Artificial Intelligence

Google Backs Away From Chatbots

A not so very long time ago, chatbots were forecast to be the next big thing in digital ecosystems. Starting with Facebook’s big announcement of its chatbot push at F8 in 2016, all of those forecasts for everything chatbots were going to do starting rolling in. And the consensus across media was that chatbots were basically going to do everything in the mobile ecosystem: replace apps, transform eCommerce, fix B2B transactions, save shopping malls – if you can name it, someone probably at some point between 2016 and 2017 predicted that a chatbot was going to do it.

“The media has declared chatbots the digital version of the little black dress: a technology staple that every brand must now have and every payment type must now commerce-enable,” Karen Webster wrote about six months into the great bot binge of the extreme excitement gunning up in the ecosystem around them.

And while some of the enthusiasm was perhaps understandable, she noted at the time, given the massive and overwhelming example of the success of WeChat, she was wondering if perhaps given the massive differences between the American and Chinese consumer ecosystems, it might make sense for everyone to slow their collective roll on the chatbot enthusiasm.

It could work, she noted. But given the massive changes U.S. customers would have to make to their digital habits to make a chat app the center of their digital lives instead of their home screens – combined with the fact that early chatbot interactions weren’t really that good – it also seemed possible that enthusiasm was getting a bit overblown.

A year later, by the fall of 2017, as consumers’ adoption of the chat economy was somewhat slower than anticipated, the tune being sung by bots’ biggest boosters had started to change. Head of Messenger David Marcus admitted that early phase of botdom has been a bit primitive — and that it has been too easy to build and release a bad bot. Head of Product for Messaging Stan Chudnovsky noted that bots were more a building block than a complete offering.

“Chatbots were always a means to an end, not an end. Our goal was always to enable meaningful and useful conversations between people and businesses. Bots were a means to achieve that goal.”

Now, it is the fall of 2018 – and these days, no one says all that much about chatbots one way or another. Mostly when people refer to “chatting” with an AI, they mean they are talking to Alexa, the Google Assistant, Siri or maybe even Bixby. They are almost certainly not talking about a Messenger bot.

In fact, the most we’ve heard about chatbots in fall of 2018 has come in the form of a negative vote – Google has officially declared defeat when it comes to trying to make chatbots a thing.

The world first met the Google Assistant in the context of the Allo messaging app, an offering Google rolled out so that users could text their questions to Google as though it were an old, extremely knowledgeable friend.

It never exactly found a wide audience.

Since then, the Google Assistant has literally found its voice, moving on to Google Home and other connected devices. Allo, according to media reports, has been officially shelved. The version of the Google Assistant on Android phones (and reportedly coming soon to iPhones) now responds to voice cues, and emphasizes interactive visual response cards for users.

“When we built the Assistant, you can clearly see inspiration from Allo in what we did, in this chatty back-and-forth model where you’re talking with an intelligent assistant,” said Chris Perry, the Google product manager who leads Assistant on Android. “And we found that was somewhat restrictive of a model for us. It ended up constraining us in a number of different ways.”

While focusing on visual elements of the conversation is useful, Ye-Jeong Kim, Google’s user experience manager for Search and Assistant, noted that keeping track of text bubbles isn’t as relevant to the user’s experience as telling and showing them information.

“Unlike spoken or written conversation, visual doesn’t have to be so ephemeral,” Kim said. “It’s lingering, and helping to aid a conversation.”

Moreover, Kim noted, as Google’s Assistant is branching out across a wider range of devices – cars, appliances, smart displays and televisions, to name a few – a chat-like interface is impossible, impractical and largely undesirable for users.

“We wanted to build a framework that can actually expand and be more fluidly adaptive to the various contexts you’re in,” Kim said.

Some of the more helpful elements of chat can be preserved, Kim noted: Users can scroll through previous questions they’ve asked. But the goal, Perry noted, is to make the Assistant more efficient, so that long chat transcripts aren’t created as the user is trying to lead the technology to the right answer.

“It’s not this back, forth, back, forth, back, forth,” Perry said. “It’s an immersive experience inside the canvas itself.”

But it’s an experience, he noted, that still needs work. The lesson of chatbots for Google is in acknowledging that while the potential to unlock conversational capability with technology exists, cracking the code on that killer use case for working with virtual assistants is still ongoing.

“It feels a lot like 2009 for me,” Perry said, “where we’re building apps, we have this new platform and everyone’s kind of trying to figure out how you should do things.”

It’s not about trying to wholesale replace apps, as was the previous hype built around the chatbot economy, but about finding the places and interactions on consumers’ phones that aren’t optimized as they are today as an app – that could be through a voice-based digital assistant.

And that goal, Perry noted, will be difficult as Google – and Amazon and Apple and Samsung – all face a common problem when it comes to recruiting both developers and users to their platform. As of right now, the regular app ecosystem is a better-known commodity that is easier for developers to stick with, as opposed to the app ecosystem where the paths to monetization are somewhat less clear. And consumers – though coming around to voice assistants slowly but consistently – are still fairly hardwired to using apps, meaning Google has to push a behavior change in them as well.

But Google remains hopeful, mainly because they believe that voice-based chat interfaces have something that their text-based counterparts lack. They are actually useful, and useful in contexts where apps aren’t.

“I want users to be able to build a mental model of this relationship with Assistant,” Ye-Jeong Kim noted, “so they can actually ask it harder things than just ‘Turn on the lights,’ ‘Set the alarm’ or ‘What’s the weather?’”

TRENDING RIGHT NOW

To Top