Community /

Merging with AI: How to Make a Brain-Computer Interface to Communicate with Google using Keras and OpenBCI


Want to search Google without having to type or say a word?

Step-by-step tutorial content like this is hard to come by in the BCI space. Full credit to Jagveer Singh for this amazing project.

Jagveer Singh, intern at Intel Corp, created a project based on work that was done at the MIT Media Lab to construct a BCI that allows you to telepathically query Google search. He wrote an end to end tutorial for anyone seeking to replicate this, along with general information about BCIs and some BCI companies. He also included a list of some resources (textbooks, websites, organizations, Github repos) that he found helpful.

“In this project, we establish a direct connection between your nervous system and an external AI agent. This agent may be anything you can get an API for: Google Assistant, Siri, Alexa, Watson, etc. Services like Dictionary or YouTube would qualify as well, but these would limit applications to content queries rather than general purpose requests.

For the purposes of this project, we will query Google Search directly as it provides the most flexibility and is the easiest to set up. Upon completion, you should be able to query a handful of terms on Google simply by thinking about them.”

Leave a Reply