March 1, 2026
Bots in class, humans in exams
10-202: Introduction to Modern AI (CMU)
Free AI class drops; commenters clash over chatbots, calculus, and a Lisp comeback
TLDR: CMU dropped a free online intro-to-AI course that teaches you to build a simple chatbot. The comments erupted over missing “symbolic reasoning,” the rule allowing AI on homework but banning it in exams, and jokes about a retro Lisp comeback—highlighting a culture clash over how we should learn AI.
Carnegie Mellon just launched a beginner-friendly AI course—with a free online version trailing two weeks—and the comment section instantly turned into a tech soap opera. The class promises you’ll build a basic chatbot and learn how tools like ChatGPT work, all in a few hundred lines of code. Cue the crowd: one early post even got flagged, setting the tone for spicy moderation as the syllabus rolled out alignment, safety, and those dreaded homework quizzes. The policy allows using AI helpers for homework (but not during exams), which triggered cries of “homework with bots, tests with brains” and a debate over whether that’s smart training or mixed messaging.
The hottest take? An impatient demand for old-school logic: “Nothing on symbolic reasoning?” Fans of retro tech piled on with jokes about a “Lisp and Prolog revival,” turning the thread into a nostalgia fest. Meanwhile, newbies begged for guidance—“Is this a good course?”—while veterans snarked that if calculus and Python scare you, maybe don’t build a robot therapist. The free online track offers videos and autograded assignments, but no midterms or final—CMU keeps the big tests on campus. In short, this course landed like a pop album: flashy, accessible, and instantly polarizing, with commenters splitting into Team Pragmatic Chatbot vs. Team Bring Back Logic, all while memeing their way through the syllabus.
Key Points
- •CMU’s 10-202: Introduction to Modern AI is taught by Zico Kolter with lectures MW[F] 9:30–10:50 in Tepper 1403.
- •A minimal free online version begins on 1/26 with a two-week delay, offering lecture videos and autograded assignments via mugrade but excluding quizzes and exams.
- •The course teaches implementation of a basic AI chatbot and an open-source LLM, covering supervised ML, transformers, tokenizers, post-training, and safety.
- •Grading is 20% homework/programming, 40% homework quizzes, and 40% exams (two midterms at 10% each, final at 20%).
- •Prerequisites include Python proficiency (15-112 or 15-122) and calculus (21-111 or 21-120); homeworks are released as Colab/Marimo notebooks and submitted via mugrade.