Everyone seems to want a piece of Artificial Intelligence (AI) to help improve their organizations and customer experiences. And why not? Gartner reported last week that global business value derived from AI is projected to total $1.2 trillion this year, an increase of 70 percent from 2017. In 2022? $3.9 trillion.
Not everyone has successfully used their resources for AI affectively. It’s a nascent time for some organizations deploying AI into customer experiences, and there are costly mistakes. Facebook in January shut down M, its virtual assistant. Most are familiar with Microsoft Tay's plight from cool Twitter bot to outright racist.
Why do organizations often fail with AI implementations for customer experience? It’s a combination of technology reliance over a human touch, unrealistic expectations and too much buying into the hype around what works for others without considering real impact internally, experts told CMSWire. We caught up with those AI experts to discuss reasons why organizations fail with AI in the area of customer experience.
Replacing Humans With AI
Cameron Weeks, CEO of Sharpen, said he’s seeing a trend of organizations trying to replace agents with artificial intelligence, machine learning and automation scripts. It doesn’t work, he said, because there is simply no replacement for human beings. “Artificial intelligence,” Weeks said, “has to empower an agent to make them stronger, make them better and make them more efficient. And in some situations you can use that to remove the human and have direct input to a customer, but you can't go about removing the agent. In general, that can't be the strategy. It will never be successful.”
AI should help the process move from bot to real life human seamlessly. That transition, Weeks said, is the most important part of an AI roll out in the customer service world.
Related Article: Artificial Intelligence: Taking the Good With the Bad
Businesses Aren't Ready for AI
AI goes wrong in customer experience when customers and/or the business isn't ready for AI, said JJ Lopez Murphy, AI and data science practice lead for Globant. “A lot of times, organizations give into the hype and invest in AI because they feel like they should, not necessarily because it's best for their end users,” Lopez Murphy said. If they are not building out AI within experiences with the end user in mind, AI can make the experience more complex and frustrating for customers. They can become uncomfortable sharing data in way that would make AI effective. “What's the point of implementing AI if the person who is affected by it doesn't want it?” Lopez Murphy asked. “Ultimately, AI adds a new element into the customer experience that organizations may not be ready for, and that can lead to a poor customer experience.”
Lack of Escalation Strategy (From Bot to Human)
Too often organizations do not have a solid strategy for human escalation — methods to move from bot to live person. The escalation will be a common occurence, Weeks said. And his organization doesn’t discourage the escalation up to an agent. It's what the bot does. Organizations often, however, don’t monitor these escalations closely enough and learn from them by applying analytics. “You have to have a strategy for moving humans into a conversation,” Weeks said. “You have to have a strategy for how the bot can continuously learn based on activity.”
Related Article: 10 Ways AI Helps Improve Customer Experience
Training the Bot Becomes Cumbersome
Too often, organizations fail with AI in the training department. Someone — or, rather, some quality data — has to train these bots, right? The learning piece of the artificial intelligence component — educating the bot — is the hardest part, Weeks said. "Actually training the bot on what to do and what not to do," he said.
Organizations often require a large scale data science team to train the bot itself, and ROI doesn't pay out, Weeks said. It's more expensive to have all these data scientists working in a room trying to program your automation bot than just to keep staffing your customer service team. Bots need to be able to learn on their own based on activity that’s already happening, Weeks said.
Related Article: Can Artificial Intelligence Help Us Tell Fact From Fiction
Being Unrealistic About AI’s Potential
Sure, AI’s been proven to help. "To keep up with rapid digital transformations, many brands have adopted AI software like chatbots to create more efficient and seamless experiences for their customers. However, human connection remains invaluable,” said Andrew Park, VP of customer experience strategy at InMoment. He cited his team’s InMoment’s 2018 CX Trends report which found the two clear leading factors driving great customer experiences were human “staff interaction” (65 percent) and "access to educators and experts" (65 percent). According to the report, despite apocalyptic talk that artificial intelligence will replace people at some point in the not-too-distant future, the human factor is actually what makes or breaks the customer experience.
Starting Too Big with AI
Abinash Tripathy, co-founder and chief strategy officer of Helpshift, said too many organizations when considering AI implementations in customer service and customer experience fail to understand the technology and its state of maturity today. They also fall for the hype of big vendors, he said, without realizing that "more pragmatic approaches have started with very simple approaches." He cited like Amazon's customer service chatbot, which does not use any Natural Language Processing (NLP) or try to have a free form conversation with the customer. Instead, it uses a simple decision tree to help customers with support inquiries.