It’s been over a month since SXSW 2016 concluded, and…well…we miss it. That’s why—as you’ve probably noticed—we’re revisiting all of the epic projects we showed off to the world back in March. The press already had the opp to test out and write about some of the amazing things we cooked up in our innovation lab, but don’t worry…we’ve got details for you here. In this post, we’re shining the spotlight on a particularly interesting use of artificial intelligence: Mockzy AI. Read on…
Mockzy AI draws on content produced by notable public personalities, living and deceased, to animate a speaking, AI-driven avatar of the individual. This gives you the chance to converse face-to-face with a famous—or in some cases, infamous—person.
Have you ever wanted to have a one-on-one conversation with a politician, celebrity or scientist? Well you don’t have to battle a bodyguard or figure out time travel—with Mockzy AI, we’ve got you covered.
Mockzy utilizes publicly available content produced by select figures, ranging from tweets to interview answers to written materials and speeches. This content—along with machine learning, tone analysis, and contextual relevance within IBM Watson’s AI services—is used to build an AI-driven avatar of each person. This allows users to ask any question to the avatar and get a response based on that person’s publicly generated content. The result? You finally have the chance to talk to Ben Franklin…which we know is something that’s totally always been on your bucket list.
Here’s the tech that makes the magic happen:
- Twitter API – Uses a Twitter handle to access and pull Tweets (if utilizing Twitter).
- Watson’s Concept Insights – Analyzes each sentence from a selected text and relates each sentence to various concepts, assigning a score. These concepts are then linked to responses, creating documents (serves as the reasoning).
- Watson’s Retrieve and Rank – Stores the answers and handles the thinking part of the conversation, providing answers to questions (serves as the memory and response).
- Watson’s Text to Speak – Translates that response from text to speech (serves as the “mouth”).
- A custom interface – Serves as the user’s way to ask questions and initiate conversations.
While Mockzy is currently focused on giving users the ability to chat with notable figures (and could, of course, allow them to talk to avatars based on living celebrities), there’s also the potential for this technology to be used in a more personal way. For instance, what if you had the opportunity to talk with the past or present version of yourself?
This technology would allow users to have a full-blown interactive conversation with themselves based on content from their social media accounts over the years, which would be…enlightening, to say the least.