Project focus: How to make it easier for people to understand, design, and critique how algorithms are designed?
Harvard John A. Paulson School of Engineering & Applied Sciences
Role: Course founder, professor
HUMAN-CENTERED ALGORITHMS DESIGN was a one-time course at SEAS that evolved out of the ai-kitchen group. Due to growing demand in 2016, the ai-kitchen discussion group expanded, becoming a course on Human Centered Algorithm Design designed and taught by Altringer at SEAS in Fall 2017. The team later worked with other groups at Harvard, MIT, and beyond that are better-positioned to take the conversation much further, namely, the BKC-MIT AI Initiative.
AI-kitchen ran from 2015-2017 and focused on how to make it easy to discuss and critique how algorithms work. We worked with collaborators to translate algorithmic ‘writing’ into plain language to help people understanding how AI works and how it could be improved. We believe that the basic inputs, assumptions, steps, and outputs of algorithms should be as accessible as sharing and discussing recipes. To do this, our team reverse-engineered popular algorithms like Tinder’s matching algorithm and AirBnB’s location algorithm, and hosted regular discussion events at Harvard where we presented on how they work in plain language. Some of our core themes included: Human Values and How They Relate to AI; Algorithmic Writing and Critique; The Design of AI; Human-Robot Interaction; Human-Computer Interaction, and How to make AI more accessible, representative, and desirable. Ai-kitchen began as a mix of professionals and students from academia and industry gathering *roughly* monthly in Harvard Square.*NEW* We recommend the syllabus that has developed out of the initiative’s work here.
Due to time constraints, we no longer update the ai-kitchen website regularly, but may occasionally post updates.