U.S. Bank, is working to change that. For the past several months, he has been using Salesforce.com’s Einstein AI/ML technology to increase personalization across the bank’s small business, wholesale, commercial wealth and commercial banking units.
For example, if a customer searched on U.S. Bank’s website for information about mortgage loans, a customer service agent can follow up with that customer the next time they visit a branch. It also helps U.S. Bank find patterns humans might not see. For example, the software can recommend that agents call a prospective client in a particular industry on Thursday between 10 a.m. and 12 p.m. because they are more likely to pick up the phone. Einstein can also put a calendar invite into the agent’s calendar to remind them to call the candidate the following Thursday.
Such capabilities get to the core of what many financial services organizations are trying to do; cultivate a 360-degree view of customers to recommend relevant services in the moment. “We are moving from a world that was describing what happened or what is happening to a world that is more about what will or should happen,” Hoffman says. “The core value is staying a step ahead, anticipating our customer needs and the channel they want to interact with us.”
Key advice: Take a test-and-learn approach to AI and ML and be patient. But also be ready to scale things that are working. “Always have the customer at the center,” Hoffman says. “Ask: How will this benefit the customer?
ML removes ‘toil,’ making work more productive
Ed McLaughlin, president of operations and technology at Mastercard, says ML “pervades everything that we do.” Mastercard is using ML to automate what he calls “toil,” or repetitive and manual tasks, freeing up humans to perform work that adds productivity and value. “It's clear we've reached a state of the art where there is a clear investment case to automate workplace tasks,” McLaughlin says.
Mastercard is also using ML tools to augment change management throughout its product and service ecosystem. For example, ML tools help determine which changes are the most risk-free and which require additional scrutiny. Finally, Mastercard is using ML to detect anomalies in its system that suggest hackers are trying to gain access. McLaughlin also put a “safety net” in the network; when it finds suspicious behavior it trips circuit breakers that protect the network. “We have fraud-scoring systems constantly looking at transactions to update it and score the next transaction that's going in,” he says.
Key advice: As far as McLaughlin is concerned, AI/ML are just tools in the payment processor’s broad toolkit. Despite all of the shiny new tools on the market, he says CIOs shouldn’t rely on them to magically fix business problems.
AI as a product and business enabler
At software maker Adobe Systems, CIO Cynthia Stoddard is reimagining her department with a “data-driven operating model,” relying on Hadoop-based analytics to gain insights to both run IT and the business better. As part of the data-driven strategy, Stoddard says she is experimenting with ML to help analyze tickets in help-desk software to look for trends in system failures. The thinking goes, if the system sees events that suggest an outage could occur, the system can be proactive to eliminate or mitigate those events before they trigger failures.
Identifying patterns in IT service failures, she says, will also empower Adobe to create some “self-healing” capabilities to absorb work that her IT staff currently does. She is also looking into chatbot technology to field employees’ IT support requests. Adobe’s commercial business has also embraced AI. In November 2016, the company introduced Sensei, a layer of AI technology it is applying to its product for creating and publishing documents, and for analyzing and tracking web and mobile application performance.
Key advice: Using ML to identify patterns is the key to creating self-healing capabilities. “If you know how you fixed it you can put self-healing component in there and take the human element out of the equation,” Stoddard says.
