Many of us following marketing or technology news can’t have helped but notice the recent flood of customer-facing chatbots and other experimental AI solutions rolled out by players such as Microsoft and Facebook. While some of these have been met with great fanfare and high expectations, the quick and spectacular failure of others have only served to highlight the infancy of the technology.
Only a few weeks ago Microsoft released its own version of a self-learning bot on Twitter called Tay. Meant as a way to do research on conversational understanding, Microsoft intended the bot to get smarter the more you interact with it, making for an increasingly personalised experience as time goes on. Unfortunately with a lot of her interactions based on mirroring users’ statements back to them, the experiment quickly turned sour with both racist and other offensive tweets coming out of the account. Within 16 hours the bot was taken offline, proving that teaching a bot what not to say is just as important as teaching it what to say.
As for a more successful example, Facebook has been trialling its own AI-assisted concierge service called ‘M’, since August last year. Integrated in its Messenger app, the key differentiator between this service and a traditional bot such as Tay, is the AI learning from a team of human trainers. The team of trainers that is monitoring all the interactions between the AI and users 24 hours a day is effectively guiding the AI into the right responses and actions. For now this trial is limited to a select number of users in California and pending on its success it can be rolled out across a larger market.
As we see more users move away from larger social networks towards messaging apps, this approach to setting up a branded AI could have tremendous commercial potential – by never straying away from the appropriate tone of voice, the AI will always effectively communicate a brand’s persona. Eventually you might even feel like you have a personal relationship with a brand.
On a much larger scale, by including the wealth of personal data available from either Facebook or Google, AI integration across the whole customer experience can represent the perfect merger between brand marketing and commerce. A branded AI could personalise branded content and then automatically serve it when you are most likely to engage with it. Taking the lead from Facebook, by having the AI shadow conversations by everyone from customer service representatives to sales executives, it can not only learn how to provide excellent assistance to customers, it can also learn to determine when intent to purchase is the highest and then finally provide the necessary sales trigger to secure the sale. This would be Brand Commerce in its most perfect form.28.04.16 Archive