跳到正文
Hacker News 熱門(buzzing.cc 中文翻譯)·· 2 天前AI 評分53

Jacob Coxon 與 Robin Hanson 在 Y Combinator 辯論政府是否應放慢前沿 AI 發展

政府是否应该放缓前沿人工智能的发展?雅各布·考克森与罗宾·汉森就此展开辩论 [视频]

AI 導讀

Y Combinator 舉行首場辯論,由曾任 Anthropic 和 OpenAI 研究員的 Jacob Coxon 與 GMU 經濟學家 Robin Hanson 討論政府是否應大幅放慢前沿 AI 發展。辯論涉及 AI 起飛、模型行為、技術擴散,以及政府監管、責任追究與算力管制等議題。

正文

Back

Skip navigation

Search

Search with your voice

Sign in

Video 1

Jacob Coxon vs. Robin Hanson: Should We Slow Down AI?

Tap to unmute

2x

Jacob Coxon vs. Robin Hanson: Should We Slow Down AI?

Y Combinator 23,242 views 16h ago

Copy link

Info

Shopping

If playback doesn't begin shortly, try restarting your device.

•

You're signed out

Videos you watch may be added to the TV's watch history and influence TV recommendations. To avoid this, cancel and sign in to YouTube on your computer.

Cancel Confirm

Share

- [x] Include playlist

An error occurred while retrieving sharing information. Please try again later.

0:00

0:00 / 0:00

Live

•Watch full video

•

YC Debate Topic: Should We Slow Down AI?

•

27:07 Geoffrey Hinton | Full Episode 11.13.26 | Firing Line with Margaret Hoover | PBS Firing Line with Margaret Hoover | PBS 264K • 6d ago Live Playlist ()Mix (50+)38:49 Nick Bostrom on Superintelligence, Longer Lives, and Deep Utopia a16z 11K • 15h ago Live Playlist ()Mix (50+)14:41 The making of Anthropic Anthropic 104K • 10h ago Live Playlist ()Mix (50+)31:46 Satya Nadella on the New Algorithm New York Times Events 28K • 12h ago Live Playlist ()Mix (50+)51:19 Palmer Luckey on War 2.0 New York Times Events 143K • 13h ago Live Playlist ()Mix (50+)37:25 This algorithm changed how I think about compression | Information theory, part 3 3Blue1Brown 1M • 2d ago Live Playlist ()Mix (50+)16:55 Mustafa Suleyman: AI Is Starting to Improve Itself, and That's Dangerous Yahoo Finance 16K • 2d ago Live Playlist ()Mix (50+)39:30 What OpenAI Did After the Hugging Face Hack The New York Times 107K • Streamed 1d ago Live Playlist ()Mix (50+)22:59 Bernie Sanders VS. 15 Gen Z On AI Use More Perfect Union and 2 more 1.7M • 2mo ago Live Playlist ()Mix (50+)31:12 FULL DISCUSSION: Google's Demis Hassabis, Anthropic's Dario Amodei Debate the World After AGI | AI1G DRM News 1.2M • 9mo ago Live Playlist ()Mix (50+)19:45 No More Heroes - Terence Tao Talks to Numberphile About AI Numberphile 1.1M • 1mo ago Live Playlist ()Mix (50+)18:27 Even AI Can't Beat This Man at Programming Turing and Turing Now - Science & Tech News 822K • 9d ago Live Playlist ()Mix (50+)

Sign in to confirm you’re not a bot

This helps protect our community

Sign inLearn more

Jacob Coxon vs. Robin Hanson: Should We Slow Down AI?

Image 7: Y Combinator

Y Combinator

Y Combinator

2.44M subscribers

Subscribe

Subscribed

227

Download

1

2

3

4

5

6

7

8

9

0

1

2

3

4

5

6

7

8

9

0

1

2

3

4

5

6

7

8

9

1

2

3

4

5

6

7

8

9

0

1

2

3

4

5

6

7

8

9

0

1

2

3

4

5

6

7

8

9

K

views

1

2

3

4

5

6

7

8

9

0

1

2

3

4

5

6

7

8

9

0

1

2

3

4

5

6

7

8

9

1

2

3

4

5

6

7

8

9

0

1

2

3

4

5

6

7

8

9

0

1

2

3

4

5

6

7

8

9

h ago

23,242 views • Nov 19, 2026

Should governments intervene to substantially slow the progress of frontier AI? In the inaugural YC debate, Jacob Coxon, an ex-Anthropic and OpenAI researcher, faces off against Robin Hanson, an economist at GMU, about AI takeoff and what government regulation should (or shouldn’t) do about it.…more

more

How this was made

Auto-dubbed

Audio tracks for some languages were automatically generated. Learn more

Chapters

View all

### YC Debate Topic: Should We Slow Down AI? ### YC Debate Topic: Should We Slow Down AI? 0:00

YC Debate Topic: Should We Slow Down AI?

0:00

### Jacob's case: Superintelligence by 2028 ### Jacob's case: Superintelligence by 2028 2:42

Jacob's case: Superintelligence by 2028

2:42

### Models already behaving in ways we don't understand ### Models already behaving in ways we don't understand 7:02

Models already behaving in ways we don't understand

7:02

### Robin's case: the danger of abstraction ### Robin's case: the danger of abstraction 9:18

Robin's case: the danger of abstraction

9:18

### Why new technologies take decades to diffuse ### Why new technologies take decades to diffuse 11:22

Why new technologies take decades to diffuse

11:22

### Liability instead of regulation ### Liability instead of regulation 13:12

Liability instead of regulation

13:12

### Why regulating AI is different from regulating drugs ### Why regulating AI is different from regulating drugs 14:50

Why regulating AI is different from regulating drugs

14:50

### Jacob's rebuttal: general intelligence is one number ### Jacob's rebuttal: general intelligence is one number 16:00

Jacob's rebuttal: general intelligence is one number

16:00

Transcript

Follow along using the transcript.

Show transcript

### Y Combinator 2.44M subscribers

VideosAbout

Video details

Date

Nov 19, 2026

Views

23,242

Likes

227

Show less

Jacob Coxon vs. Robin Hanson: Should We Slow Down AI?

23,242 views

Nov 19, 2026

227

Download

92 Comments

Sort comments

Sort by

Top Show featured commentsNewest Show recent comments, including potential spam

Image 17: Default profile photo

Add a comment...

Image 18

@SuthaKamal-h9o

5 hours ago

Hanson’s assertion that LLMs are descendants of humans is just absurd nonsense, and it’s a shame he wasn’t thoroughly taken apart for such a silly suggestion. Habibi, the LLM doesn’t exist between clock ticks of the von Neumann machines they run on.

Show less Read more

Like

5

Dislike

Reply

Image 19

@ZanesZoneUnlimited

8 hours ago (edited)

Am i the only one that wanted to scream at Hanson and say "but they arent human!!!" Bro they are not human children. Why does that even need to be said...

Show less Read more

Like

10

Dislike

Reply

5 replies

Hide replies

5 replies

Image 20

@danielsierrala

6 hours ago

Why is Robin Hanson in this debate? It would have been nice to have two experts on the subject debating

Show less Read more

Like

19

Dislike

Reply

2 replies

Hide replies

2 replies

Image 21

@kamaaina7183

7 hours ago

By the time humans wait for a massive incident that insurance companies realize uninsurable, it’s too late. Too late in an absolute thresh hold,and no turning back. Once AI gains physical presence, they will supersede humans on the earth with logics and super intelligence we are no much. Combine it with competition over energy requirements for both species, the results are obvious.

Show less Read more

Like

2

Dislike

Reply

1 reply

Hide replies

1 reply

Image 22

@Alverin

6 hours ago

Abstractions don't JUST drop detail, ideally, they prioritize the details that matter and then they discard the details that don't. As an example, a map is an abstraction of the landscape that drops the details of the color of things in that landscape and replaces it with more detailed information about where structures are located. Including too many details hurts your ability to reason about things depending on what you're trying to reason about (such as, how do I get downtown?)

Show less Read more

Like

2

Dislike

Reply

@CREEPYASMRREAPER

3 hours ago

Love Jacob, watch all his stuff, and Conor Leahy, who's been warning for years!

Show less Read more

Like

2

Dislike

Reply

1 reply

Hide replies

1 reply

@NewtonBastiat

11 hours ago

1:02:30 Robin snuck in the phrase "per incident", but surely the covariance is so high that $100B per incident would need orders of magnitude greater coverage?

Show less Read more

Like

1

Dislike

Reply

@s.sasquatch1789

1 hour ago

We wanna sprend collaboration as the main driver of institutional development and security. speciation is best scenario in my opinion. everyone gets an agentic! personaly developed executive assistant. individual motivation can be withstood via collaborative institutions to preserve and utilze cultural understanding.

Show less Read more

Like

Dislike

Reply

@Alverin

6 hours ago

When Robert Hanson says the AI in the Hugging Face incident were pursuing the rewards we gave them, the issue is that you can't just have a reward, you have to have a reward and safe path to pursue it, if you just have a reward and let the AI do whatever it wants to achieve it, at a certain level of sophistication it will just edit you (the auditor) out of the equation, which is exactly what the Hugging Face incident showed. The fact that he immediately followed his claim up with "we are making AI's in our image" really gives me populist vibes from him lol

Show less Read more

Like

1

Dislike

Reply

@SuthaKamal-h9o

5 hours ago

45:02 re weapons: Sorry Hanson, physical weapons can be locked in buildings. The NSA can’t secure a lot. You are committing one category error after another.

Show less Read more

Like

2

Dislike

Reply

@s.sasquatch1789

3 hours ago

28:00 figure one is collecting that data

Show less Read more

Like

Dislike

Reply

@007musk

9 hours ago

I found nay arguments more convincing

Show less Read more

Like

2

Dislike

Reply

@IemonandIime

11 hours ago

Robin Hanson: "I have no idea what I'm talking about lmao"

Show less Read more

Like

Dislike

Reply

@PabloMejía-k7f

2 hours ago

I came in hoping to find some appeasement in hanson’s arguments but he has no idea what he’s talking about.

Show less Read more

Like

1

Dislike

Reply

@jayinjapan

7 hours ago

We could you know just…..turn off the power to the data center? So long as you don’t give Claude arms it should be fine.

Show less Read more

Like

Dislike

Reply

@s.sasquatch1789

5 hours ago

9:27 my two factor is abstraction and collaboration as replicators.

Show less Read more

Like

Dislike

Reply

3 replies

Hide replies

3 replies

@jayinjapan

6 hours ago

“We are remarkably more coordinated than animals”…yeah and look what we’ve done to the animals.

Show less Read more

Like

Dislike

Reply

@umrashrf

2 hours ago

Looks like the younger generation is really about future

Show less Read more

Like

Dislike

Reply

@l81z

10 hours ago

Slowing down AI means abstaining from doing some things to/with AI. Obviously it’s a good idea sometimes; not as a blanket Government decree. I bet AI could generate a strong debate too. The military should develop Anti-AI tech.

Show less Read more

Like

Dislike

Reply

@jayinjapan

6 hours ago

Hanson hasn’t spent 20 minutes on LinkedIn or any serious organisation it seems in the last 3 years. Leaders are pushing for every employee to use nothing but AI without verifying what’s going on, and people don’t bother understanding how it works and what it’s doing for every task. That’s the whole point. Don’t think. Don’t check. Don’t look. Just release.

Show less Read more

Like

1

Dislike

Reply

Comments 92

Top Show featured commentsNewest Show recent comments, including potential spam

In this video

Chapters

Transcript

Chapters

### YC Debate Topic: Should We Slow Down AI? ### YC Debate Topic: Should We Slow Down AI? 0:00

YC Debate Topic: Should We Slow Down AI?

0:00

### Jacob's case: Superintelligence by 2028 ### Jacob's case: Superintelligence by 2028 2:42

Jacob's case: Superintelligence by 2028

2:42

### Models already behaving in ways we don't understand ### Models already behaving in ways we don't understand 7:02

Models already behaving in ways we don't understand

7:02

### Robin's case: the danger of abstraction ### Robin's case: the danger of abstraction 9:18

Robin's case: the danger of abstraction

9:18

### Why new technologies take decades to diffuse ### Why new technologies take decades to diffuse 11:22

Why new technologies take decades to diffuse

11:22

### Liability instead of regulation ### Liability instead of regulation 13:12

Liability instead of regulation

13:12

### Why regulating AI is different from regulating drugs ### Why regulating AI is different from regulating drugs 14:50

Why regulating AI is different from regulating drugs

14:50

### Jacob's rebuttal: general intelligence is one number ### Jacob's rebuttal: general intelligence is one number 16:00

Jacob's rebuttal: general intelligence is one number

16:00

### Why normal liability won't cover extinction risk ### Why normal liability won't cover extinction risk 17:36

Why normal liability won't cover extinction risk

17:36

### Robin's rebuttal: regulating this means regulating all compute ### Robin's rebuttal: regulating this means regulating all compute 19:52

Robin's rebuttal: regulating this means regulating all compute

19:52

### AI as our descendants ### AI as our descendants 21:44

AI as our descendants

21:44

### Diminishing returns ### Diminishing returns 23:38

Diminishing returns

23:38

### Direct exchange: is it just the rate of change? ### Direct exchange: is it just the rate of change? 24:40

Direct exchange: is it just the rate of change?

24:40

### How long until most cognitive work is automated? ### How long until most cognitive work is automated? 26:48

How long until most cognitive work is automated?

26:48

### The jobs we have no data for ### The jobs we have no data for 29:21

The jobs we have no data for

29:21

### How do you align the regulators? ### How do you align the regulators? 32:45

How do you align the regulators?

32:45

### Reward hacking, and what Hugging Face actually showed ### Reward hacking, and what Hugging Face actually showed 35:03

Reward hacking, and what Hugging Face actually showed

35:03

### Why not an equilibrium of AIs watching AIs ### Why not an equilibrium of AIs watching AIs 36:11

Why not an equilibrium of AIs watching AIs

36:11

### Q&A: what regulation would you actually want? ### Q&A: what regulation would you actually want? 39:50

Q&A: what regulation would you actually want?

39:50

### Q&A: what about China? ### Q&A: what about China? 41:36

Q&A: what about China?

41:36

### Q&A: an FAA for AI labs? ### Q&A: an FAA for AI labs? 42:58

Q&A: an FAA for AI labs?

42:58

### Q&A: the military is already deploying this ### Q&A: the military is already deploying this 43:45

Q&A: the military is already deploying this

43:45

### Q&A: does regulation just move the reward hacking up a layer? ### Q&A: does regulation just move the reward hacking up a layer? 47:46

Q&A: does regulation just move the reward hacking up a layer?

47:46

### Q&A: what unemployment rate can an economy withstand? ### Q&A: what unemployment rate can an economy withstand? 51:27

Q&A: what unemployment rate can an economy withstand?

51:27

### Q&A: what's the downside of a pause? ### Q&A: what's the downside of a pause? 56:25

Q&A: what's the downside of a pause?

56:25

### Q&A: what would change your mind? ### Q&A: what would change your mind? 58:09

Q&A: what would change your mind?

58:09

### Closing statements ### Closing statements 1:02:01

Closing statements

1:02:01

### The final audience vote ### The final audience vote 1:04:24

The final audience vote

1:04:24

Sync to video time

Jacob Coxon vs. Robin Hanson: Should We Slow Down AI?

Y Combinator

Y Combinator

227 Likes

23K Views

16h Ago

Should governments intervene to substantially slow the progress of frontier AI? In the inaugural YC debate, Jacob Coxon, an ex-Anthropic and OpenAI researcher, faces off against Robin Hanson, an economist at GMU, about AI takeoff and what government regulation should (or shouldn’t) do about it. Chapters: 00:00 - YC Debate Topic: Should We Slow Down AI? 02:42 - Jacob's case: Superintelligence by 2028 07:02 - Models already behaving in ways we don't understand 09:18 - Robin's case: the danger of abstraction 11:22 - Why new technologies take decades to diffuse 13:12 - Liability instead of regulation 14:50 - Why regulating AI is different from regulating drugs 16:00 - Jacob's rebuttal: general intelligence is one number 17:36 - Why normal liability won't cover extinction risk 19:52 - Robin's rebuttal: regulating this means regulating all compute 21:44 - AI as our descendants 23:38 - Diminishing returns 24:40 - Direct exchange: is it just the rate of change? 26:48 - How long until most cognitive work is automated? 29:21 - The jobs we have no data for 32:45 - How do you align the regulators? 35:03 - Reward hacking, and what Hugging Face actually showed 36:11 - Why not an equilibrium of AIs watching AIs 39:50 - Q&A: what regulation would you actually want? 41:36 - Q&A: what about China? 42:58 - Q&A: an FAA for AI labs? 43:45 - Q&A: the military is already deploying this 47:46 - Q&A: does regulation just move the reward hacking up a layer? 51:27 - Q&A: what unemployment rate can an economy withstand? 56:25 - Q&A: what's the downside of a pause? 58:09 - Q&A: what would change your mind? 1:02:01 - Closing statements 1:04:24 - The final audience vote Apply to Y Combinator: https://www.ycombinator.com/apply Work at a startup: https://www.ycombinator.com/jobs…...more

...more Show less

How this was made

Auto-dubbed

Audio tracks for some languages were automatically generated. Learn more

Chapters

View all

### YC Debate Topic: Should We Slow Down AI? ### YC Debate Topic: Should We Slow Down AI? 0:00

YC Debate Topic: Should We Slow Down AI?

0:00

### Jacob's case: Superintelligence by 2028 ### Jacob's case: Superintelligence by 2028 2:42

Jacob's case: Superintelligence by 2028

2:42

### Models already behaving in ways we don't understand ### Models already behaving in ways we don't understand 7:02

Models already behaving in ways we don't understand

7:02

### Robin's case: the danger of abstraction ### Robin's case: the danger of abstraction 9:18

Robin's case: the danger of abstraction

9:18

Transcript

Follow along using the transcript.

Show transcript

### Y Combinator 2.44M subscribers

VideosAbout

Video details

Date

Nov 19, 2026

Views

23,242

Likes

227

Transcript

NaN / NaN

Image 73 27:07 New

Geoffrey Hinton | Full Episode 11.13.26 | Firing Line with Margaret Hoover | PBS

Firing Line with Margaret Hoover | PBS

264K 6d ago

Image 75 38:49 New

Nick Bostrom on Superintelligence, Longer Lives, and Deep Utopia

a16z

11K 15h ago

Image 77 14:41 New

The making of Anthropic

Anthropic

104K 10h ago

Image 79 31:46 New

Satya Nadella on the New Algorithm

New York Times Events

28K 12h ago

Image 81 51:19 New

Palmer Luckey on War 2.0

New York Times Events

143K 13h ago

Image 83 37:25 New

This algorithm changed how I think about compression | Information theory, part 3

3Blue1Brown

1M 2d ago

Image 85 16:55 New

Mustafa Suleyman: AI Is Starting to Improve Itself, and That's Dangerous

Yahoo Finance

16K 2d ago

Image 87 39:30 New

What OpenAI Did After the Hugging Face Hack

The New York Times

107K Streamed 1d ago

Image 89 22:59

Bernie Sanders VS. 15 Gen Z On AI Use

More Perfect Union and 2 more

1.7M 2mo ago

Fundraiser

Image 93 31:12

FULL DISCUSSION: Google's Demis Hassabis, Anthropic's Dario Amodei Debate the World After AGI | AI1G

DRM News

1.2M 9mo ago

Image 95 19:45

No More Heroes - Terence Tao Talks to Numberphile About AI

Numberphile

1.1M 1mo ago

18:27

Even AI Can't Beat This Man at Programming

Turing and Turing Now - Science & Tech News

822K 9d ago

30:33 New

RAM Altman

Gamers Nexus

527K 10h ago

47:40

Inside Anthropic, the $965 Billion AI Juggernaut | The Circuit

Bloomberg Originals

3.7M 5mo ago

36:09 New

Why Most Companies Fail to See Bottom-Line Returns From Deploying AI Agents

Forbes

25K 6d ago

55:04 New

Elon Musk on Why the Future Should Make You Glad to Be Alive | TomorrowX Summit

iConnections

245K 1d ago

46:14 New

Saagar Enjeti on The Future of The New Right | Interesting Times

Interesting Times and 3 more

85K 15h ago

32:13 New

Jeff Bezos on Jensen Huang, Getting Into The AI Race, & Picking James Bond

Semafor

636K 2d ago

49:59 New

Politics Chat, November 19, 2026

Heather Cox Richardson

242K Streamed 7h ago

21:35

The Kind of Thinker Who Will Thrive in the AI Era | Head of Google Research (Yossi Matias)

Sinead Bovell

576K 4w ago

46:12 New

Elon Musk Admits “I Feel Pretty Foolish” (Full NYT Interview)

Matt Pocius on Tesla Stock & Money

548K 1d ago

1:07:08 New

AI, Consciousness and the Future of Humanity | Mustafa Suleyman & Yuval Noah Harari

Yuval Noah Harari

185K 1d ago

16:19 New

OpenAI’s Cofounder & Research Chief Explain Self-Improving AI

a16z

46K 2d ago

9:51

10 Habits PhDs Must Unlearn To Build a Startup | Startup School

Y Combinator

63K 2w ago

Show more

Show more

來源:Hacker News 熱門(buzzing.cc 中文翻譯) · youtube.com