Quest Bridge Updates

2019 Q1: Software Prototypes Completed for Five Labs

By Nicholas Roy, Joshua Joseph and Cynthia Breazeal

April 16, 2019

As the spring semester draws to a close, we’d like to report on our progress to date.

Bridge platform development:

  • We have been developing Jupyter notebooks to help MIT research labs automate time-consuming tasks and accelerate their research. So far, we have prototypes for five labs, with tasks that include: discovering causal relationships in gene regulation; identifying the types of cracks created when carbon laminates are put under stress; generating synthetic data to complement small data sets; integrating the tool, GANpaint, into the coding platform, Scratch, for K-12 education; and inferring the pose of soccer players for modeling player performance.  
  • We continue to focus on building prototypes and demonstrating what the Bridge can offer to MIT researchers. We are waiting for resources to take on new projects. 
  • We moved into our new offices in NE36, and are now also hosting the undergraduate club, the Machine Intelligence Community.

Bridge support of students and faculty:

  • We began distributing Google Cloud credits in mid-October and have now given away the full donation to faculty and students.
  • We have started building a similar model to distribute IBM cloud credits, and have moved three groups from Google to IBM.
  • This spring, 18 UROP students are working with us to build prototypes and other tools.

Progress on Bridge mission to bring AI into K-12 classrooms

  • We have started building a portal and AI curriculum for K-12 students under our “Big Ideas of AI” initiative that integrates an online mentor-training course with Scratch 3.0, Hal Abelson’s App Inventor and Jupyter Notebooks. Bridge engineer Katherine Gallagher is working with a team of UROPs to build the portal. 
  • We have begun building a website to support communication and fundraising under our “Innovating learning and education in the era of AI” initiative covering K-12 and vocational AI education. MIT Open Learning will host the site and Quest will link to it.
  • We’ve started collaborating with i2Camp to develop a one-week AI curriculum targeting 5th and 6th graders and covering AI concepts and ethics. We plan to pilot the course in about 10 Boston-area schools in October. 

Goals for 2019 Q2:

  • Continue to focus on building workflows for campus, with the goal of achieving five operational workflows by the end of June, as well as hardening and modularizing components that support multiple workflows.
  • Streamline management of Google and IBM cloud credits.
  • Finish a K-12 portal prototype, and launch website: “Innovating learning and education in the era of AI.”
  • Provide support for the use of IBM’s donation to MIT of 5 computer racks modeled after its Summit supercomputer. 

2018 Q4: Getting Started!

By Nicholas Roy and Cynthia Breazeal

December 24, 2018

It’s been a busy four months since our launch in August. As the year draws to a close, we wanted to provide an update on our activities so far.

Bridge platform development: 

  • Welcome to our first software engineers Josh Joseph and Katherine Gallagher!
  • We have developed a roadmap for the Bridge platform of tools and services, which we will present to Quest leadership next month. 
  • We have met with labs and units across campus to assess their machine learning needs, and identified five labs in which we can make an immediate impact. We are now working with two of them.
  • The design and prototyping of the Bridge’s services and workflows is underway. We have conducted in-depth user interviews and begun collaborations with UX designers at both IBM and the MIT Libraries to ensure the services we build and provide will be useful. Our goal is to complete the first iteration of design by mid-January. We expect to put three workflow prototypes in front of our users by end of January for feedback and revision. 

Bridge support of students and faculty: 

  • We began distributing Google Cloud Platform credits in mid-October and have given away about half of our credits to faculty and students.
  • We have also worked closely with the Google Cloud engineers to build significant infrastructure to handle the distribution and monitoring of credits as well as guided support for AI researchers. 
  • We are still working with IBM to establish a mechanism for distributing their cloud resources to campus.
  • We have established a collaboration with an undergraduate machine learning club, the Machine Intelligence Community. We ran one training session with MIC in November and a research blitz in December. A training session is planned for early February with monthly meetings thereafter. 
  • We have 10 UROPs currently working with Bridge engineers to build prototypes and other tools. This team will grow in the new year. 

Bridge first mission: Bring AI into K-12 classrooms

  • We set as our first mission the goal of bringing AI into K-12 classrooms. We have formed an advisory committee comprised of academic experts and AI industry leaders.
  • We are designing a portal and AI curriculum called “Big Ideas of AI” for K-12 students which integrates an online mentor-training course with Scratch 3.0, Hal Abelson’s App Inventor and Jupyter Notebooks. We’ve received Microsoft Azure credits to support the project.