Ever felt frustrated by a chatbot that just doesn’t seem to get what you’re saying? You’re not alone. Traditional chatbots often struggle with natural language and can misinterpret even simple questions. But there’s a new sheriff in town: Generative AI (GenAI). This cutting-edge technology is expected to revolutionise chatbot interactions, making them more natural, engaging, and effective.
GenAI improves our understanding of user Intent
While traditional conversational AI can handle basic, straightforward questions, it often falls short when things get complex or require context. Simon Kriss, Chief AI Officer at simonkriss.ai, comments, “GenAI may not provide any better understanding of a user’s intent in the first circumstance. Conversational AI can do a great job of understanding natural language. The difference occurs when something is not clear or the customer wants to engage in a true conversation”.
“Conversational AI is built on traditional AI models (RNN or CNN technology mostly) and therefore tends to handle questions one at a time, treating each one as discreet. By comparison, Generative AI uses all of the context of the entire discussion to answer a question (much like a human does). So, if your call types or intents are simple and straightforward, then Conversational AI will suffice. If your product or service needs clarification or has complexity (as most do) then you may need to consider Generative AI”.
GenAI analyses the entire conversation, just like a human would. This allows it to understand the user’s true intent, even if their phrasing isn’t crystal clear. Imagine asking a chatbot about nearby stores and then following up with “Which one is closest?”. GenAI, with its contextual awareness, would understand that “one” refers to a store, not a product. This context-driven approach makes interactions feel more natural and eliminates the frustration of robotic, one-question-at-a-time responses.
Beyond intent – Creating engaging conversations
GenAI’s power goes beyond understanding intent. It can also be used to craft natural and engaging responses. Traditional chatbots often sound stilted and impersonal. GenAI, however, can use the conversation history to tailor its responses, making them more relevant and even humorous (though it might not understand the joke itself!). Remember the store example?
GenAI could respond with “Sure, the store on Elm Street is closest to your current location,” instead of just providing a list of stores. This adds a layer of personalisation that makes the interaction feel more human-like. Kriss advises, “This is how humans communicate, we use heuristics (shortcuts) that depend upon context. This is why most current chat bots feel stilted, clunky and non-human – treating each question discreetly without context will always feel strange to us.”
Selecting the right GenAI model for your chatbot
While companies may not directly choose specific GenAI models, understanding the selection process is helpful. Most will rely on vendors who use pre-trained models best suited for the task. These models often work in tandem: a large language model for understanding user intent, a knowledge base for retrieving relevant information, and a smaller model for crafting the final response, ensuring brand voice and tone.
Open-Ended Questions? No Problem for GenAI
When faced with open-ended questions, GenAI’s vast knowledge base proves invaluable. Trained on massive datasets of text and code, these models possess a remarkable understanding of human language. This allows them to comprehend user intent and generate informative, comprehensive answers, even for complex or unexpected queries.
Kriss comments, “GenAI models are like super-powered language learners. They’ve been trained on a mountain of text data, dwarfing what any human could read in a lifetime. This makes them exceptional at understanding language, but it’s like learning a language in a classroom – they miss the cultural subtleties like humor or sarcasm”.
“GenAI’s advantage lies in its ability to grasp human intent. Imagine a student who aced an English exam but wouldn’t get a joke down the pub. That’s GenAI – it can use its vast knowledge to craft the perfect response, be it informative, serious, or even seemingly humorous (though it wouldn’t understand the humour itself)”.
Personalisation
GenAI holds immense potential for personalisation. Imagine a chatbot that not only remembers your past interactions and preferences but also adapts its communication style to your liking. Do you prefer a playful tone? GenAI can handle that. Need a more formal approach for serious inquiries? It’s got you covered. This level of personalisation, moving beyond “mass personalisation” to “mass individualism,” will redefine the way we interact with chatbots.
Chatbots often get a bad rap for being frustrating and unhelpful. Kriss observes, “Let’s be honest: many organisations turn to chatbots with cost reduction or deflecting calls in mind. While these goals are valid, they shouldn’t be the sole motivator. Organisations need to consider the bigger picture. What will happen to freed-up human resources? Will this lead to a genuine improvement in service quality? Can we finally move beyond metrics like Average Handling Time (AHT) and focus on a richer customer experience? Ultimately, AI adoption should be about creating a win-win for both businesses and customers. Transparency is key – we need to be clear about the “why” behind our AI choices”.
Instead of solely aiming to deflect traffic and reduce costs, organisations should leverage AI to genuinely improve the customer experience. This means being transparent about AI adoption and utilising freed-up human resources to offer superior service. Ultimately, the goal should be a win-win situation for both customers and businesses.