Chatbots And Commerce

Chatbot Tracker: Facebook’s Plan For Meaningful Conversations

While chatbots tend to make almost anything customer-related more efficient, the technology behind them is still in its infant stages. With voice-recognition software like Alexa and Siri moving the ball forward on interactive speaking, consumers’ demand for similar interactions with chatbots may be on the rise.

As we reported earlier this year, it seems that Facebook is becoming the central conduit for the advancement of chatbots via its Messenger integrated offering announcement. After welcoming companies to build out their own version of chatbots into its Messenger application, Facebook is taking the chatbot industry one step further.

The social media giant just announced last week its plans to set up a new training ground with data and programs to assist artificial intelligence researchers and developers to not only build out new chatbots, but to learn from one another in a community space. It will also enable these two groups to continually build on top of others’ work to help push the chatbot initiative forward more quickly. Dubbed ParlAI, this new open-source offering comes directly from the company’s artificial intelligence research arm Facebook Artificial Intelligence Research (FAIR).

For this new offering, the goal is to make text-based conversations with consumers more meaningful via an amped up human-esque interaction. Whether it’s trying to figure out what movie to go see, trying to get feedback on a new product or requesting an interview with someone at a company, chatbots have a myriad of ways in which to help consumers. Facebook AI research scientist Jason Weston commented on the importance of better understanding text-based lanugage and how this project will help pull pertinent information out of respective silos and into a wider pool of understanding.

He said, “Researchers often focus on one of these things alone, and that could be a fundamental mistake. We need to look at dialogue as a whole, and so what we’re trying to do in this new software platform, ParlAI, is to put all these things together and unify this research.”

The 20 different data sets and tests offered through ParlAI will help move the ball for conversational AI (at least that’s Facebook’s hope). Through these natural language data sets, the hope is that machine learning will help enable chatbots to broaden their focus when interacting with consumers. As ParlAI is integrated with Amazon’s outsourcing Mechanical Turk, developers will have the ability to have feedback from people on how to best train chatbot dialogue.

While the ParlAI offering is currently a standalone system meant for research purposes only, there is the potential to have it integrated into Facebook’s Messenger system, which could in turn significantly reduce consumer irritation with robotic company responses.

With 60 percent of adults using online messaging, it’s likely that this new open source research offering will be the beginning of what AI and machine learning is truly capable of accomplishing. As we reported just last week, the chatbot industry is set to save $8 billion annually by the year 2022. Through ParlAI, it may be safe to say that this initiative has helped propel that research into reality.

Although there’s much excitement around chatbot development, research shows that 56 percent of consumers still prefer to speak with a human over a chatbot. Given this fact, it seems that Facebook is on the right track to ensure more meaningful conversation occurrences with its new ParlAI research community announcement.


Featured PYMNTS Study: 

With eyes on lowering costs to improving cash flow, 85 percent of U.S. firms plan to make real-time payments integral to their operations within three years. However, some firms still feel technical barriers stand in the way. In the January 2020 Making Real-Time Payments A Reality Study, PYMNTS surveyed more than 500 financial executives to examine what it will take to channel RTP interest into real-world adoption. Here’s what we learned.

Click to comment