Hi! I'm Phillip Compeau, I teach in Carnegie Mellon's Griffin School of Computer Science, and I run CS Learn, which launched today.
We're releasing full semesters of lectures from current CMU computer science courses across a variety of topics, with one course premiering each day this week. More courses come in January.
Yes, MIT got here 25 years ago :). What I'm excited about is how we're doing it: a few of our students built tools that automate a lot of the production work, so a small team can get a course out quickly and still have people review every lecture.
We'll have lots more coming across a wide array of subjects. I'd love to know if there are topics you would like to see us cover!
CMU has some of the best courses on advanced topics, but the ones you have put out so far are the introductory courses. Would love to see courses on advanced topics such as database internals and distributed systems. Also please release materials such as slide decks and PDF course notes.
I plan to put my distributed systems lectures online the next time I teach it, but not sure when that will be after I rotate out of teaching 15-213. One of my colleagues who also teaches it might do it before then.
Do you/could you release any information about the automation of production work? I'd be interested in what those tools look like and how they're improving your workflow.
It's still very much a work in progress as we scale up for next semester. When things get settled I would love to publish this because it's been a very interesting problem to solve how to present all the different things that can arise in a university classroom. We benefit greatly from having an existing team of course videographers who were previously keeping recordings internal.
Hi! Love to see these released, thanks to all at CMU who help make it happen. Would y'all be willing to host this on a CMU CS Peertube instance as well versus only on Youtube?
You could push the videos to CloudFlare R2 and not pay egress fees, and provide an RSS feed for updates. An XML based index of the videos with transcriptions and video markers would be great.
Is your workflow open source? I have had great results building bespoke image editing pipelines using claude and ffmpeg.
It would be awesome to be able to ask an LLM questions and have it respond with lecture snippets.
In this day and age, why learn computer science? I learned to program before there was Google but with AI, someone who doesn't know the difference between an int and a boolean can make a digit thing. Where do you see the importance of big O notation and, say, eating?
In the age of LLMs and Multi-modal models, do you think students across the world will benefit from this? If so, how do you think they will benefit.
For some context, when others released courses in youtube/coursera/udacity decades ago, many students, across the world, did not have any real access good teachers and qualified lecturers who could explain the concepts well. Those releases were boon to them.
Now, we learn by asking AI against anything, include youtube content and AI has been best teacher so far.
> I tried testing each frontier llm on questions w.r.t deferred tax and it made up stuff
This would be a stronger example if it were related to Comp Sci. It’s entirely possible that an LLM would perform better on CS fundamentals than details about taxes.
You're going to get flagged for asking - it does seem to be in bad taste IMO - but I do think it is the elephant in the room.
"AI" will teach you anything at a undergrad level - customized to your style.
The real reason you'd go to CMU or MIT is for credentials, or, more honestly, for networking opportunities. That doesn't exist for watching YouTube videos any more than prompting an AI.
Hard disagree on the reason to attend a school like CMU or MIT (disclaimer, I’m a CMU MatSci PhD)
Networking is a big value proposition, but the biggest benefit IMO is learning what questions are important and learning how to think independently about them. LLMs are great for learning but I don’t think they fully replace the skills you need to ask important questions. (This perspective is probably slanted towards research degrees but I think it’s at least somewhat transferable to engineering)
"AI" cannot teach a person what they don't know they don't know.
Human curation will always have a place in passing on knowledge for people who desire deep understanding of a topic, particularly young people and novices in a field.
You can ask ChatGPT for a detailed listing of all the areas within a discipline and it will give a totally decent answer, better than you'll likely find anywhere else. You can ask it to list disciplines that might be useful, that are taught at university, or whatever.
You can even just ask it "help me know what areas/subjects/topics are useful that I wouldn't even know exist"
On a long enough timeline, we are all dead. In the meantime, come check out Graham Neubig and Daniel Fried teach AI agents, and you will see that we have value still.
You have an amazing dataset, handed out for free by a top rated institution like CMU that can be used as the baseline for whatever learning flow you want to do with your AI.
You do understand that data like this is precisely the goldmine that makes AI good at answering general knowledge questions?
I haven’t watched any of these yet but seeing this takes me back to 2008/2009 when I was just learning to Java in high school and my teacher showed us Randy Pausch’s last lecture and the Alice project. That resonated with me for so much of my life. Also Alice was a really cool way to learn programming and probably still great to teach it today.
On the contrary to speculations in the comments, i think this is great and shows integrity. TheY are offering their knowledge to the world. In the meantime we can still appreciate real human knowledge being shared. AI has it's place.
I remember back in the late 2000s/early 2010s many many universities started putting much of their courses on youtube for free. But then they were made aware they legally liable in some jurisdictions for not having accessibility (youtube didn't auto-caption back then) and that liability pressured most universities into taking their courses down.
Glad to see things finally starting to get back to normal.
We're releasing full semesters of lectures from current CMU computer science courses across a variety of topics, with one course premiering each day this week. More courses come in January.
Yes, MIT got here 25 years ago :). What I'm excited about is how we're doing it: a few of our students built tools that automate a lot of the production work, so a small team can get a course out quickly and still have people review every lecture.
We'll have lots more coming across a wide array of subjects. I'd love to know if there are topics you would like to see us cover!
I plan to put my distributed systems lectures online the next time I teach it, but not sure when that will be after I rotate out of teaching 15-213. One of my colleagues who also teaches it might do it before then.
In particular, I would love to see you cover CAD, esp. if you could work up a way to use FreeCAD as the subject.
Is your workflow open source? I have had great results building bespoke image editing pipelines using claude and ffmpeg.
It would be awesome to be able to ask an LLM questions and have it respond with lecture snippets.
In the age of LLMs and Multi-modal models, do you think students across the world will benefit from this? If so, how do you think they will benefit.
For some context, when others released courses in youtube/coursera/udacity decades ago, many students, across the world, did not have any real access good teachers and qualified lecturers who could explain the concepts well. Those releases were boon to them.
Now, we learn by asking AI against anything, include youtube content and AI has been best teacher so far.
Even if your premise is 100% correct, releasing this will still help, because LLMs will ingest it or use it as a source and provide better answers.
Just yday I tried testing each frontier llm on questions w.r.t deferred tax and it made up stuff.
Fundamentals should never be learned via llm’s.
This would be a stronger example if it were related to Comp Sci. It’s entirely possible that an LLM would perform better on CS fundamentals than details about taxes.
"AI" will teach you anything at a undergrad level - customized to your style.
The real reason you'd go to CMU or MIT is for credentials, or, more honestly, for networking opportunities. That doesn't exist for watching YouTube videos any more than prompting an AI.
Networking is a big value proposition, but the biggest benefit IMO is learning what questions are important and learning how to think independently about them. LLMs are great for learning but I don’t think they fully replace the skills you need to ask important questions. (This perspective is probably slanted towards research degrees but I think it’s at least somewhat transferable to engineering)
Human curation will always have a place in passing on knowledge for people who desire deep understanding of a topic, particularly young people and novices in a field.
You can even just ask it "help me know what areas/subjects/topics are useful that I wouldn't even know exist"
You do understand that data like this is precisely the goldmine that makes AI good at answering general knowledge questions?
I don't know who is we besides you, but was it the AI that told you you're being taught well?
Glad to see things finally starting to get back to normal.