Until A.I. gets better, buttons are a bridge for chatbots

Facebook announced a pretty radical change to chatbots.

Instead of improving the A.I. or adding better natural language processing, the social networking giant has provided a way for chatbot developers to make things easier on users. At the same time, the Facebook Messenger platform update announced last Friday adds “quick replies” and other buttons to guide users to select the right option. In the end, it’s some hand-holding for both the bots and the users, as the world of chatbot artificial intelligence slowly improves. It’s also a nod to the wide popularity of the platform.

I’ve tested many of the recent chatbots, including a few that help with travel plans and with ordering a book or some flowers. The more I use them, the more I can see the value and the more the tech impresses me.

Yet, I’ve also watched as friends and family have tried to talk to a chatbot, and the interactions tend to veer off course. With the Poncho cat, for example, I’ve watched as people have tried to make jokes back to the cat. It doesn’t really work, because the cat is not really parsing a conversation about anything other than the weather.

What people want is the A.I. assistant from the Spike Jonze movie “Her.” What they get is sometimes a bit like an ATM machine. I’m not saying that buttons and other guides are a downgrade, because I want more developers to experiment with bots and figure out how to add intelligence to the conversations over time. And part of the equation is that humans don’t quite know how to chat with bots yet. It’s the same problem encountered by even the most powerful voice-activated assistants, like Amazon Alexa and Apple Siri. You can’t just say anything you think of to these bots, even if the options seem unlimited. With Amazon Alexa, most people start out asking for things that don’t even work, thinking the A.I. is all-powerful and can tell you the store hours for a local pawn shop or make predictions about the universe.

It’s one step forward, one slight shovel-step back.

Fortunately, Facebook is also adding some extra chatbot features. A bot can now send GIFs, audio, video, and other files to the user as a way to appear more convincingly human. You can also mute a bot in the same way you can mute a human chat or a group.

As someone who led a usability and design team for about ten years in the corporate world, I know it’s important to keep the focus on the users. If they are confused, we should help them. Sticking to an engineering formula that’s “better” or “cooler” or even “innovative” is not always the best approach. You want users to see the light, but they can’t see anything.

This is Facebook helping the developers help the users. I like the change, but it’s also a bit depressing. Buttons are controversial, I’ve noticed, because they dumb the conversation down. It’s like going back to the ’90s (or maybe even the ’70s) and giving people an A-B-C choice, which is not exactly advanced. Maybe the focus needs to be on tech that provides the answer to a button push for now, although people usually do come around and “get” what Siri and Alexa really want.

I’m curious about your view and whether you think the button options are a downgrade. Post in comments or send me an email directly to discuss and I promise to engage in a conversation.