woman on laptop

Among several mildly frightening future innovations mooted at Facebook’s F8 Developer Conference this week (5G internet for the entire world using planes, VR glasses, and a section which, really, was just titled ‘drones, satellites, lasers’), one of the most realistic – and interesting from a grocery retail perspective – was the idea of chatbots.

Put simply, chatbots are to customer service what mechanisation was to the factory floor – an innovation for replacing a human-led workforce with a terrifyingly efficient inexhaustible army of artificial intelligence.

Facebook thinks it’s getting close to a chatbot that won’t annoy customers during interactions with brands, and while there are still a good few teething problems, Facebook certainly has the resources to iron them out. Shopify, a platform used by circa 250,000 small businesses to sell their wares, clearly sees this potential too, just this week snapping up Kit CRM, a start-up specialising in “conversational commerce”.

Brands have already started messaging people within Facebook’s billion-user-strong Messenger app, and you’d have to be crazy not to see the potential for fmcg companies to save on customer service and deliver a cleaner, more efficient CS experience without all the foibles that come with human interaction.

Brands should beware, however, giving over customer service to machines.

A consumer is relatively unlikely to stop and think whether or not a human or a robot produced their £3 frozen pizza. In fact, it’s almost become part of the accepted bargain of cheap products that machinery was used heavily in the production process. It’s also easier for consumers to ignore because they only deal with the end product.

With a lot of CS interactions, on the other hand, the consumer is already mildly annoyed and wants some form of redress. Certainly, there are those consumers who contact a business to ask a question like “what are your opening hours?” or “do you have product X in stock?”, and undoubtedly these questions could be easily dealt with by AI. These are probably the people Facebook founder Mark Zuckerberg had in mind when he said at the conference: “We think you should just be able to message a business the same way that you message a friend”.

Say my £3 frozen pizza contains a scrap of metal or plastic, though. When I message a company to express my concerns with their product and request, at the very least, a refund, do I want human empathy and understanding, or a stock response to a complaint from artificial intelligence? I’d like to think that my complaint had been seen by human eyes, at the very least. Imagine the damage if, like so many current chatbots, the AI doesn’t quite get what I’m saying and gives an irrelevant or incorrect response to my query.

These sorts of negative interactions with a brand could be enough to drive customers away – not just once but for good. Just think of what happened with Microsoft’s accidentally racist attempt at a chatbot.

The same human fallibilities that make a chatbot so appealing also mean a company can often repair the damage from a bad CS interaction with a single representative. Letting an AI loose on the system doesn’t offer the same protection.

The future’s coming but it’s not quite ready yet, and any brands looking to mechanise their CS should wait until they’re sure there’s a perfect solution. Of course, by the time AI is perfect, we’ll probably have a few more pressing, Skynet-esque concerns to deal with.