The chatbot revolution continues, but poor quality control and confusion of terms weighs heavily on adoption.
Facebook has put chatbots at the center of its refreshed Messenger platform, creating a directory where users can easily find useful applications. But the platform is only as good as the bots created for it, and many were lacking.
The Information tested a number of Facebook Messenger bots and found they could only ” fulfill only about 30 per cent of requests without human agents,” per a Register [UK] report.
“There’s not much quality control on Facebook,” said Jordi Torras, botmaster at Inbenta, a maker of customer support, e-commerce and conversational chatbots. “The team in charge of quality control at Facebook was likely overwhelmed. Trying to have a centralized approach to having quality control is very difficult. It’s a huge task to embrace.”
The ravine between a well-designed, intelligent bot and a substandard one is quite significant.
“Many of these bots have the minimum capacity to understand human language,” Torras said. “They use a common line interface. And so they can only understand commands if you type lines exactly as they’re programmed to understand.”
“If it’s [just a] line of commands; it’s not a true chatbot,” Torras said.
Command-line bots are often the likely setup for weather and news apps.
“They just send you news that might be relevant to you,” Torras said. “They call it chatbot as well, even though there’s not chatting with you- you’re restricted to a few commands.”
Despite the gap and public perception, Torras is bullish on the future of bots.
“I believe that chatbots are here to stay,” and will evolve, Torras said. “If you look at any science fiction movie, [the interactions with machines are] always conversations. It’s going to happen – it’s in our human nature [to build these].
He added: “Forget screens, forget keyboards; the only efficient way to interact with machines will be talking.”
