Social networking giant Facebook is always on the top of its game when it comes to dishing out features that would help users with their Facebook experience. This time, the Mark Zuckerberg led organisation has successfully unveiled ‘human-like’ bot. Yes, you read that it.
Most of the times, our problems with bots, is that they churn out really organic replies in a really organic way, without any change in facial expressions. So, whether it’s a sad conversation or a joyful conversation, their expressions would be the same. Taking cue of this problem, Researchers at Facebook’s AI lab decided to work on a ‘human-like’ bot. In order to give the bot its human touch, the researchers controlled it by an artificially intelligent algorithm which trained after it watched hundreds of videos of human Skype conversations.
According to New Scientist web portal, Facebook’s AI lab researchers decided to train the algorithm on Skype so as to make it understand, learn and then mimic the way humans adjust their expressions in a one-on-one conversation with each other.
From the Skype conversations, the algorithm grasped and learned to do normal human acts like naturally producing blinks, nods and doing various mouth movements in order to show that they are fully immersed in the conversation that they’re having with a particular person. Not only this, Facebook’s AI researchers have also made the bot learn to speak like human beings and choose in real time what would be the most apt facial response depending on the nature of the conversation it is having with a person.
In order to make the best effective use of its learning, the algorithm decided to divide the human face into 68 key points that it then monitored throughout each Skype conversation. The data collected from this exercise would soon be presented to the world by Facebook at the International Conference on Intelligent Robots and Systems scheduled to take place in Vancouver, Canada, from September 24-28.
Facebook’s human-like bots have tested out well. In order to see how effective the bots are, the Facebook team decided to test the system with panels of people who watched animations that included both the bot reacting to a human, and a human reacting to a human. The final verdict was, that the volunteers tested out that both the bot and the human were equally natural and realistic.
Prior to this, in July, Facebook’s AI efforts had hit a temporary bump when it had to shut down one of its AI systems after chatbots started speaking in their own language rather than in the codes that were fed into them. What made everyone nervous in the world of tech
It wasn’t that the AI became destructive and went on a spree to shut down computers all around the world or anything of that sort, but what made people in the tech world nervous was the fact that it just stopped using English and started conversing in a language that it had created. Fearing an AI uprising was in the vicinity, Facebook decided to pull the plug on the project immediately.
Like this content? Sign up for our daily newsletter to get latest updates.