Why smart enterprises are thinking AI

Companies are leveraging artificial intelligence and machine learning technologies to bolster customer experience, improve security, and optimize operations

Why smart enterprises are thinking AI
Taylor Callery

TGI Friday's may have a reputation as a casual restaurant and watering hole, but its messaging to customers was hardly conversational. The well-known chain sent out regular blasts through traditional broad-reach media and, more recently, social media, yet it increasingly wanted to re-create the banter that happens organically when regulars belly up to the bar.

In lieu of hiring a battalion of customer service "bar keeps," TGI Fridays recruited an enterprise conversation platform infused with a shot of machine learning and artificial intelligence (AI) to personalize its messaging and overall customer experience. Now, patrons can chat up the AI for happy hour suggestions and appetizer specials, engage in small talk using emojis, make reservations, and order takeout via social media channels and through Amazon Alexa.

"We thought about how technology could help us create that one-on-one personalized messaging outside of the bar without having to hire 1,000 people to respond to individual guests," says Sherif Mityas, vice president of strategy and brand initiatives, as well as acting CIO, at TGI Fridays. "We wanted to be part of the conversation when someone was thinking about where to go for happy hour or get recommendations on the most popular drink. That's where the initial power of chatbot technology comes into play."

The restaurant chain's chatbot, created with Conversable, is just the appetizer in what is expected to be a full course meal as AI and machine learning capabilities take root in other enterprise systems, from security platforms to sales systems. While hardly newcomers to the technology scene, AI and machine learning have burst into the mainstream in recent months. Stories about robots, autonomous vehicles and smarter consumer products are grabbing headlines, and voice-powered digital assistants like Alexa and the recommendation engines of companies like Netflix and Amazon have become familiar parts of our everyday lives.

At the same time, technologies such as Google Deep Mind and IBM Watson, once ivory tower research projects, are also gaining notice as the engines that power a variety of applications in sectors like healthcare and finance (H&R Block's tax preparation service is one example).

Early days still

Despite the hype, it's still early days for AI, especially in the enterprise. The technologies are still evolving, although much more rapidly today, thanks to nearly unlimited computational power, the collection of vast amounts of data and advances in neural network capabilities. While the terms AI, machine learning and deep learning are used somewhat interchangeably, there are differences among them, and failure to grasp those differences can lead to confusion.

AI constitutes the broader concept of employing machines or systems to carry out tasks intelligently. Machine learning is an application of AI whereby a system learns how to act on its own based on the data being collected. Deep learning, a subset of machine learning, applies many layers of neural network models and algorithms to solve highly complex and data-intensive problems.

In a recent Forrester Research survey, just 17 percent of the respondents said that they will be implementing or expanding their use of AI systems over the next year. However, 55 percent said they intend to invest in the technology over the same time frame. Nearly half of those polled said they hadn't yet seen any results from their AI initiatives, and the lion's share have invested or plan to invest less than $1 million in such efforts through 2018.

Mastering AI takes time

One factor holding up the spread of AI in the enterprise is the learning curve, because most IT leaders and executives still don't fully comprehend the nuances of the AI stack, much less understand how to apply the technologies to solve real business problems, experts say. On top of that, organizations are grappling with the usual budgetary, business case and talent gap concerns that remain barriers to implementing many cutting-edge IT projects.

"Last year, everyone got so focused on chatbots, machine learning and AI that they started to use [the terms] all magically and interchangeably, but it created massive market confusion," says Ben Lamm, CEO of Conversable. "Every major company now gets the point that AI can have massive implications to the business -- they just don't know how to get there."

The first wave of interest seems to be around leveraging AI technologies to improve customer experience and support. Fifty-seven percent of respondents to the Forrester survey cited improving the customer experience as a reason for using AI, with 37 percent reporting that they're implementing or planning to implement intelligent assistants, and 35 percent saying they're working to develop cognitive products for customers.

One area where machine learning and smart algorithms are starting to have a significant impact is in detecting known and unknown attacks, thereby allowing IT security professionals to take a more proactive security posture. Sales and customer service are other areas where AI technologies are starting to deliver results: In a survey conducted by the Accenture Institute for High Performance, 40 percent of companies said they're using machine learning to improve sales and marketing performance.

"The interest in the enterprise in doing something with AI is really high," says Joshua Feast, CEO and co-founder of Cogito, which markets a platform that leverages real-time intelligence and machine learning to help call center workers better engage with customers. "The problem CIOs are finding is that a lot of the things they want to do are on the margin, like implementing a different type of UI on a website. The only way to move core metrics is to impact major operations . . . and CRM, sales, security and call centers are the best way to do that."

AI gets to work

Dan Olley, CTO of Elsevier, is exploring enterprise AI use cases on a number of fronts. The information services provider is test-driving AI-based security software to up its cybersecurity game. It also has interest in leveraging AI capabilities in its CRM platform to improve lead generation, and is actively evaluating tools such as chatbots and knowledge management systems to improve its customer support experience. For example, Elsevier's technology and developer group is already benefiting from an AI-based knowledge management system that dynamically identifies relevant content and delivers it to the appropriate person without human involvement, Olley says.

Elsevier is also using AI to improve existing product offerings and monetize new ones, Olley says. In one example, the company used AI to extract medical images out of decades-old content, classify and annotate the material, make it searchable, and repackage it as a new offering in a matter of only four weeks, he says, noting that such an undertaking would previously have involved years of effort to unearth and assemble the relevant images.

The keys to more widespread application of AI, Olley contends, are making sure his technology team understands the organization's data and getting staffers up to speed on emerging AI capabilities. "Once they're trained in the art of the possible, they find new applications almost daily," he says. "Start small and it becomes a self-perpetuating engine."

Intelligent banking

Capital One, widely heralded for its technology-driven approach to banking, is already well down the path of using AI and machine learning to transform customer service and banking systems, says Adam Wenchel, the company's vice president of AI and data innovation. Last March, Capital One announced that it was integrating Amazon Alexa into its IT systems to establish a foundation for introducing new services that will allow customers to do their banking in conversational, hands-free ways, regardless of environment, Wenchel says.

In the future, Capital One plans to use machine learning to analyze call center conversations and identify major themes in an effort to improve customer service, detect fraud and identify new business opportunities. For example, that type of analysis could figure out which type of banking customers have a propensity to become investing customers, Wenchel says. In another example, Capital One is using machine learning to identify characteristics of a neighborhood or to uncover reasons why residents are moving into or out of areas to help optimize its home loan underwriting process in new and existing markets, Wenchel says.

Huge potential

TGI Fridays also sees huge potential for AI, says Mityas. For example, beyond improving the customer experience, the data collected via the conversation platform could help the restaurant chain better understand its customers and their requirements. For example, data could be used to identify the most frequented social media forums to ensure TGI Fridays is actively participating in the conversations.

The machine learning capabilities go beyond traditional analytics in that the more data TGI Fridays collects on customers and their behavior, the more the technology gleans insights that can help the company tailor outreach or serve up specific offers in near real time. "The more we know about you, the more we can personalize messages and not just respond," Mityas says.

Already, customers are reacting positively to the higher level of engagement, he says. For example, TGI Fridays has seen a 500 percent spike in engagement with customers on social media channels since deploying the new conversation-based customer experience tools. It has also seen a bump in conversations initiating commerce activities such as ordering food or making reservations through social media platforms, he adds.

The same AI-driven capabilities have a lot of potential to help TGI Fridays develop future product offerings. "We are garnering consumer insights that can help us understand trends in taste, preferences for time of day to come in, or what kind of burgers are popular. We can feed [this data] to our culinary and marketing teams to create more relevant products," Mityas says. "The beauty of machine learning is that it gives us guidance beyond just the data. It becomes a feedback loop."

TGI Fridays also envisions gaining insights that can drive operational efficiencies by, for example, incorporating AI and machine learning capabilities into enterprise tools such as point-of-sale (POS) systems, Mityas says. "There's a lot going on within our four walls on a busy Friday night, from the front of the house, like how we allocate tables, to the back of the house, such as the flow of meals or what gets cooked," he says. "We can analyze all of that and learn how to run a more efficient restaurant."

The CIO's role

How successful TGI Fridays or any other company will be with AI technologies depends, in part, on the CIO. While line-of-business leaders can set up stand-alone systems piecemeal, an AI undertaking won't become a transformational initiative unless the CIO takes the lead to ensure that officials engage in the proper planning and strategic thought processes to support an enterprise view, experts say.

"CIOs who can recognize and figure out how to create business value are the ones that can position themselves across the enterprise," says Matthew Guarini, an analyst at Forrester Research. "CIOs need to be thinking about how to put the right road map in place, how to leverage data assets and how to get the right governance practices implemented."

You should start by identifying which business processes have cognitive bottlenecks and where fast, accurate decision-making can make a difference, especially those cases that involve too much data for humans to analyze or where it's too expensive to hire people with specific expertise, says Tom Davenport, a professor of IT and management at Babson College.

Davenport also advises CIOs to attack AI as a portfolio of projects. For example, they could do something in statistical machine learning along with efforts involving, say, chatbots, image recognition or speech recognition in cases where those types of technologies would meet specific needs. "Don't put all your eggs in one basket -- start to learn what types of use cases make sense for what technologies," Davenport says. "Unless you're really ambitious and are trying to totally transform the business model, it makes sense to be more conservative and have a portfolio of projects that is less dramatic than trying to pull off a moon shot."

Building an AI team

Assembling the right talent is another critical component of an AI initiative. While existing enterprise software platforms that add AI capabilities will make the technology accessible to mainstream business users, there will be a need to ramp up expertise in areas like data science, analytics and even nontraditional IT competencies, says Guarini.

"As we start to see the land grab for talent, there are some real gaps in emerging roles, and those that haven't been as critical in the past," Guarini says, citing the need for people with expertise in disciplines like philosophy and linguistics, for example. "CIOs need to get in front of what they need in terms of capabilities and, in some cases, identify potential partners."

1 2 Page 1
Page 1 of 2
InfoWorld Technology of the Year Awards 2023. Now open for entries!