Source: Google is Working on Chatbots Which Can Engage in a More Genuine, Human Conversation | Social Media Today. Google is working on the next generation of chatbots which would actually be able to hold a genuine, responsive conversation.
While there are now more than 300,000 active chatbots live on Facebook Messenger, chatbots, as a functional option, have never really gained traction the way that Facebook might have hoped.
Initially, after the launch of its Bot Platform, Facebook envisioned an all-encompassing future for Messenger bots, modeled on the success of similar tools in the Chinese market, though with expanded functionality. Indeed, former Messenger Boss David Marcus (who’s now in charge of its Libra cryptocurrency), remarked that Messenger bots would “transcend their Clippy origins and appeal to the masses”.
But they haven’t, at least not yet. Part of the problem is that people don’t seem to have warmed to engaging with bots, or brands in general, via Messenger, where they more commonly speak with close friends. But another key issue could be that the existing bots themselves simply aren’t very engaging.
Talk to your home speaker device and you’ll see what I mean – unless you say the exact right thing, and it hears you exactly right, the device won’t understand what you’re asking.
This is a key limitation that every bot and AI developer is trying to address. And this week, Google has shown a glimpse of the potential way forward, via a new bot system that it’s developing, which can recognize significantly more patterns of conversation, and hold a discussion with you to an almost human degree.
As per Google:
“Current […] chatbots have a critical flaw – they often don’t make sense. They sometimes say things that are inconsistent with what has been said so far, or lack common sense and basic knowledge about the world. Moreover, chatbots often give responses that are not specific to the current context.”
To address this, Google has built a new chatbot model called Meena, which is trained on a “2.6 billion parameter end-to-end trained neural conversational model”, and can conduct conversations that are more sensible and specific than existing bots.