The Andrej Karpathy episode. During this interview, Andrej explains why reinforcement learning is terrible (but everything else is much worse), why AGI will just blend into the previous ~2.5 centuries of 2% GDP growth, why self driving took so long to crack, and what he sees as the future of education. It was a pleasure chatting with him. Watch on YouTube ; read the transcript . Sponsors * Labelbox helps you get data that is more detailed, more accurate, and higher signal than you could get by default, no matter your domain or training paradigm. Reach out today at labelbox.com/dwarkesh * Mercury helps you run your business better. It’s the banking platform we use for the podcast — we love that we can see our accounts, cash flows, AR, and AP all in one place. Apply online in minutes at mercury.com * Google’s Veo 3.1 update is a notable improvement to an already great model. Veo 3.1’s generations are more coherent and the audio is even higher-quality. If you have a Google AI Pro or Ultra plan, you can try it in Gemini today by visiting https://gemini.google Timestamps (00:00:00) – AGI is still a decade away (00:29:45) – LLM cognitive deficits (00:40:05) – RL is terrible (00:49:38) – How do humans learn? (01:06:25) – AGI will blend into 2% GDP growth (01:17:36) – ASI (01:32:50) – Evolution of intelligence & culture (01:42:55) - Why self driving took so long (01:56:20) - Future of education Get full access to Dwarkesh Podcast at www.dwarkesh.com/subscribe

Mention timeline

Jump between the book moments.

View every mention

Scale

The host mentions Scale while discussing the limitations of current LLMs as judges in reinforcement learning setups a…

Card

Book mentions1

Media mentions2

Quick answers

Quick FAQ

Answers to common summary, books, and takeaway questions for this episode.

What is Andrej Karpathy — AGI is still a decade away about?

Summary, books mentioned, transcript quotes, and timestamps for Andrej Karpathy — AGI is still a decade away on Dwarkesh Podcast.

Listen to the episode

What are the main takeaways from Andrej Karpathy — AGI is still a decade away?

These are the strongest takeaways surfaced by the transcript, summary copy, and linked mentions for Andrej Karpathy — AGI is still a decade away.

The conversation centers on LLM adversarial examples.
Referenced books include Scale.
The strongest audience signal points to AI researchers and engineers working on LLM training, robustness, and RL-from-feedback.
The Andrej Karpathy episode. During this interview, Andrej explains why reinforcement learning is terrible (but everything else is much worse), why AGI will just blend i…

Listen with timestamps

Which books are mentioned in Andrej Karpathy — AGI is still a decade away?

Scale are the clearest linked books in this episode, each tied back to transcript timestamps and quote cards.

Jump to books mentioned

Why are listeners searching for Andrej Karpathy — AGI is still a decade away?

Andrej Karpathy — AGI is still a decade away keeps attracting summary-style searches because this page combines episode context, transcript quotes, book references, and direct jump links back into the audio.

See topic signals

Topic and sentiment signals

Aggregated from transcript-derived mention metadata for better topical navigation and citation.

Top triggers

LLM adversarial examples(1)

Mention sentiment

Deep Dive(1)

Audience signals

AI researchers and engineers working on LLM training, robustness, and RL-from-feedback

Books Mentioned

The full list below is ranked by how useful each mention is to a listener: stronger recommendation language, clearer quote context, and better timestamp support rise first.

Scale

Best for AI researchers and engineers working on LLM training, robustness, and RL-from-feedbackOften cited around LLM adversarial examples

“Mentioned as a "really good book" by a physicist talking about biology; recommended for reading and about scaling laws of animals.”

View mention details

Sentiment: Deep Dive

Trigger: LLM adversarial examples

For: AI researchers and engineers working on LLM training, robustness, and RL-from-feedback

Key quote: “When you're reading a book, I almost don't even feel like the book is like exposition I'm supposed to be attending to and training on.”

The host mentions Scale while discussing the limitations of current LLMs as judges in reinforcement learning setups and the prevalence of adversarial, out-of-distribution examples that get erroneously high rewards. They use the book as a jumping-off point to argue that LLMs need new training regimes—synthetic problems, reflection-like processes, or meta-learning—to better generalize and evaluate solutions.

Direct Amazon listing · affiliate link

Check price

Weekly source-backed picks

Get the strongest books from new Dwarkesh Podcast episodes.

A short weekly email with transcript-backed book recommendations, source quotes, and exact moments from recently indexed episodes.

Scale

Dwarkesh Podcast · 2:16:10

When you're reading a book, I almost don't even feel like the book is like exposition I'm supposed to be attending to and training on.

One useful email a week. Unsubscribe anytime.

Shop This Episode

Pick up the books after you hear them in context.

Buy links stay tied to the same book cards and timestamps shown above.

Mentioned at 2:16:10

Scale

The host mentions Scale while discussing the limitations of current LLMs as judges in reinforcement learning setups and the prevalence of adversari…

Check price Jump to mention

Direct Amazon listing · affiliate link

Movies & Documentaries Mentioned

Movie

Idiocracy

Confidence: 90%

“Mentioned alongside WALL-E as a film depicting humanity being on the side of technological progress.”

Check price

Movie

WALL-E

Confidence: 90%

“Mentioned as an example of a movie depicting a future where humanity is sidelined by technology.”

Check price