Artificial Intelligence Gets Real
For many people, the phrase “artificial intelligence” conjures up images of human-like robots and self-driving vehicles. For Kerrie Holley, AI has a much more human meaning.
Holley had a long career, first as a distinguished engineer and later IBM fellow, before joining Eden Prairie-based Optum, a UnitedHealth Group unit that provides information and technology-based health services. At Optum, he’s focused on developing technologies that make the health system work better and people’s lives healthier. AI is one of those technologies.
AI can make better predictions about who might develop a condition such as diabetes based on a vast universe of digitized patient data. “This has a huge positive upswing in people’s health,” Holley says, “because now I have more people whom I can accurately predict [will develop a disease]—people whom I can call and do an intervention on. And I’ve also been able to do that faster than before.”
All the talk of androids and automatons aside, AI in real life means technologies that process and find useful patterns in huge amounts of digital data, whether it originates from sensors, smart devices, online social interaction, or, increasingly, voice and video. High-powered cloud computing can now take that disparate data and glean new insights into customer preferences, show where production processes are likely to break down, and better organize employee scheduling, among other business uses. AI is a technology that can write and rewrite its own algorithms as it takes in more data.
Optum is far from the only Twin Cities company developing AI capabilities and incorporating them into its business. The list includes startups such as Minneapolis-based Inspectorio, which is using machine learning—one of the building blocks of AI—to help retailers and vendors verify the quality of the products they sell. Numerous established businesses (notably Target, Land O’Lakes, and the Mall of America) also are exploring ways to put AI to work. This summer, Eden Prairie-based Starkey Hearing Technologies introduced its Livio AI hearing aid, which incorporates machine learning to continually improve its performance and to measure the wearer’s brain activity.
In other words, AI is already here. “It’s a nascent field—still young,” Holley notes, but businesses of all kinds are finding ways to use it.
Beyond the buzz
“The use of the term ‘artificial intelligence’ right now is very broad in the media—it’s become a buzzword,” notes Ravi Bapna, professor of business analytics and information systems at the University of Minnesota’s Carlson School of Management. (Bapna also is academic director of the Carlson Analytics Lab, whose students have worked on several AI-related projects for area businesses. See “Minnesota’s AI Vanguard,” below.)
“In reality,” he adds, “when we think about AI, we think about machines with human-like capabilities” to both discover hidden patterns in data and then use those discoveries to make better predictions.
“We use machine learning to achieve the goals we want with AI,” Bapna says. “Machine learning is either about finding new and interesting patterns of association in your data, or about making predictions based on those patterns.” In AI, the technology is able to “learn” from its past errors and to incorporate new data.
Another way to look at AI or machine learning, says Minneapolis AI entrepreneur Mitch Coopet, is that “it unlocks the ability of people to look at unstructured data sets and make them useful.” Unstructured data are bits and bytes that aren’t automatically organized into pre-defined categories. In the not-so-distant past, they were treated more or less as noise. But data scientists have discovered that there is a lot of useful information in that noise.
As Coopet and others in the field note, even those who have some idea about what AI involves have numerous misconceptions about the kinds of technologies that fall into that bucket. It does not include seeming humanoid technologies like chatbots or digital algorithms—technologies “that have been around since 1991,” he jokes.
With that in mind, here’s a look at how Minnesota companies are using AI to boost business.
Better consumer insights
Start with Coopet’s own company, Rambl. Officially launched in March, Rambl’s current product is “an AI-powered inside-sales phone system,” Coopet says. With Rambl, each sales call “is turned into a processable data stream using machine-learning techniques,” he says.
Most sales calls, Coopet notes, are not being processed or analyzed. When the sales manager wants to evaluate how well the caller is performing, he or she often listens in. The same is true of a new salesperson who needs to learn the ropes. That makes learning and training ad hoc, Coopet says. What Rambl’s technology promises is a more data-driven approach to determining what works on a call—the kind of wording, style of conversation, and type of product features that are most likely to appeal to certain customers. Rambl’s technology, in a sense, does the listening and picks out patterns in the salesperson’s patter that make for a successful or unsuccessful call.
Coopet sees Rambl’s market opportunity as just about any company that does selling by phone, particularly where it is engaged in high-volume dialing. Rambl’s technology also integrates with CRM (customer relationship management) tools such as Salesforce and HubSpot to streamline the salesperson’s data entry and logging, so salespeople can focus on their primary task, not administrative chores. The latter, says Coopet, can take anywhere from 10 to 20 minutes per call.
Because Rambl is rooted in machine learning, Coopet argues it will only get stronger as it learns what to listen for in spoken conversation. He notes that’s very different from written communications. “Five years from now, we believe that businesses will be able to search and learn from voice conversations as easily as email,” he says.
Rambl is a homegrown AI firm. Coopet co-founded Rambl after nearly 15 years at Minneapolis-based backup and security software company Code42, which he helped launch in 2001. A newcomer to the local AI scene with origins outside the state is Cludo, a Copenhagen-based company that opened its U.S. office in Minneapolis in February 2017. Cludo has developed a site search and analytics platform that incorporates machine learning to continuously improve the site’s response to search queries, with a goal of improving customers’ online experiences as well as boosting website sales conversion rates.
In addition to making company websites more useful, Cludo’s technology also captures visitor data that is quite valuable, CEO Philip Andersen says. Machine learning and AI can give structure to this unstructured data, which, Andersen says, can help companies understand customers’ needs and purchasing habits at an increasingly deeper level.
With AI and machine learning’s capabilities for finding patterns in digital consumer data from purchases, social media, and so on, it’s not surprising that retailers and marketing firms have been exploring its uses. Their goal, of course, is to better target products to people who are most interested in buying them.
Andrew Eklund, founder and CEO of Minneapolis-based digital agency Ciceron, has been working with a variety of AI platforms. One of its key partners is St. Louis Park-based Equals 3 LLC and its Lucy technology, an AI-based “intelligent agent” powered by IBM’s Watson supercomputer. Lucy’s ability to parse large amounts of consumer data “helps us make more informed, faster media-buying decisions,” Eklund says.
Ciceron also uses AI to pick through online data to discover “how pockets of people are having conversations about brands,” Eklund says, and to learn the language that consumers actually use. For instance, Ciceron has numerous health care clients. To reach those clients’ customers and patients, the marketing firm needs to differentiate between how consumers talk about their health concerns and the terminology that physicians and nurses use. AI technology helps Ciceron sort through the conversations that consumers have so that the agency can address them in ways they can comprehend. That’s instead of “expecting everyone to have a medical degree to understand what we’re talking about,” Eklund says.
For marketers, there’s one notable drawback to AI. “We have several very large platforms, specifically Amazon and Facebook, that are walled gardens of data,” Eklund notes. “You can’t get information out of those systems.” And that closes off a vast amount of particularly useful consumer information.
The human factor
Perhaps in part because of the misconceptions that surround them, AI and machine learning do have a certain “cool” factor. But any business considering adding AI and machine learning “should have a very specific business problem in mind,” Rambl’s Coopet says. Ultimately, the human beings using the technology need a clear idea of what they want to accomplish.
The data gathering and analysis capabilities born of AI promise to change how businesses market and sell, how they operate on the production floor, and how they manage their supply chains. And AI won’t necessarily be visible. “AI technology will be embedded into everything we do over time,” Cludo marketing director Sahil Merchant asserts. “It will be a fundamental way that we interact with the world.”
The Carlson School’s Bapna sees potential uses of AI in “every industry and every function.” He offers as one of many examples a cheese company seeking to avoid production failure that would make its products unsalable. Using the data in the sensors embedded in each stage of the manufacturing process, machine-learning technology can help the cheese company generate an algorithm that shows “that if these conditions are present, you’re likely to have poor-quality cheese,” Bapna says. From there, the company can develop a predictive model that can prevent that undesirable outcome.
Financial services firms also are exploring AI. Seeing a need for employees who can handle increasingly complex data analytics, professional services and auditing firm PwC developed its Digital Accelerator training program, in use in several PwC offices worldwide, including Minneapolis. The program provides “digital upskilling” training to select staffers in areas including machine learning and advanced data analysis in topics such as tax and advisory.
At Optum, Holley’s team has been developing a number of AI-based projects. One is “network steerage,” which means guiding patients to health provider networks where outcomes for certain conditions are likely to be better. Optum also is using AI and machine learning to improve its fraud detection techniques and beef up customer service. Looking ahead, Holley sees numerous other applications. For instance, machine learning with voice recognition and natural-language processing could be combined in ways that could help physicians detect the onset of Alzheimer’s by the way people talk on the phone.
Holley stresses that none of this is science fiction. “Artificial intelligence will work hand in hand with people,” he says. “It won’t be an existential threat to humanity. Instead, it will advance our ability to solve some of the most wicked problems we face, whether in health care or any other industry. It’s because people will be working with AI, not because AI is doing something to people.”
Minnesota’s AI Vanguard
The Carlson School of Management Analytics Lab at the University of Minnesota has worked with numerous Minnesota businesses on projects incorporating AI and machine learning:
- Shortly after adding free Wi-Fi, the Mall of America in late 2015 worked with the Carlson Analytics Lab to find patterns in the vast amount of user data for insights into foot traffic patterns. Students mapped the Wi-Fi data and discovered increasingly predictable patterns, such as the impact of promotions on cross-selling and the affinities certain types of shoppers have for particular brands.
In addition, the study provided information that MOA managers never had before about how people move around the space, says Carlson’s Ravi Bapna. “If they find there are certain areas getting unusually high traffic at certain times, those areas might need more janitors,” he says.
- Carlson Analytics Lab students worked with Hennepin County Medical Center (HCMC) to forecast patient populations across 17 different nursing units. The students combined five years of historical patient “census data” with data on weather, population, and community events (like Twins and Vikings games) that can affect patient counts. The students then developed a dashboard that helps HCMC managers create schedules to avoid both inadequate nursing coverage and costly overstaffing.
Why Is Blockchain a Big Deal?
Artificial intelligence isn’t the only new technology that’s generating buzz in the business world.
People who have heard of blockchain typically associate its use with cryptocurrencies such as bitcoin. But James Mitchell, president of Minneapolis-based software development firm Bitwise IO Inc., says there are numerous potential business applications that have nothing to do with those mysterious alt-coins.
“I like to describe currencies like bitcoin as platforms and applications,” Mitchell says. Bitcoin, for example, is the software that manages the exchange of information between the computers that are running the bitcoin network. But it’s also the application—meaning the business rules that determine what you can do using the system, he says.
In other words, a blockchain platform doesn’t need to use bitcoin. For the past few years, Bitwise IO has been collaborating with California-based Intel to develop a blockchain platform called Hyperledger Sawtooth. Mitchell says it is designed to enable any business-to-business transaction.
Hyperledger is an open-source blockchain project launched in 2015 by the Linux Foundation. Bitwise IO also has been working with a number of businesses, including Minnesota companies, on potential blockchain applications that might use Sawtooth as its platform.
One way to think of a blockchain is a shared (distributed) database or ledger that records all transactions between the various parties involved. One attribute that makes blockchain different from standard online transaction records is that it bypasses a clearinghouse that keeps all the players in the transaction honest. Instead, a blockchain network has a built-in cryptography and security rules that all the connected parties are required to follow.
Mitchell cites some examples of how businesses could use an enterprise blockchain platform. For instance, logistics companies and others involved in highly complex sources of supply could use it to verify the origin and quality of the goods as they move through the supply chain. And businesses involved in international transactions such as land sales, where the clearinghouse is less than trustworthy (a corrupt government, for instance), could find a blockchain-based accounting system more reliable and honest.
At this stage, blockchain is “a niche technology,” Mitchell says. It has numerous potential uses. At the same time, like AI, blockchain is still evolving, and it’s likely many businesses will not need it. But companies involved in highly complex transactions of money and data might soon find it a useful connection.
Gene Rebeck is TCB’s northern Minnesota correspondent.