Generative AI is the hottest innovation in conversational AI today. It’s a type of AI technology that produces human-like content (text, images, code, synthetic data, and audio) based on natural language inputs and contextual information. An excellent example of generative AI is ChatGPT, an OpenAI offering that has taken the AI industry by storm since its release in November 2022.
While ChatGPT can handle fluent conversations impressively well, it’s not quite ready for the enterprise space. In its current version, ChatGPT lacks the governance, tooling, reliability, knowledge base, foundation models, and brand differentiation needed to run an enterprise-based virtual assistant or chatbot. However, ChatGPT can be paired with enterprise-ready interactive AI systems via API integration to introduce generative capabilities to otherwise robotic conversational interfaces. That’s exactly what IBM is doing with Watson Assistant.
Integrating Watson Assistant with generative AI is pretty exciting, technically speaking. But what does this integration look like, and what does it mean from a business perspective? Read on to learn how you can apply generative capabilities in Watson to achieve tangible business outcomes.
Augmenting IBM’s Watson with Generative AI
Watson Assistant is a well-established and mature virtual interaction tool capable of conversing with users, understanding their intentions, and providing relevant information based on each query’s entities. It can also automate tasks by connecting to an enterprise’s back-end systems. Furthermore, Watson is reliable and brand-differentiated since it is trained on the company’s own knowledge base, workflows, and tone of voice.
However, despite its robustness, Watson Assistant lacks that human feel in conversations. Although logical and accurate, Watson’s responses barely have a natural conversational flow and tend to sound more robotic than human. That’s where generative AI comes in. Adding a layer of generative AI on top of Watson’s search results produces more conversational, human-like responses.
Generative AI is filling a gap that chatbots and virtual assistants had: the ability to generate content based on their knowledge. — Alexandre Lanoue, VP & Leader, Business Reimagination, SIA.
When integrated with generative AI, Watson remains the same trusted and dependable virtual assistant we all know and love, but with the following additional capabilities:
- Natural-sounding conversational responses, even when replying to untrained queries
- Personalized answers based on the user, contextual data, and tone
- Easier and faster authoring of conversational flows and AI-assisted customer journeys
The Business Value in Generative AI
Watson is already a good listener; with generative AI, he becomes a good speaker too. Why is that important, and how does it translate to business outcomes? Here are five illustrations of the unique business value generative AI presents when paired with consumer or employee-facing enterprise AI applications:
Reaching a Wider Variety of Audiences
Conventional virtual assistants and chatbots are rather static when it comes to nuanced language characteristics such as tone, voice, and complexity level. In most cases, the source material in the AI’s training data or knowledge base dictates the responses’ tone and complexity.
Obviously, this is not how humans talk.
People naturally adjust their tone and language depending on the topic of conversation, their feelings towards the subject matter, and the listeners. Generative AI mimics that level of language adaptability when crafting query responses. The beauty of such a system is it can seamlessly converse with just about anyone, from industry experts and new prospects to the average Joe, based on how they phrase their questions or their place in the company.
Summarizing Bulky Information into Digestible Snippets
Large language models, such as ChatGPT, are pretty good at condensing vast amounts of information into brief and concise summaries while conserving meaning and essential details. This is a valuable ability for extracting helpful facts and figures from large and unstructured data sources.
Watson Assistant relies on automated search engines (Watson Discovery and other third-party solutions) to scour assorted data sources for queried information. These search engines do a great job identifying and filtering relevant information, but they could use generative AI’s help to present the information in easily digestible blurbs.
With more concise responses, AI interactions become faster, more fulfilling, and less frustrating to the user.
Providing Factual Responses for Untrained Inquiries
What happens when Watson is asked a question he doesn’t understand or hasn’t been trained on? Although responses to such queries vary depending on the conversational authoring, most sound too mechanical. This is another area where generative AI can make responses read more naturally to encourage users to rephrase their query or take a different navigation route.
You can think of this scenario as the infamous “404 Error” in websites. Rather than exposing users to the somewhat alarming 404 default page, UX designers redirect 404 traffic to a branded page explaining what might have happened and advising the user on what to do next. Generative AI does the same thing for queries outside Watson’s scope.
Generating Straightforward Answers to Complex Questions
Generative AI can help your customers or employees get simple and accurate answers to complex questions using Watson Assistant.
Let’s assume that the answer to a query is scattered across multiple documents or data sources in the company’s knowledge base. Naturally, Watson will present the answer in several information cards with links pointing to the various sources. But with generative capabilities, Watson will extract and consolidate all the relevant information from each source into a single coherent answer. Providing straightforward answers goes beyond merely summarizing information. It’s a neat trick that even web search engines like Google and Bing have conveniently incorporated into their search results pages.
Again, this saves time by creating a more fluid and satisfactory search experience for the user.
Integrating Omnichannel Support
Generative AI can match the conversational demands of multiple channels at once. Watson Assistant already supports omnichannel deployment. So, it’s only a matter of customizing each channel’s interactions with the corresponding conversational flows and navigation cues. In other words, generative AI enables you to set the specific conversational characteristics for each audience channel, be it email, live chat, SMS, or even voice.
Conclusion
Ultimately, generative AI boosts customer and employee experiences by introducing a more natural and human-like way to interact with the business. Moreover, generative responses make conversational interfaces more resourceful, relevant, and valuable in various use cases.
AI has made a major leap in content generation. But AI is not the only thing stirring in the enterprise tech space. Enterprise-based technologies are evolving so rapidly and drastically that keeping track of digital business solutions is becoming increasingly difficult.
But don’t worry. Your trusted IT partner, SIA Innovations, is here to sharpen your digital edge with the latest business technologies. Book a consultation with our IT experts and learn how to accelerate digital transformation through automation, cloud migration, data modernization, DevOps, and more.
