New Course 6 major proposed in AI+Decision Making

Major will be implemented in Fall 2022 if approved

MIT Professor of Computer Science and Engineering Leslie Kaelbling presented a proposal for a new undergraduate major in Course 6 (Electrical Engineering and Computer Science) to the Undergraduate Student Advisory Group in EECS (USAGE) on Nov. 17. If approved, the major will be known as 6-4, or Artificial Intelligence and Decision Making (AI+D). 

The AI+D major would include machine learning, symbolic reasoning, computer vision, natural language, robotics, and medical AI. It aims to “integrate disciplines typically taught in different departments” including electrical engineering, computer science, statistics, operations research, and brain and cognitive sciences. 

The AI+D curriculum committee is working on presenting to the necessary MIT committees and piloting CI-M subjects and the new AI center in Fall 2021. If approved in Spring 2022, first-year students would be able to declare the new major. The major would be officially implemented in Fall 2022.

The proposal comes as AI+D is becoming an area of demand by students, where enrollment for related classes is growing. The number of students who declare EECS-related majors at MIT is also growing rapidly, from 23% of the undergraduate population at MIT in 2012 to 42% in 2021. 

The Course 6 majors currently consist of 6-1 (Electrical Science and Engineering), 6-3 (Computer Science and Engineering), and their overlap, 6-2 (Electrical Engineering and Computer Science). Blended majors include 6-7 (Computer Science and Molecular Biology), 6-9 (Computation and Cognition), 6-14 (Computer Science, Economics, and Data Science), and 11-6 (Urban Science and Planning with Computer Science). 

Kaelbling pointed out that the original computer science majors “mostly focus on the inside of the computer,” such as “making a compiler or coming up with an algorithm.” The intention is to “try to draw a box around it and analyze it internally.” The AI+D major, on the other hand, would focus on the “analysis and synthesis of systems that interact with an external world” via “perception, communication, and action.” 

Kaelbling said that the committee tried to focus on “exposure to different attitudes” and “ways of thinking about problems” in developing the curriculum. For example, when approaching a problem with a model, they asked “what the system is actually going to do in the world, what choices it should make, and how systems connect to humans around it.” 

Though 6-4 may seem similar to 6-9, Kaelbling noted that their fundamental difference is that 6-4 is related to AI, whereas 6-9 is related to humans. A student in 6-9 can do the “brain and cognitive sciences part without doing anything computational.”

Pranav Krishna ’23 wrote in an email to The Tech, “It's great to see that MIT is moving with the times with emphasizing AI and Machine Learning, though it's a shame that all the cool Machine Learning AUS subjects are becoming CI-Ms” in response to the proposed major. 

6-4 Degree Proposal

The current proposal for 6-4 contains 14.5 subjects, divided into categories of Fundamentals, Centers, Communication Intensive in the Major (CI-M), Application CI-M, Advanced Undergraduate Subjects (AUS), and Flex. 

Fundamentals contain 5.5 required subjects in math, probability, algorithms, programming, and discrete and continuous math: 6.0001 (Introduction to CS and Programming in Python), 6.042 (Mathematics for Computer Science), 6.006 (Introduction to Algorithms), either 6.008 (Introduction to Inference) or 6.041 (Introduction to Probability) or 18.05 (Introduction to Probability and Statistics), 6.009 (Fundamentals of Programming), and either 18.061 (Linear Algebra and Optimization) or 18.06 (Linear Algebra). 

Centers provide breadth in AI+D at an intermediate level. It contains five areas: data-centric, model-centric, decision-centric, computation-centric, and human-centric. 

Data-centric focuses on “building models and drawing conclusions from data” (statistics and machine learning). Decision-centric focuses on “choosing actions that affect an external world” (control theory, reinforcement learning, decision theory, game theory). Model-centric focuses on “specification and representation of models and priors” (classical AI, graphics). Computation-centric focuses on algorithms and implementation strategies (optimization algorithms, vector/tensor programming). Human-centric focuses on the interplay between humans and computational systems (societal impact, bias and fairness, user interfaces, cognitive science). 

In Centers, students would be required to take 5 total subjects with one from each area. Data-centric includes 6.036 (Introduction to Machine Learning) and 6.401 (Introduction to Statistical Data Analysis). Model-centric has 6.837 (Computer Graphics), 6.003 (Signal Processing), and 6.038 (AI Representation and Reason). Decision-centric includes 6.038, 6.302 (Feedback System Design), 6.207 (Networks), and 6.215 (Optimization Methods). Computation-centric has 6.046 (Design and Analysis of Algorithms), 6.215, and 6.837. Human-centric includes 6.s080 (Software Systems for Data Science), 6.804 (Computational Cognitive Science), 6.805 (Foundations of Information Policy), and 6.207.

The Application CI-M category would be satisfied by a non-concrete list of either 6.800 (Robotic Manipulation), 6.806 (Quantitative Methods for Natural Language Processing), 6.819 (Advances in Computer Vision), 6.835 (Intelligent Multimodal User Interfaces), or 6.141 (Robotics: Science and Systems I). 6.806 and 6.800 are currently ongoing pilots in Fall 2021, 6.819 will be offered in Spring 2023, and 6.835 is aiming to be offered in Spring 2023. 

Tentatively, 6.800, 6.806, and 6.819 are to be offered as 15-unit undergraduate subjects that meet with their respective 12-unit graduate subjects. The additional 3 units would come from added Communication Intensive material. 

The regular Course 6 CI-M requirement would be completed by either 6.UAT (Oral Communication) or 6.UAR (Seminar in Undergraduate Advanced Research). 

To satisfy their AUS, students can select any one course from the AI+D AUS list or the Application CI-M list.

Flex subjects would include any additional subject satisfying a degree requirement in Course 6 or Course 18.