Creating the best AI chatbot means understanding AI

Creating the best AI chatbot means understanding AI

SnatchBot TeamSnatchBot Team, 13/03/2019

Creating the best AI chatbot means understanding AI


It was only a matter of time before the acceleration of computing power towards the end of the last millennium led to a computer that could beat the best humans at chess. In fact, the issue was not so much could it ever happen, but whether it would happen before the year 2000. And in 1997, Deep Blue beat Gary Kasparov. This seminal moment did not provide evidence that humanity had created artificial intelligence. Complex – and beautiful – as chess can be, it is still a game with narrow set of rules and distinct boundaries around the set of successful strategies.


This is why Alan Turing was surely closer to defining the tipping point in terms of artificial intelligence as being the ability of the computer to comprehend and react to language with sufficient skill that a human could not tell it apart from other people. A much better game test for AI then would be Diplomacy, since it relies on negotiation between the players as much as smart tactics. And interestingly, despite some efforts to create computer programs that play the game, AI is nowhere near being able to master the game.


Do you need a chatbot that can pass the Turing Test?


In business, there is a different definition of AI, a much more practical one than theoretical. It is the automation of sentence analysis and here, while Natural Language Processing (NLP) has made very rapid advances even in the last few years, the point is to create operational software that delivers return on investment, rather than take a step towards self-awareness among machines. In every actual use-case in 2019, an NLP model that can, for example, identify with near-human accuracy (which is now possible) that a customer has lost his or her luggage and needs assistance is more effective than one that is attempting to impersonate a human and only succeeding to a limited extent.


It’s a case of not needing a (costly) sledgehammer to crack a nut.


In this sense, SnatchBot’s AI is state-of-the-art. Our NLP engine represents a major investment by us, but it is the perfect tool with which to build chatbots that deliver immediate ROI. You can hone in on the exact topics that you need the chatbot to understand and you can do so without having to employ people with coding skills. Just follow our guide and train your NLP models, or ask us to do this for you.


This is why SnatchBot recently appeared in a Forrester report: Now Tech: Chatbots For IT Operations, Q1 2019.


Below, is an infographic based on a design by, which details the milestones and indicates the future of AI.