Facebook addresses the status of bots on Messenger
When Facebook introduced an API for developers to build purpose-driven bots in Messenger last year it gained tremendous hype. That led to almost immediate and sustained criticism about the capability and viability of bots overall, but Facebook now is trying to reframe how bots are perceived by the public and somewhat downplay their utility.
“We’ve tried to be careful in terms of how we will be talking about bots,” Chudnovsky said. “Everyone expected a human-like conversation which was hard for developers to do. Expectations were high and that led to some people’s disappointment.”
Bots were never the end goal for Facebook, but rather a means to an end, according to Chudnovsky. “The plan was always to connect people and businesses, and in many cases there is no bot needed.” The user who bought solar panels on Messenger in Asia never engaged with a bot, for example, he said. Businesses can build bots to increase efficiency, particularly on simple questions such as hours of operation.
“They’re just a tool, they’re not necessarily an end,” he said. “When you start thinking about them this way, they’re not a disappointment necessarily. From that standpoint it’s a completely different story and that’s what seems to be working right now.”
What’s next for M, the human-powered AI assistant
M, which Chudnovsky describes as a human-powered artificial intelligence assistant, is also a bot-like interface that Facebook has been developing and piloting with a smaller user group for almost two years. Last month Facebook announced that M is becoming a platform in its own right and select developers are starting to build upon that framework.
“The plan with M was that we can’t yet do everything perfectly for everybody so we’re going to do everything for a small group of people,” Chudnovsky said. When M is enabled for all Messenger users it could eventually replace the more rudimentary or underperforming bots, he added.
M is purely text-based, which makes it stand out and seem less advanced than personal assistants from Amazon, Apple, Google and others that operate via voice. Chudnovsky described a three-step funnel to explain why Facebook hasn’t jumped on the voice-controlled bandwagon of late. Voice introduces complexities and uncertainties that will undoubtedly lead to lower quality outcomes, he said. Focusing on text recognition and delivering suggestions based on specific requests also gives M the ability to not react when it doesn’t understand the user’s request.
“If you don’t have to act on voice every single time -- when M cannot make a suggestion it doesn’t make any suggestions because it doesn’t know what’s going on,” Chudnovsky said. “We don’t want to go in the world where we teach people what we can’t know and do well. We want to do whatever we do very well and build up. Otherwise we’re going to be in a world where people very quickly realize certain things that we don’t do well yet and then they may not give us another try.”
Finally, when Chudnovsky was asked if Messenger will eventually incorporate voice control into the M assistant, he confirmed that to be the case without providing any further detail. “Voice eventually,” he said.
Sign up for CIO Asia eNewsletters.