Will conversational AI resolve chatbot fails
Right now, interest in the Conversational artificial intelligence AI (CAI) space is booming. Getting machines to grasp language has been a momentous challenge in AI, but all evidence points towards some important breakthroughs in this space.
Recently, for example, it was announced that Brain Technologies had raised an impressive $50 million in funding for their natural language processing (NLP) search engine solution, with other companies like Soffos also at the forefront of the race to deliver an enhanced user experience.
Yet the innovation taking place within the sector today risks being tarnished by the bad rep of traditional ChatBots. While ChatBots have been around for quite some time now, having grown from interactive voice response (IVR) systems more than 20 years ago, along the way, they have been subject to a few hiccups. Historically, these technologies may not have always provided customers with the exact answers they have needed, or floundered when faced with a more complex set of demands.
From providing seamless customer service, to delivering nuanced support employees, no doubt, recent advances in the CAI space will be good for business. However, customers and colleagues alike might take a little more convincing when it comes to adopting these new technologies.
Chatbots and Conversational AI are worlds apart
First, it is important to acknowledge the differences between the ChatBot technologies of yesteryear and the CAI technologies we are seeing today.
While on the surface these technologies may seem similar, this is not the case: architecturally, these systems are poles apart. No doubt, there continues to be a clear role for ChatBots in limited, transactional customer support activities such as paying a bill or adding a new service to an account. Indeed, many people will have had positive experiences with online ChatBots when completing routine tasks. However, their ability to mimic complex human interactions remains limited.
This is because common ChatBots are keyword-driven and rely substantially on manual curation to work, responding only to a rigid set of conversational cues and reverting a predefined set of responses. In this regard, there are human beings under the hood, so to speak, building these keywords into the system.
On the contrary, newer and more sophisticated Conversational AI systems are more convoluted from a linguistic standpoint, and have a firmer grasp of semantics than earlier, keyword-driven models. As these models rely on neural network technologies, rather than a set of inbuilt buzz words and synonyms, CAI can decode the meaning of words and consistently learns from exchanges to curate more coherent responses.
Even systems like Siri and Alexa have limited utility in the wider business context. They work from a prefixed set of mappings based on the types of questions they are likely to be asked. However, these personal assistants are not inherently capable of making abstract decisions or accumulating aggregated combinations of information to revert a simple answer to a query. Most likely, they will simply direct a user to a resource where their answer can be found.
That’s where Conversational AI changes the game. CAI can understand the context of a conversation without relying on specific phrases to offer a contextual framework, and will synthesise all the knowledge stored in its database to revert precise answers even to unexpected questions.
The result of this is that new technologies are widening the use-cases for ChatBots into domains where less structured support might be required, where they haven’t traditionally served very well.
The hybrid working revolution has created a clear need for Conversational AI
While ChatBots and Conversational AI technologies will always be necessary to provide customers with first-line support, now that more businesses are moving towards hybrid working arrangements in the aftermath of COVID-19, these technologies will be relied upon in a whole different set of contexts.
Scattered workforces and remote practices mean that organizations must find new and innovative ways to ensure that their employees are armed with all the knowledge they require to thrive in their roles, without necessarily having to rely on lengthy video calls with line managers, or gratuitous training programs.
Instead, up-and-coming NLP technologies will allow users to quickly ask questions about a product’s USP, say, and receive the answers immediately – which should be a great help to those working in the B2B sales arena, for example. What’s more, they will be able to work with open-ended questions and allow for more back and forth with the end user. If the conversational agent doesn’t understand something, it is capable of asking the user to clarify or provide more information so it can offer better support.
From a training perspective, Conversational AI will open up a whole host of new opportunities for employees. Some systems have the inbuilt capability to store exchanges in their memory bank, which means the conversational agent will be able to pick up where it left off and recall previous information to support more seamless and dynamic exchanges. What’s more, they will be able to pick up on the minutiae of conversation, interrogating knowledge graphs to understand dialect, abbreviations and even regional accents, when providing expertise – far beyond the capacity of simple keyword databases.
Ultimately, as more businesses adopt and invest in Conversational AI, the more individuals will begin getting to grips with, and eventually embracing, these technologies. In many ways, the proof will be in the pudding, but there is little doubt that these technologies will be a strong and growing presence across all areas of business in the years to come.
Nikolas Kairinos is the chief executive officer and founder of Soffos, the world’s first AI-powered KnowledgeBot. The platform streamlines corporate learning and development (L&D) to deliver seamless professional training for employees. You can register for beta here.