Best explanation
"I update my… What are the missing variables in my internal model"
Adam Marblestone – AI is missing something fundamental about the brain
Public index moment — strongest composite transcript signals for this topic (heuristic).
Topic research hub
Dwarkesh Patel
Dwarkesh Patel: the numbers. I feel about 10% to 20%, if I had to guess, is only knowledge work, someone could work from home and perform tasks, something like that. It's still a really large market. What is the size of the economy, and…
High-signal research hub
Best transcript moments
Canonical moments ranked from the public index — preferring multi-word, semantic excerpts where available.
the size of the economy, and what is 10% or 20%
Andrej Karpathy — “We’re summoning ghosts, not building animals”
1:09:09
I update my… What are the missing variables in my internal model
Adam Marblestone – AI is missing something fundamental about the brain
24:02
your advice as a student to other students, if you don't have a Karpathy who is doing the exposition of an idea
Andrej Karpathy — “We’re summoning ghosts, not building animals”
2:23:51
provable, and what is difficult
Terence Tao – How the world’s top mathematician uses AI
39:10
the types of bottlenecks that remain, and what are the prospects for getting past them
Michael Nielsen – Why aliens will have a different tech stack than us
48:24
a kind of specific qualitative change where you'd be training Llama-5 or Llama-4
Mark Zuckerberg — Llama 3, $10B models, Caesar Augustus, & 1 GW datacenters
38:43
the model quality impact
How GPT, Claude, and Gemini are actually trained and served – Reiner Pope
28:53
time, as a limited resource, what is the biggest difference
Terence Tao – How the world’s top mathematician uses AI
1:13:07
that large powerful models trained
Fully autonomous robots are much closer than you think – Sergey Levine
1:04:41
hard and what is easy, we're struggling quite a lot
Terence Tao – How the world’s top mathematician uses AI
21:19
building models and abstractions and understanding that there's a first-order
Andrej Karpathy — “We’re summoning ghosts, not building animals”
2:16:23
you're predicting profit is that you are systematically underinvesting in compute
Dario Amodei — “We are near the end of the exponential”
1:01:28
Videos discussing this topic
Terence Tao – How the world’s top mathematician uses AI
9 indexed moments · Dwarkesh Patel
Adam Marblestone – AI is missing something fundamental about the brain
8 indexed moments · Dwarkesh Patel
Fully autonomous robots are much closer than you think – Sergey Levine
6 indexed moments · Dwarkesh Patel
Andrej Karpathy — “We’re summoning ghosts, not building animals”
5 indexed moments · Dwarkesh Patel
Mark Zuckerberg — Llama 3, $10B models, Caesar Augustus, & 1 GW datacenters
4 indexed moments · Dwarkesh Patel
Michael Nielsen – Why aliens will have a different tech stack than us
2 indexed moments · Dwarkesh Patel
How GPT, Claude, and Gemini are actually trained and served – Reiner Pope
2 indexed moments · Dwarkesh Patel
Dario Amodei — “We are near the end of the exponential”
2 indexed moments · Dwarkesh Patel
Dylan Patel — The single biggest bottleneck to scaling AI compute
3 indexed moments · Dwarkesh Patel
China is killing the US on energy. Does that mean they’ll win AGI? — Casey Handmer
1 indexed moment · Dwarkesh Patel
AI Scaling, Alignment, and the Path to Superintelligence — With Dwarkesh Patel
1 indexed moment · Dwarkesh Patel
Ilya Sutskever – We're moving from the age of scaling to the age of research
1 indexed moment · Dwarkesh Patel
Related topics
Related searches
the size of the economy, and what is 10% or 20%I update my… What are the missing variables in my internal modelyour advice as a student to other students, if you don't have a Karpathy who is doing the exposition of an ideaprovable, and what is difficultthe types of bottlenecks that remain, and what are the prospects for getting past thema kind of specific qualitative change where you'd be training Llama-5 or Llama-4the model quality impacttime, as a limited resource, what is the biggest differencethat large powerful models trainedhard and what is easy, we're struggling quite a lot
Creators discussing this topic
No mapped creator profiles for these moments yet.
Continue researching
Jump into search, other hubs, and the public moment index to go deeper in-session.
Research lens
Best answer across indexed moments
Grouped by transcript heuristics only — not generative summaries and not fact-checking. Empty slots mean we did not find a confident match for that role in this hub.
Beginner explanation
"the size of the economy, and what is 10% or 20%"
Andrej Karpathy — “We’re summoning ghosts, not building animals”
Public index moment — beginner / definitional wording in the excerpt.
Technical explanation
"the model quality impact"
How GPT, Claude, and Gemini are actually trained and served – Reiner Pope
Public index moment — technical vocabulary or systems detail in the excerpt.
Counterpoint / caveat
"that large powerful models trained"
Fully autonomous robots are much closer than you think – Sergey Levine
Public index moment — hedging, disagreement, or risk language detected (possible caveat).
Compare explanations
Different experts and framings on the same topic — compare before you decide.
"I update my… What are the missing variables in my internal model"
Adam Marblestone – AI is missing something fundamental about the brain
Tutorial / walkthrough style
The Bayesian inference problem, which is basically the problem of perception, given some model of the world and given some data, how should I update my… What are the missing variables in my internal model?
"the size of the economy, and what is 10% or 20%"
Andrej Karpathy — “We’re summoning ghosts, not building animals”
Tutorial / walkthrough style
I feel about 10% to 20%, if I had to guess, is only knowledge work, someone could work from home and perform tasks, something like that.
"time, as a limited resource, what is the biggest difference"
Terence Tao – How the world’s top mathematician uses AI
Tutorial / walkthrough style
If civilization could from first principles decide how to use Terry Tao's time, as a limited resource, what is the biggest difference?
"that large powerful models trained"
Fully autonomous robots are much closer than you think – Sergey Levine
Possible caveat or counterpoint
Meta learning is emergent, as you pointed out before. LLMs essentially do a kind of meta learning via in-context learning.
"the model quality impact"
How GPT, Claude, and Gemini are actually trained and served – Reiner Pope
Possible caveat or counterpoint
This paper is "Unified Scaling Laws for Routed Language Models." It's a somewhat old paper by this stage, but one of the things they looked at is if I keep increasing sparsity, what is the model quality impact?
"a kind of specific qualitative change where you'd be training Llama-5 or Llama-4"
Mark Zuckerberg — Llama 3, $10B models, Caesar Augustus, & 1 GW datacenters
Possible caveat or counterpoint
Help grow the technical explanation index
Get notified as we index more engineering talks and tutorials for in-video transcript search.