REAL

Holds up: the 'Stanford lecture that explains how ChatGPT and Claude work' is Yann Dubois's free CS229 guest lecture on YouTube, though he was a PhD student, not a professor, and 'comment STANFORD' is a hook for a public link

Facebook · Oct 2026

The claim'Comment STANFORD below and I'll send you the link.' 'This is the best explanation of how LLMs (like ChatGPT and Claude) actually work. A Stanford professor broke it down so clearly that even if you've never touched AI, you'll get it.' On screen: 'Skip Netflix today and watch this Stanford lecture instead, it's less than 2hrs and explains how ChatGPT and Claude work better than anything else I've found.' Follow for daily prompts and resources.

The lecture is real, free and good. The laptop in the reel is playing 'Stanford CS229 I Machine Learning I Building Large Language Models (LLMs)' on the Stanford Online YouTube channel, video ID 9vM4p9NN0Ts, 1 hour 44 minutes, about 2.9 million views. The slide on screen reads 'Introduction to Building LLMs, CS229: Machine Learning, Yann Dubois, Aug 13th 2024'. It covers tokenization, pretraining, scaling laws, post-training with SFT and RLHF, and evaluation, in plain language, and it is widely recommended for exactly the reason the caption gives. Two corrections. The speaker was not a Stanford professor; Stanford's own description says the guest lecture was delivered by Yann Dubois, then a fourth-year CS PhD student. He now leads a post-training team at OpenAI, which if anything makes the talk more relevant, but 'professor' is wrong. And 'comment STANFORD and I'll send you the link' is a gate on a link that is one YouTube search away; the page is harvesting comments and DMs to grow a prompts-and-resources follow, which is the actual product. The 'under 2 hours' line is accurate. The 'best explanation I've found' line is opinion, but it is a defensible one. The claim stands; the funnel around it is unnecessary.

What holds up

Reel read frame by frame: a laptop on a bed plays a lecture video with a 'Stanford' watermark; the title slide reads 'Introduction to Building LLMs, CS229: Machine Learning, Yann Dubois, Aug 13th 2024', followed by an 'LLMs' slide with a ChatGPT example.
YouTube, Stanford Online: 'Stanford CS229 I Machine Learning I Building Large Language Models (LLMs)', video ID 9vM4p9NN0Ts, 1:44:31, about 2.9M views; description: 'This guest lecture was delivered by Yann Dubois in Stanford's CS229: Machine Learning course, in Summer 2024', a fourth-year CS PhD student at the time.
yanndubs.github.io: 'I'm a researcher at OpenAI, where I lead the Post-training Frontiers team. I was a PhD student at Stanford University.' He was not a professor.
The link is public and free; the 'comment STANFORD' hook exists to generate comments and DMs for a page that sells follows for 'daily prompts and resources'. No course or product is attached to this reel.

What doesn’t

'A Stanford professor' is wrong; the speaker was a PhD student giving a guest lecture (now at OpenAI).
'Comment STANFORD for the link' gates a free public YouTube video; the comments are the product.
The lecture is from August 2024 and predates GPT-5, GPT-6 Astra and Claude 5; the fundamentals still hold, the model names do not.

The catch

The lecture is worth the two hours and costs nothing. Search 'Stanford CS229 Building Large Language Models' on YouTube and skip the comment. Just know it is a 2024 talk, so take the model names as examples, not the current lineup.

How to actually do it

Open YouTube and search 'Stanford CS229 Building Large Language Models Yann Dubois' (Stanford Online channel, 1 hr 44 min); watch at 1.25x if you only want the pretraining and post-training sections.
Pair it with one hands-on hour: open ChatGPT or Claude and test what you learned about tokens, context and post-training on your own documents.
If you want the current picture, follow the lecture with the vendors' own model pages (OpenAI, Anthropic) rather than a prompts page's DM link.

The video, the length and the Stanford origin all check out, and it is a genuinely clear explainer. The speaker was a PhD student rather than a professor, and the comment-for-link hook is a follower-growth tactic wrapped around a free link.

Confidence: High
Posted by: an 'AI for beginners' prompts-and-resources page (598K views on the reel); reel sent to Buddy Sat Sep 26, 2026, ~5:25 PM

See the original claim →

We test hype for free. We build the real thing for a living.

Thirty minutes, no pitch — and you'll leave with something useful either way.

Book a call with Todd or start with the free Business Checkup →

The Verdict Weekly

Three verdicts every Friday. Free forever, unsubscribe anytime, no spam — that would be ironic.