When a human has been taken for a bot
My story begins one week ago. At SAP Conversational AI, the startup where I work, we opened up a Slack to welcome our community, so they could ask questions and give feedback. Last week one of our interns, Marian, started welcoming all the users arriving on our Slack, by saying “Hello” at every first connection 🙂 And very quickly the first jokes started appearing: “Marian, the bot”.
The day after, some users even asked us for the link to install “Marian the bot”..
Maybe you’ve heard about chatbots, and maybe more so in these last few weeks. Microsoft released its bot framework, Facebook opened its Messenger platform to custom bots, and so many startups have been popping up in this incredible bot world.
Before going further, let me briefly explain what a chatbot is. It’s a service powered by artificial intelligence, that you interact with via a chat or messaging interface.
This service fits into your daily life and responds to many different needs, ranging from very precise and useful things to fun things. It can live in any major messaging application like Facebook Messenger, Slack, Telegram, Text Messages, etc…
What’s happening with our intern is that in people’s minds, bots are not so smart; they only understand what you are saying if you say the right command. They respond with the same sentence and they don’t do humour or irony.
So when Marian started to repeat the same expression in the same context, people detected the pattern, the automation. And when you think automation, you think machine and bots.
Conversational VS Non-Conversational bot
Remember the first chatterbot, a program that gives the impression of conversing like a human. It only works with some keywords extracted from the user input but doesn’t understand or analyse the user’s input to provide an authentic conversation.
Today, real conversational bots exists, those which are powered by artificial intelligence.
No need for particular syntax or words when you interact with them.
Thanks to artificial intelligence, bot conversation has come very close to human conversation. It’s more natural for the user, and the gap between bot and human conversation is closing.
So at this stage, I have a question for you. Would you prefer to know if you chatting with a bot or not? What impact will this issue have on the service and the user experience?
Humans are ready to forgive bots
Many services and startups provide assistants or bots with this headline: Human-supervised artificial intelligence. This begs the question: am I speaking with a bot, or a human? These services always arouse suspicion and it’s exciting to solve the mystery. The user’s emotion at this point is really interesting, and in itself is a factor that contributes to a great user experience.
What if you know you’re chatting with a bot? You will probably be more demanding, it’s a process, a service. You are likely to have less empathy than when you talk with a human, and maybe you will be more frustrated if it makes mistake. But the bot is still learning, and I think people are willing to forgive. It’s like a child, he’s allowed to make mistakes, he will learn. This image of a smart funny child is really important.
Bots can, with a touch of humor, and with great conversation, impress users. Like a child who suddenly did something very smart and whose parents are very proud.
On the other hand, a bot that tricks us into thinking its a human, by adopting human characteristics and not revealing its true identity, will seriously disappoint the user and ultimately provide a poor experience.
What happens when you speak with other humans?
You share your feelings, your wishes, your understanding of the world. If you chat with someone you think is human but in fact isn’t, or a bot with a human “appearance”, like a little avatar with a human name, you will be extremely frustrated.
Bots must adapt to humans, but should definitely not pretend to be one.