Vectors of Innovation with Conversational AI
Conversational AI is a huge technology advancement – as momentous as the unveiling of the Internet in 1983 or when Steve Jobs launched the iPhone in January 2007. But within the last year or two, Conversational AI has evolved into a cornerstone of innovation. Gone are the days of single-use chatbots that execute pre-scripted, single-path programs or recite your service manual to customers.
With Conversational AI, we are talking about complex, Machine Learning (ML)-powered, intelligent Digital Assistants that can drive unmatched customer and employee interactions based on the current context, past history, even predicting the flow of conversations and delivering next best actions – based on naturally expressive voice or text. Where the engagement with a digital assistant is intelligent enough that you think of it as a “cobot” – a co-pilot in your journey as a customer, employee, vendor or partner.
This article walks through the journey from a single-use chatbot to today’s intelligent Conversational-AI based digital assistants and explores the different vectors of innovation that may completely transform the way we engage with Conversational AI and with our world. We will share real-world use cases and actual product development efforts that are making this utopian AI world a reality.
From Single-Use Chatbot to Intelligent Conversational-AI Based Digital Assistants
The first Chatbot, ELIZA, was invented in 1966. Then, in 1995, Robert Wallace introduced ALICE, which worked with the XML schema Artificial Intelligence Markup Language (AIML), to help specify conversation rules. ALICE simulated chatting with a real person over the Internet and had a distinct personality.
This type of structured, specific use chatbot remained the standard until about ten years ago when we saw the dawn of a new era for Conversational AI with the birth of consumer digital assistants, as well as viable use cases for the enterprise. The introduction of Apple’s Siri in 2011 gave rise to a rapidly evolving world of AI-powered assistants that lived in the palm of our hand, and where you could simply “talk to the phone” to get answers.
Interactive Voice representatives began to provide customer service when customers had a problem or needed pre-determined information, and were intelligent enough to know when to pass off the user to a human agent. Clear boundaries were established with digital assistants as first-line customer service and human agents handling more intricate calls. But true innovation happened with seamless bot-agent handoff where intelligent digital assistants hands off an inquiry to live agents, sharing context, conversation history, and relevant details so that the customer does not have to repeat the conversation they just had. Then, the digital assistant can continue to listen and learn for future interactions.
Towards the end of the last decade, advancement in Machine Learning made Conversational AI increasingly relevant in the enterprise world – not just as a consumer-enterprise conduit but also for employees and internal functions.
There are three key differences between consumer-grade and enterprise-grade digital assistants – first, enterprise digital assistants need to draw on domain/industry-specific language, and trigger processes across the enterprise’s diverse set of applications. Second, an enterprise digital assistant must be intelligent enough to parse very specific and non-linear requests delivered in natural language. Support of multiple languages, complex vernacular and structures are key. Third, privacy and compliance are critical considerations. Even beyond GDPR and industry policies, an organization’s data and conversations are sacrosanct.
Vectors of Innovation
The evolution of Conversational AI will continue at breakneck speed now that Computational AI has reached new speeds and intelligence is converging across devices and systems and addressing the complexity of the enterprise. Currently, four key vectors of innovation are transforming the way we engage with Conversational AI.
- Naturally Conversational – Digital assistants now use voice or text to communicate bi-directionally with the user. They can discern real meaning from context when there might be more than one meaning for what a user is saying, and are able to converse out of sequence. They are personalized and human-like, rather than a limited chatbot with only a few potential responses and actions. For example, an employee may enquire about their vacation balance, then apply for leave and transfer pending tasks to a peer for the vacation period all using the same digital assistant.
- Understanding of the User – Today’s digital assistants know the user really well. They understand roles, behaviors, traits, and preferences, and their interactions span multiple channels and apps. A modern digital assistant knows how and when it should communicate and can recommend actions based on real-time needs. For example, if an employee is transferred to a location in a different state and updates their address in the HR system, the digital assistant could recommend actions like updating a W4 form in accordance with state and local tax rates.
- Knowledge, Memory, and Reasoning – Digital assistants today have a semantic understanding of various domains, which is critical in the enterprise where acronyms could have several meanings, and specialized knowledge is paramount. They understand events and benefit from both long- and short-term memory. For example, a travel service’s digital assistant might learn a frequent traveler’s preferred airline, times to fly, hotel room preferences and prompt the user to book a car service to the airport.
- Proactive behavior – an intelligent digital assistant watches for things the user cares about and takes actions autonomously or alerts the user. For example, when a delivery person reports an incident, the digital assistant can adjust the delivery time or route, alert the supply chain personnel and customer, and help locate help nearby. The same delivery person can order an alternate ride, scan the ride receipt to file expense using the same digital assistant or trigger recommendations for a service or sales executive to initiate a plan to improve customer satisfaction, loyalty and improve the company’s top line.
As we usher in the new decade, the good news is that these are no longer hypotheticals. Oracle Digital Assistant is tackling these use cases and more in enterprises around the world. We are at an exciting time for Conversational AI. As it continues to evolve, the potential use cases will grow exponentially, driving unmatched customer and employee interactions and opening up enterprises to new levels of satisfaction and productivity.