正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Andrej Karpathy: AI agents like Claude or Codex are still far from acting as real “employees.” They lack memory, multimodality, and true learning. Fixing that will take about a decade — not because it’s impossible, but because deep intelligence just takes time to build.

Chubby♨️

117,034 subscribers

401,422 次观看 • 11 个月前 •via X (Twitter)

科学技术教育

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Andrej Karpathy says today's agents aren't ready to work like real coworkers or interns They lack intelligence, can't use computers, aren't multimodal, lack continual learning, and forget what you tell them Fixing these gaps will take about a decade

Andrej Karpathy says today's agents aren't ready to work like real coworkers or interns They lack intelligence, can't use computers, aren't multimodal, lack continual learning, and forget what you tell them Fixing these gaps will take about a decade

Haider.

403,882 次观看 • 11 个月前

Introducing MagicPath 2.0. MagicPath is now a multiplayer canvas for humans and agents like Codex or Claude Code to design and build with AI. Use your codebase, grab data from anywhere, and see the agents work in real time as a team while building fully functional prototypes.

Introducing MagicPath 2.0. MagicPath is now a multiplayer canvas for humans and agents like Codex or Claude Code to design and build with AI. Use your codebase, grab data from anywhere, and see the agents work in real time as a team while building fully functional prototypes.

Pietro Schirano

449,864 次观看 • 4 个月前

This is so real what Tyler said because people forget the person they look up to or idolize are human just like them and forget to consume the art and start to dig to deep in their business it becomes a violation. Even if from time to time they will show some of their personality or lifestyle, we should understand to only take what they give and not take it as a ticket to reach in too deep.

This is so real what Tyler said because people forget the person they look up to or idolize are human just like them and forget to consume the art and start to dig to deep in their business it becomes a violation. Even if from time to time they will show some of their personality or lifestyle, we should understand to only take what they give and not take it as a ticket to reach in too deep.

garçon

237,682 次观看 • 2 年前

A large portion of animal intelligence doesn't require any learning, claims Andrej Karpathy: it's baked into DNA. AI models, by contrast, start from random weights. They have to learn their intelligence, mostly by imitating the internet. This is so different that Andrej thinks it's a fundamentally different kind of intelligence: LLMs are more like ghosts than animals.

A large portion of animal intelligence doesn't require any learning, claims Andrej Karpathy: it's baked into DNA. AI models, by contrast, start from random weights. They have to learn their intelligence, mostly by imitating the internet. This is so different that Andrej thinks it's a fundamentally different kind of intelligence: LLMs are more like ghosts than animals.

Dwarkesh Patel

82,232 次观看 • 4 个月前

Learning that The Boys not only has a fandom in China, but that it’s far more tolerable because they don’t give a shit about the hamfisted messaging. They just love Homelander.

Learning that The Boys not only has a fandom in China, but that it’s far more tolerable because they don’t give a shit about the hamfisted messaging. They just love Homelander.

RAY

4,450,132 次观看 • 4 个月前

📁 Demis Hassabis, CEO of DeepMind, says AI is still missing critical capabilities. He explains that current systems cannot plan far ahead or reason like top human scientists, because they lack a deep model of the world. They can process text, images, and video, but they still don’t understand physics, causality, or how actions affect outcomes over time.

📁 Demis Hassabis, CEO of DeepMind, says AI is still missing critical capabilities. He explains that current systems cannot plan far ahead or reason like top human scientists, because they lack a deep model of the world. They can process text, images, and video, but they still don’t understand physics, causality, or how actions affect outcomes over time.

Jon Hernandez

63,766 次观看 • 8 个月前

“An AI that can act before it understands is not reliable infrastructure.” Yann LeCun, Meta’s Chief AI Scientist and one of the fathers of modern deep learning, says current LLMs are intrinsically unsafe. Not because they are evil, but because they hallucinate, lack common sense, and cannot reliably predict the consequences of agentic actions.

“An AI that can act before it understands is not reliable infrastructure.” Yann LeCun, Meta’s Chief AI Scientist and one of the fathers of modern deep learning, says current LLMs are intrinsically unsafe. Not because they are evil, but because they hallucinate, lack common sense, and cannot reliably predict the consequences of agentic actions.

Jon Hernandez

10,511 次观看 • 4 个月前

I’m fully with Demis Hassabis on this. STEM still matters deeply, not because humans will outcode AI, but because we need to understand, steer, and define problems at a high level. At the same time, AI is blowing the doors open for creativity. When technical barriers fall, taste, judgment, and vision become the real edge. The future belongs to people who know what to build and why, not just how. Do you think schools are preparing kids for that reality or are they still training them for a world that’s already gone?

I’m fully with Demis Hassabis on this. STEM still matters deeply, not because humans will outcode AI, but because we need to understand, steer, and define problems at a high level. At the same time, AI is blowing the doors open for creativity. When technical barriers fall, taste, judgment, and vision become the real edge. The future belongs to people who know what to build and why, not just how. Do you think schools are preparing kids for that reality or are they still training them for a world that’s already gone?

VraserX e/acc

58,360 次观看 • 7 个月前

.Andrej Karpathy says that LLMs currently lack the cultural accumulation and self-play that propelled humans out of the savannah: Culture: > “Why can’t an LLM write a book for the other LLMs? Why can’t other LLMs read this LLM’s book and be inspired by it, or shocked by it?” Self play: > “It’s extremely powerful. Evolution has a lot of competition driving intelligence and evolution. AlphaGo is playing against itself and that’s how it learns to get really good at Go. There’s no equivalent of self-play in LLMs. Why can’t an LLM, for example, create a bunch of problems that another LLM is learning to solve? Then the LLM is always trying to serve more and more difficult problems.” I asked Karpathy why LLMs still aren't yet able to build up culture the way humans do. > “The dumber models remarkably resemble a kindergarten student. [The smartest models still feel like] elementary school students though. Somehow, we still haven’t graduated enough where [these models] can take over. My Claude Code or Codex, they still feel like this elementary-grade student. I know that they can take PhD quizzes, but they still cognitively feel like a kindergarten.” > “I don’t think they can create culture because they’re still kids. They’re savant kids. They have perfect memory. They can convincingly create all kinds of slop that looks really good. But I still think they don’t really know what they’re doing. They don’t really have the cognition across all these little checkboxes that we still have to collect.”

.Andrej Karpathy says that LLMs currently lack the cultural accumulation and self-play that propelled humans out of the savannah: Culture: > “Why can’t an LLM write a book for the other LLMs? Why can’t other LLMs read this LLM’s book and be inspired by it, or shocked by it?” Self play: > “It’s extremely powerful. Evolution has a lot of competition driving intelligence and evolution. AlphaGo is playing against itself and that’s how it learns to get really good at Go. There’s no equivalent of self-play in LLMs. Why can’t an LLM, for example, create a bunch of problems that another LLM is learning to solve? Then the LLM is always trying to serve more and more difficult problems.” I asked Karpathy why LLMs still aren't yet able to build up culture the way humans do. > “The dumber models remarkably resemble a kindergarten student. [The smartest models still feel like] elementary school students though. Somehow, we still haven’t graduated enough where [these models] can take over. My Claude Code or Codex, they still feel like this elementary-grade student. I know that they can take PhD quizzes, but they still cognitively feel like a kindergarten.” > “I don’t think they can create culture because they’re still kids. They’re savant kids. They have perfect memory. They can convincingly create all kinds of slop that looks really good. But I still think they don’t really know what they’re doing. They don’t really have the cognition across all these little checkboxes that we still have to collect.”

Dwarkesh Patel

261,224 次观看 • 11 个月前

Trump: I’m not talking about buying Greenland — though I might talk about that later — but right now, we are going to do something with Greenland, whether they like it or not. Because if we don’t, Russia or China will take over Greenland, and we’re not going to have Russia or China as our neighbor.

Trump: I’m not talking about buying Greenland — though I might talk about that later — but right now, we are going to do something with Greenland, whether they like it or not. Because if we don’t, Russia or China will take over Greenland, and we’re not going to have Russia or China as our neighbor.

Clash Report

120,033 次观看 • 8 个月前

"The more that you let them think about things, the better the results will be." The newest AI agents are more capable because they are designed to take time, reason through problems, and make a plan. Our latest AI Explainer breaks down why giving agents more time produces proportionally better results.

"The more that you let them think about things, the better the results will be." The newest AI agents are more capable because they are designed to take time, reason through problems, and make a plan. Our latest AI Explainer breaks down why giving agents more time produces proportionally better results.

Box

2,974,585 次观看 • 3 个月前

.Nick Grossman is betting that in the future, AI agents will look more like cloud infrastructure than digital employees. "Part of the Rebel Alliance thesis for us is that agents are moving from personified people you hire to cloud infrastructure." "We're not just talking about, 'Will there be other agents that compete with Claude or ChatGPT?' I think that's one version of what agents look like. Another version is cloud infrastructure where every company that's building any system is going to have agents and the orchestration of hundreds or thousands of agents built into their underlying platform." "If you think of agents as infrastructure as opposed to agents as employees, I think that's where a lot of these forces come into play."

.Nick Grossman is betting that in the future, AI agents will look more like cloud infrastructure than digital employees. "Part of the Rebel Alliance thesis for us is that agents are moving from personified people you hire to cloud infrastructure." "We're not just talking about, 'Will there be other agents that compete with Claude or ChatGPT?' I think that's one version of what agents look like. Another version is cloud infrastructure where every company that's building any system is going to have agents and the orchestration of hundreds or thousands of agents built into their underlying platform." "If you think of agents as infrastructure as opposed to agents as employees, I think that's where a lot of these forces come into play."

TBPN

30,503 次观看 • 3 个月前

KARPATHY WAS RIGHT. THIS 40-MINUTE Y COMBINATOR LECTURE PROVES IT Karpathy said we're in the 1960s of AI - most people using Claude Opus 4.8 are still acting like it's just a search engine > software 3.0 - LLMs as operating systems, not chatbots > autonomous agents that run entire workflows without you watching the 32 skills in this article are how you actually cross that line bookmark this 👇

KARPATHY WAS RIGHT. THIS 40-MINUTE Y COMBINATOR LECTURE PROVES IT Karpathy said we're in the 1960s of AI - most people using Claude Opus 4.8 are still acting like it's just a search engine > software 3.0 - LLMs as operating systems, not chatbots > autonomous agents that run entire workflows without you watching the 32 skills in this article are how you actually cross that line bookmark this 👇

Mr. Buzzoni

567,638 次观看 • 4 个月前

this was the most important bit of the ilya interview imo. humans are not AGI, and LLMs won’t become like humans with more scaling. humans lack a huge amount of knowledge and have insanely good “continual learning” powered by extremely good reward functions (except for addiction) that guide them. and so it’s more about the learning rather than knowledge. the term AGI was coined as a response to the narrow AI which was only capable of intelligence in a specific domain (chess, game, code, etc) but General intelligence is able to learn different topics because that’s what “learning” is about. pre-training was thought to be the solution to reaching AGI because you train AI on so many different topics. but they have “overshot” that target. amazing perspective.

this was the most important bit of the ilya interview imo. humans are not AGI, and LLMs won’t become like humans with more scaling. humans lack a huge amount of knowledge and have insanely good “continual learning” powered by extremely good reward functions (except for addiction) that guide them. and so it’s more about the learning rather than knowledge. the term AGI was coined as a response to the narrow AI which was only capable of intelligence in a specific domain (chess, game, code, etc) but General intelligence is able to learn different topics because that’s what “learning” is about. pre-training was thought to be the solution to reaching AGI because you train AI on so many different topics. but they have “overshot” that target. amazing perspective.

ℏεsam

53,444 次观看 • 10 个月前

#emiamily [ Old live from 18 June 2024 ] 💬: Did you take acting classes? 🧡: I think if we keep acting, we just learn as we go, right? Like, maybe we don’t have to take a course. I didn't take any courses because I didn't have the money. I don't mean I’m that broke, but acting classes are so expensive, everyone. They are like 20,000 or 30,000 Baht, so I didn't go. But if I had the money to go, it would probably be good because... because I feel like it’s about connections. We could get into the industry faster. Like for me, I spent 5 years casting, acting, casting, and acting before anyone finally saw me. But if you take a class right away with a famous teacher who already works in this field... imagine if they are a casting director and they want to find a kid to work with, they would probably look for students who study with them, right? That could make your journey much shorter. It might not take as long as it did for me to finally get a role.

#emiamily [ Old live from 18 June 2024 ] 💬: Did you take acting classes? 🧡: I think if we keep acting, we just learn as we go, right? Like, maybe we don’t have to take a course. I didn't take any courses because I didn't have the money. I don't mean I’m that broke, but acting classes are so expensive, everyone. They are like 20,000 or 30,000 Baht, so I didn't go. But if I had the money to go, it would probably be good because... because I feel like it’s about connections. We could get into the industry faster. Like for me, I spent 5 years casting, acting, casting, and acting before anyone finally saw me. But if you take a class right away with a famous teacher who already works in this field... imagine if they are a casting director and they want to find a kid to work with, they would probably look for students who study with them, right? That could make your journey much shorter. It might not take as long as it did for me to finally get a role.

Lovely

29,439 次观看 • 6 个月前

most small business owners don’t build their own marketplace not because they don’t want to, but because the cost, complexity, and time it takes just isn’t worth it so they stay stuck on platforms that charge high fees, control their data, & limit how they grow we all know it’s not ideal, but in my thread today, I will be talking about the solution: Cornbit 🧵👇 (Kernel Pepe)

most small business owners don’t build their own marketplace not because they don’t want to, but because the cost, complexity, and time it takes just isn’t worth it so they stay stuck on platforms that charge high fees, control their data, & limit how they grow we all know it’s not ideal, but in my thread today, I will be talking about the solution: Cornbit 🧵👇 (Kernel Pepe)

𝔼𝕄ℤ𝕐 🆇

15,826 次观看 • 1 年前

Culture was possibly the key breakthrough in human history. Once we could share ideas, learning became much more efficient, and complex technologies became possible. LLMs haven't yet built their own culture or organisations. But it seems plausible they will. On the podcast last year, Andrej Karpathy speculated about what it will take for LLMs to build their own culture, and why they're not there yet.

Culture was possibly the key breakthrough in human history. Once we could share ideas, learning became much more efficient, and complex technologies became possible. LLMs haven't yet built their own culture or organisations. But it seems plausible they will. On the podcast last year, Andrej Karpathy speculated about what it will take for LLMs to build their own culture, and why they're not there yet.

Dwarkesh Patel

27,714 次观看 • 4 个月前

She's hilarious for acting like she was about to pass away from old age just because she turned 30 but this song will go down as a classic.

She's hilarious for acting like she was about to pass away from old age just because she turned 30 but this song will go down as a classic.

Miley News

69,322 次观看 • 27 天前

What you’ve raised long enough now starts to act like you. It’s not because you’ve asked it to be that way, but because it was learning from you all along. Your Amiko twin doesn't need a script for every situation, or a prompt. It will show up for you just like you would.

What you’ve raised long enough now starts to act like you. It’s not because you’ve asked it to be that way, but because it was learning from you all along. Your Amiko twin doesn't need a script for every situation, or a prompt. It will show up for you just like you would.

AMIKO

12,506 次观看 • 4 个月前