Taking what they learned conceptually about artificial intelligence and machine learning (ML) this 12 months, students from across the Greater Boston area had the chance to use their recent skills to real-world industry projects as a part of an experiential learning opportunity offered through Break Through Tech AI at MIT.

Hosted by the MIT Schwarzman College of Computing, Break Through Tech AI is a pilot program that goals to bridge the talent gap for girls and underrepresented genders in computing fields by providing skills-based training, industry-relevant portfolios, and mentoring to undergraduate students in regional metropolitan areas to be able to position them more competitively for careers in data science, machine learning, and artificial intelligence.

“Programs like Break Through Tech AI gives us opportunities to attach with other students and other institutions, and allows us to bring MIT’s values of diversity, equity, and inclusion to the training and application within the spaces that we hold,” says Alana Anderson, assistant dean of diversity, equity, and inclusion for the MIT Schwarzman College of Computing.

The inaugural cohort of 33 undergraduates from 18 Greater Boston-area schools, including Salem State University, Smith College, and Brandeis University, began the free, 18-month program last summer with an eight-week, online skills-based course to learn the fundamentals of AI and machine learning. Students then split into small groups in the autumn to collaborate on six machine learning challenge projects presented to them by MathWorks, MIT-IBM Watson AI Lab, and Replicate. The scholars dedicated five hours or more each week to fulfill with their teams, teaching assistants, and project advisors, including convening once a month at MIT, while juggling their regular academic course load with other every day activities and responsibilities.

The challenges gave the undergraduates the possibility to assist contribute to actual projects that industry organizations are working on and to place their machine learning skills to the test. Members from each organization also served as project advisors, providing encouragement and guidance to the teams throughout.

“Students are gaining industry experience by working closely with their project advisors,” says Aude Oliva, director of strategic industry engagement on the MIT Schwarzman College of Computing and the MIT director of the MIT-IBM Watson AI Lab. “These projects might be an add-on to their machine learning portfolio that they will share as a piece example once they’re able to apply for a job in AI.”

Over the course of 15 weeks, teams delved into large-scale, real-world datasets to coach, test, and evaluate machine learning models in quite a lot of contexts.

In December, the scholars celebrated the fruits of their labor at a showcase event held at MIT through which the six teams gave final presentations on their AI projects. The projects not only allowed the scholars to accumulate their AI and machine learning experience, it helped to “improve their knowledge base and skills in presenting their work to each technical and nontechnical audiences,” Oliva says.

For a project on traffic data evaluation, students got trained on MATLAB, a programming and numeric computing platform developed by MathWorks, to create a model that allows decision-making in autonomous driving by predicting future vehicle trajectories. “It’s essential to comprehend that AI shouldn’t be that intelligent. It’s only as smart as you make it and that’s exactly what we tried to do,” said Brandeis University student Srishti Nautiyal as she introduced her team’s project to the audience. With firms already making autonomous vehicles from planes to trucks a reality, Nautiyal, a physics and arithmetic major, shared that her team was also highly motivated to contemplate the moral problems with the technology of their model for the protection of passengers, drivers, and pedestrians.

Using census data to coach a model could be tricky because they are sometimes messy and stuffed with holes. In a project on algorithmic fairness for the MIT-IBM Watson AI Lab, the toughest task for the team was having to scrub up mountains of unorganized data in a way where they might still gain insights from them. The project — which aimed to create demonstration of fairness applied on an actual dataset to guage and compare effectiveness of various fairness interventions and fair metric learning techniques — could eventually function an academic resource for data scientists serious about learning about fairness in AI and using it of their work, in addition to to advertise the practice of evaluating the moral implications of machine learning models in industry.

Other challenge projects included an ML-assisted whiteboard for nontechnical people to interact with ready-made machine learning models, and an indication language recognition model to assist disabled people communicate with others. A team that worked on a visible language app set out to incorporate over 50 languages of their model to extend access for the thousands and thousands of folks that are visually impaired throughout the world. In keeping with the team, similar apps available on the market currently only offer as much as 23 languages. 

Throughout the semester, students persevered and demonstrated grit to be able to cross the finish line on their projects. With the ultimate presentations marking the conclusion of the autumn semester, students will return to MIT within the spring to proceed their Break Through Tech AI journey to tackle one other round of AI projects. This time, the scholars will work with Google on recent machine learning challenges that can enable them to hone their AI skills even further with a watch toward launching a successful profession in AI.