Where AI Is Working, Where AI Needs Work

Matthew Bieber | Apr 10, 2018

If you’re a sports fanatic, you probably were glued to your television watching this year’s March Madness matchups. And more than likely, you saw IBM Watson’s latest TV commercial, “Watson at Work: Basketball.”

In the commercial, Watson, the IBM system capable of answering questions posed in natural language, is helping two professional basketball scouts find data on NBA prospects. When Watson gets some terminology wrong, one of the scouts points out the mistake. Watson’s response? “I’m still learning.”

That’s an important point to keep in mind with regard to artificial intelligence (AI) and natural language processing (NLP), and a statement that sums up the current state of affairs with the technology. 

Still, we’ve come a long way. The concept of AI dates back to ancient Greek and Chinese engineers, who had the idea of inanimate objects coming to life as intelligent beings. However the field of AI wasn’t formally founded until 1956, at a conference at Dartmouth College. The concept was fairly simple until natural language processing started to be incorporated into AI, making things a bit more complicated. According to Wikipedia NLP is known as “computer activity in which computers ... analyze, understand, alter or generate natural language.” NLP is a key component in AI in terms of ensuring a seamless digital experience. The first successful demonstration of NLP took place in 1954, in a Georgetown University-IBM experiment in which computers translated more than 60 Russian sentences into English. 

Fast-forward to today. How has AI evolved and gotten more sophisticated in the millennia since ancient times — and in the six decades since it first became a field of modern scientific study? Let’s take a look into how AI has supported the digital experience of today, how it has succeeded, how it has failed and where we can expect it to go.

Where AI Is Working

There’s no question that AI is playing a tremendous role in the customer experience, and with powerhouses like Microsoft, Facebook, Google and Amazon implementing AI, it’s impossible to deny that AI has the potential to have a significant impact on business. 

Take call centers for example. Customers who contact companies by phone are usually looking for immediate feedback and accurate results. How do businesses make sure that their call center operations fulfill cusomers’ expectations? With the help of AI, of course.

Gone (mostly) are the days where “intelligent” customer service consisted of pressing 1 for sales and 2 for technical support. During my time as executive director of product management and marketing at AT&T, we realized that customers were not happy with that model. I knew that it was time to implement an interactive voice response (IVR) system, a move that would uniquely position AT&T at the forefront of NLP technology. IVR technology offers an error-free alternative to the mindless pressing of buttons. When we deployed our system, it allowed us to route calls to the correct department, making it more likely that customer queries would be resolved on the first try. The personalized interaction brought our company’s core voice and excelling standards into the limelight while going above and beyond to fill the customer’s requests. 

As consultant Keith Kirkpatrick wrote in an article in Communications of the ACM, "Rather than simply being used to replace contact center workers, Al-based technologies, including machine learning, natural language processing and even sentiment analysis, are being strategically deployed to improve the overall customer experience by providing functionality that would be too time-consuming or expensive to do manually."

Having AI provide data on demand isn’t just benefiting the customer, it’s benefiting businesses that strive to connect with consumers and build a smooth experience for both parties. AI is saving customers and companies time and money, and — thanks to NLP — it is eliminating language barriers. It’s swiftly turning call centers into a proactive environments that collect data and turn it into a conversational interaction between the agent and the purchaser.

Where AI Needs Work

In order to fully understand the capabilities of AI, we must also understand its limitations. And what better way to understand its limitations than to examine its failures? Who can forget the time that personal banking company, HSBC, raised security measures by implementing voice recognition ID only to have it cracked after a BBC reporter’s twin brother successfully accessed his account by mimicking his voice? 

Or how about when a Vietnamese company used a 3-D printed mask to trick the facial recognition technology that unlocks the iPhoneX? 

These examples reiterate what we learned in IBM’s March Madness commercial: AI is still learning, and mistakes and ongoing refinements are a necessary part of its development. 

So, where do failures come in in relation to the call center? We know AI has its place, but it’s still missing one important type of functionality: It cannot consider the feelings of other people like a human call center agent could.

Sure, the twin brother must have been amused after he was able to rupture the “high-tech” security measures, but I doubt the owner was as entertained after he realized how easy it was to break into his account. If AI was capable of sensing specific voice changes or subtle differences, this dilemma could have been avoided. 

AI can’t feel empathy and sympathy for others. It doesn’t have humor or trust building qualities and it can’t problem-solve or critically think like a person. It can’t ask, “Hmmm, this voice seems slightly off … I wonder if it could be a difference person.”

While this example was thankfully caught in time, there could be serious consequences in other situations or industries — like healthcare. Those industries — and industries that deal with sensitive topics daily — will always require real human intelligence and EQ. An AI-powered bot isn’t going to be delicate or conscious of the situation; it’s going to do exactly what it was programmed to do and trying to use it for something it’s not meant for will only turn into an IVR disaster.

The Road Ahead for Artificial Intelligence

It’s hard to predict exactly where AI will end up, but we do know it will only continue to move forward. For now, AI has its place in simplistic, straightforward and routine tasks, and while eventually it could advance the customer experience from end-to-end, that won't be anytime soon. It’s likely that AI will revolutionize industries such as medical and engineering and technology will adapt to its growth. But if AI ever plans to fully replace human voice it will need to implement affective computing which could eventually lead to predicting customer queries and providing them with a tailor-made solution. 

Whether it’s transportation, customer interaction or the workforce, AI will gradually become part of our everyday activity. So for now, we can continue to take AI with a grain of salt as we get a realistic handle on what the future of AI might look like — and, just as Watson taught us, always be learning. 

About the Author

Matthew Bieber is the CEO of CDC Software, a provider of SaaS solutions that create real-time, events-based links between leading telephony systems, CRM systems and other mission-critical contact center data solutions. A strategic, business-minded thinker with over three decades of experience in customer care and technology, Bieber oversees and shapes CDC Software’s product offerings by anticipating customer needs and innovating next-generation solutions to enable the delivery of unparalleled customer experiences.