If you’ve ever chatted with a chatbot, you’ve realized they lack a certain…awareness, especially when you’re getting confused or pissed. No surprise that Gizmodo writer Darren Orf describes chatbots as “frustrating and useless” and likens conversations with them to “talking politics with a toddler.” When you feed a chatbot lines like “this is stupid” or “that doesn’t answer my question,” most of them will respond with lame, repetitive answers like “I don’t understand.”

Surprisingly, these robotic answers are easily to fix by adding emotional awareness. While you may feel daunted by the task of granting a computer program emotional sensitivity, a trait many humans lack, the reality is that plenty of powerful services exist to simplify the process. We’ll show you how we built a caring robot with just 150 lines of code using 4 different IBM Watson APIs.

Granting technology emotional sensitivity has many benefits. Users feel understood and acknowledged, leading to superior UX and less frustration. In customer service or digital health contexts, emotional awareness can make or break your bot experience.

Here’s how we built our caring robot:

  1. Create a simple conversation using Watson Conversation to solicit the user’s feelings.
  2. Use Watson Speech to Text and a microphone to turn user voice input into text.
  3. Run the text in  Watson Tone Analyzer to extract emotions, personality traits, and other attributes.
  4. Have TJBot speak an empathetic response using Watson Text To Speech.

 

The Creation Story

Watch our video to learn how we built both the software and the hardware components of our emotional robot:

 

The Photo Evidence

Here are some photos we took along the way while building TJ Bot:

Emotional Robot

tjbot_halfway_web

tjbot_almostdone_web

tjbot_me_web

 

The Tutorials

If you’re interested in building the robot yourself, check out these tutorials written by Josh Zheng from IBM Watson.

Part One: Write The Code

Part Two: Set Up The Hardware

Did you make your own caring & emotional robot from our tutorials? Send us pictures or links to your experience in the comments below!