AI: useful tool or amusing toy?

What does Artificial Intelligence potentially mean for brand personality? Rosanna Beart looks further.

Have you ever asked Siri stupid questions just to see their response? Asked them to be your friend? Said you love them?

That last one might be just me, but I'm guessing most of us have tested the 'off-script' abilities of an artificial intelligence (AI) platform before. Because we don't just accept a bot for the functions we know they can perform – we like to see what sprinkle of personality they might also have.

This opens a big door for brands. When they create bots that help users perform tasks or buy products, they can add a distinct personality and tone of voice to their bot to enhance and strengthen the brand experience. With a chatbot, not only is there a direct channel to a consumer engaging with your brand, but it is also facilitated by a bot – and who could be more consistently on brand than an actual robot?

But this is the hard bit. The opportunity to display personality in a chatbot is pretty much limited to greetings, emojis, jokes, and error messages. Distilling everything you want your brand to be and stand for into this 'microcopy' is challenging. To tackle this, companies are hiring writers to create what is essentially a fictional brand character. From comedians and satirists to novelists, poets and playwrights, the language team craft how the bot reacts to the user and with what language. This is being termed 'interactive scriptwriting' and Seth Greenfield, co-founder of Imperson, a company that specialises in building these bots for brands, predicts it will soon become a profession taught at universities.

So there is a lot of investment in building brand characters – but these characters need to be authentic to the established brand. If, for instance, the McDonalds bot was a clean eating freak pushing the healthier options as you order, you wouldn't identify with the McDonalds you know and love. If the CNN bot was juvenile and gossipy, you might be a bit confused.

Experimenting with some existing bots, this doesn't seem to be too much of a problem. If the bot is built by an internal team, it is likely to reflect the internal culture of the company, which should naturally reflect the brand. The brand has filtered down to the bot almost by default. For example, Slack's Howdy is very self-deprecating when he can't answer, with quips such as "I'm just a bot – I don't even have arms", whereas Facebook Messenger's M, a personal assistant-type bot, will never admit it can't do something. Intentional or not, to me this difference directly reflects the attitude of their respective real life brands. A smaller, newer, still learning, start up versus Facebook, the dominating tech conglomerate that is everywhere and can do anything.

Whether by design or default, these two bot personalities seem to accurately reflect their respective brands. But will that personality appeal to everyone? The user might be getting a consistent and engaging brand experience, but do they actually like their bot? Even though we know a bot is a bot, because a bot behaves like a human, humans automatically treat it like one. We develop an emotional affinity, which ranges from the millions of us that needlessly say 'thank you' to Siri, to soldiers holding funerals for their bomb disposal robots. We are almost incapable of being indifferent when something is doing something for us, so liking the bot or not is probably inevitable.

This is why the people at Slack are building personalisation settings into Howdy. Sound familiar? Yep, they are trying to replicate TARS from Interstellar, the robot who Matthew McConaughey engineers to be 90% honest, 65% humorous etc. Soon, Howdy users will be able to make their bot more funny, less polite, less patient, more arrogant – whatever they want. This is incredibly complex language development, where the full range of microcopy needs to be nuanced for an almost infinite range of personality combinations. But aside from massive financial investment – does this personalisation also implicate the brand experience?

Right now, I can't personalise Howdy. This means he gives me a consistent brand experience, builds the Slack brand, and strengthens my relationship with it. I probably like Slack more as a company. But then again, I'm already getting bored of Howdy, his jokes, his Monday morning enthusiasm and his slightly 'dude' attitude and soon it is likely I will stop engaging with Howdy and the Slack brand altogether.

On the other hand, if I could personalise Howdy, chances are I'm going to like him more, use him more, and therefore continue to interact with the brand more. All this is good news for Slack.  But that bot would then have the personality I want, not the personality of Slack.

Am I really building a relationship with Slack, or did I just create a robot friend for myself?