Why AI is Not Right for your Chatbot

Last week ROKO Labs had our unofficial launch for InstaBot at New York TechDay. Our team put together a great InstaBot presentation and gave away some bot related swag. We also met a lot of engaging people from all sorts of companies.

Naturally we were there to talk to anyone interested in chatbots and how Instabot could solve all of their engagement problems. We received a lot of interesting questions about use cases, implementation, and the conversation building experience. One question that seemed to come up the most was if Instabot used AI (Artificial Intelligence). The answer - it does not and I’ll tell you why your app doesn’t need it.

AI is a buzzy term in tech and a lot of folks see it as a requirement for chatbots to work for their app – without understanding their own use cases for a bot. Most of the consumer applications on the market are straightforward and have no use for AI. So my answer to those at TechDay is the same I give to you as to why you don’t need AI powered chatbots.

For starters - AI technology, specifically machine learning and natural language processing (NLP), is not ready to be consumer available. With the exception of Alexa, Siri, and Google Assistant (which have all only recently begun opening up their API to third party development) most companies do not have the resources to support the massive work that AI and machine learning require.

Communication is a complex combination of language, tone, body language, and timing. Any variation of these characteristics can alter the meaning of a statement. (This is the reason those text message jokes that I sent to my girlfriend had to be smoothed over with flowers.) Not to mention the different languages, accents, and dialects that exist – it can become a frustrating loop of miscommunication that disengages your users.

Why not just cut right to the chase and provide value to your users?

We have seen time after time that when users are able to provide open ended responses to chatbots, the value to the both the user and app owner gets diluted and unclear. Chatbots should be able to both provide user value and app value at the same time.

By building a predefined conversation with a range of values, you can quantify the results and make actionable decisions. Users are not stuck thinking about what they want from the bot, and instead they make a couple of decisions and a clear value is presented to them. Where is the AI required here?

Instabot is a little different than most bots available on the market in that applications can build a conversation directly into their workflow instead of through a 3rd party messaging service. So our conversations can engage with users in a seamless way where the results relate directly back to the UX.

Last but not least, creating the logic behind AI powered bots is no easy task. It will most likely require a developer to be involved each time a new conversation is needed. This limits the resources you have dedicated to building a great product.

So next time you think about utilizing chatbots in your app or website, think about what exactly you are looking to get from your users and decide if AI is really needed. My guess is no and Instabot will be here waiting.