)
Many experts deem artificial intelligence (AI) to be the most important technology of our time. However, the concept of AI is not new at all. In fact, it’s been around for nearly 75 years.
During that time, we’ve experienced periodic phases of “AI winters,” where interest and investment waned, contrasted with periods of “AI springs” where interest intensifies again. The current AI spring is perhaps the most prolific phase of development so far, where maturity and advancement are rapidly increasing.
While the topic of AI is the subject of many conversations around the world, have machines advanced far enough to truly understand those conversations? In some ways, the answer is yes, but in general, it’s more complicated than anyone believed. Let’s consider this in more detail.
Narrow AI Sees Extraordinary Progress
Over the past decade, AI has proliferated throughout our everyday lives — recommending songs, translating languages, answering questions or executing tasks using voice command, dispatching the closest Uber or Lyft driver, and myriad other applications, many of which we are usually unaware.
The biggest advances and achievements have been in narrow AI, that is, areas that target specific tasks, such as:
- Identifying cancer in a mammogram and reducing false positives and false negatives, outperforming human radiologists
- Capturing and processing imagery of vehicle damage to automatically predict repair costs for insurers
- Picking, packing, sorting, and transporting products and packages autonomously for fulfillment
- Inspecting finished goods for quality using smart cameras and AI-enabled quality control software at speeds beyond the abilities of humans
Easy Things for Humans Are Hard for Computers
The world is transforming in many ways thanks to the benefits of narrow AI. However, more generalized AI (where intelligence includes human-like creative and associative thinking) has not yet achieved similar results.
In her paper “Why AI is Harder Than We Think,” computer scientist and author Melanie Mitchell explains, “… the things that we humans do without much thought— looking out in the world and making sense of what we see, carrying on a conversation, walking down a crowded sidewalk without bumping into anyone—turn out to be the hardest challenges for machines. Conversely, it’s often easier to get machines to do things that are very hard for humans; for example, solving complex mathematical problems, mastering games like chess and Go, and translating sentences between hundreds of languages have all turned out to be relatively easier for machines.”
Mitchell cites carrying on a conversation as an example of what’s easy for humans and hard for AI. It turns out that understanding human language is the most difficult problem to solve in the AI world. Even the smartest linguists don’t understand why language works the way it does. We can’t create machines that operationalize true human conversation if we don’t have a fully developed theory of language (we don’t) that would enable machines to understand our conversations.
That’s why we still can’t talk to machines the way we would talk to another human. When you ask a question or give a command to a voice assistant, it often requires dumbing down and slowing down the conversation, making it closer to talking to a child instead of an adult.
Conversations That AI Can Understand
While current advances in the field are still far removed from a generalized AI that can mimic human intelligence, domain-specific conversational AI is making huge leaps in sophistication and already delivering transformational results for contact center use cases.
As an application of narrow AI, specialized or domain-specific conversational AI combines a set of advanced AI technologies to understand, optimize, and automate human conversations and actions within a certain environment.
The contact center is one such environment where techniques such as reinforcement learning and machine learning using millions of conversations have helped advance AI’s abilities to understand and optimize human-to-human conversation within contact centers. Narrowing the domain specialization to a particular industry further improves AI’s accuracy and understanding.
For example, AI technology can be trained to understand terminology specific to a sector such as wealth management. One of the many use cases for conversational AI within this sector is automating compliance with strict regulations. A conversational AI platform can identify when an order is being taken and whether the agent taking the order is complying with regulations regarding disclosures.
“What is Conversational AI?”A set of advanced AI technologies that recognizes and comprehends human language in multiple languages and uses this understanding to optimize and analyze conversations in and across multiple channels.
The Outlook for Conversational AI in the Contact Center is Bright
As domain-specific conversational AI capabilities continue to become increasingly more advanced and sophisticated, they are poised to transform the customer and agent experience. By augmenting humans with AI and automation, contact centers can drive strategic outcomes such as improved customer satisfaction and Net Promoter Score, greater sales effectiveness and higher revenues, lower costs and improve agent productivity, and much more.
Learn more about the state of conversational AI for the contact center in the ebook “The Future of AI for Contact Centers.”