Basil Halperin

Andrey Fradkin@AndreyFradkin

0

19

764

Herbie Bradley@herbiebradley·3 Mar

Agree with all of this I am increasingly sure that the "Anthropic house view" on takeoff is overly confident and unhedged for a disappointment moment if the result of automating AI R&D is underwhelming economically

Last week, @jamescham and I hosted a dinner discussing the economics of fast takeoff with representatives of leading camps from tech, economics, and EA (not mutually exclusive). Some themes from the discussion.

English

5

0

50

6.7K

Basil Halperin@BasilHalperin·2 Mar

@RandomSprint Share of GDP rather than dollar value but (from Benedict Evans):

English

0

2

98

RandomSprint🧭@RandomSprint·2 Mar

@BasilHalperin I wonder what it looks like stacked with traditional advertising.

English

0

3

130

Basil Halperin@BasilHalperin·2 Mar

Internet advertising revenue growth before vs. after dot-com Boom: +175%/yr Bust: +25%/yr

English

0

33

1.7K

Basil Halperin@BasilHalperin·23 Şub

@arpitrage If we're talking about aggregate demand -- rather than talking about insurance or redistribution -- why not our usual tool for managing AD?

English

Depressing but intriguing piece on how far AI white collar worker disruption might take us by @Citrini7 and @alapshah1 citriniresearch.com/p/2028gic

0

2

360

Arpit Gupta@arpitrage·23 Şub

Having digested this a bit more, I think the main thing it misses is the role of fiscal policy transfers to make up for aggregate demand. In a world of turbo charged AI productivity, fiscal credibility should enable keeping up AD by writing checks

Arpit Gupta@arpitrage

English

13

3

96

23.4K

Basil Halperin@BasilHalperin·13 Şub

@pawtrammell @AndreyFradkin @dwarkesh_sp Glen Weyl has convinced me "neoliberal capitalism offers no framework for grappling with increasing returns" -- more broadly on your last point -- though, if that's sort of what you're getting at! x.com/glenweyl/statu…

(((E. Glen Weyl/衛谷倫))) ⿻ 🇺🇸/🇩🇪/🇹🇼@glenweyl

3) standard neo-liberalism capitalism, no matter how much social insurance you pile on top, offers no framework for grappling with increasing returns, 4) that the @NiskanenCenter project makes no progress on this and in fact willfully ignores pressing related policy questions

English

0

134

Basil Halperin@BasilHalperin·13 Şub

(0) Not sure I follow what you're saying -- "single-sector" r would be lower than traditionally estimated (ie phi would be higher). Intuition might be: if Moore's Law has been happening NOT JUST by sending more scientists into the semiconductor mines, BUT ALSO by getting spillovers from other sectors, then diminishing returns must be worse than previously believed Maybe you're talking theoretically not empirically -- in which case yes!, *holding r constant*, introducing spillovers makes it easier to generate a singularity ("lesson 2" basilhalperin.com/papers/singula…) (1) 👍👍. (My idiosyncratic view is that all optimal tax papers should start with a sort of 'land acknowledgment' to Pigou and George [...pun intended 😏]) (2) I didn't see your writing on this increasing returns point before -- makes sense to me what you write, though to be clear we can have mechanisms for dealing with increasing returns that aren't "full expropriation and national champions" (eg letting the private sector do it and just eating any markups that results, or lighter touch regulation)

English

0

66

Basil Halperin retweetledi

Andrey Fradkin@AndreyFradkin·10 Şub

New pod up with @BasilHalperin. We discuss the assumptions needed to allow for explosive economic growth, Basil's extrapolations from the METR graph, and his criticism of @dwarkesh_sp and @pawtrammell's blog post.

English

5

12

68

21.3K

Basil Halperin@BasilHalperin·12 Şub

@Jsevillamol But what convinced you! I feel like work, both theoretical and empirical, for tracking and forecasting robotics progress is ~nonexistent (where's the bio anchors for robotics)

English

0

5

348

Jaime Sevilla@Jsevillamol·12 Şub

Last autumn, I became more convinced that we are not that far from practical, widely deployed everyday robots. My colleagues @datagenproc and Yann Riviere have now taken a deep dive into the state of the field. Big thanks to @girishsastry for generously funding this work!

Epoch AI@EpochAIResearch

How far are autonomous robots from real-world deployment? We reviewed robot capabilities across 14 concrete tasks. We find that while navigation is deployed at scale, manipulation mostly is not.

English

4

58

5.3K

Basil Halperin@BasilHalperin·12 Şub

@oliverwkim Re: demonstration effects / the power of ideas Buera et al (2011), "Learning the Wealth of Nations", is interesting on this -> on countries "learning to be neoliberal" from neighbors, after 70s/80s Would be interesting to port over... (& great post! / you may be familiar!)

English

global-developments.org/p/how-africa-w…

0

6

455

Oliver Kim@oliverwkim·10 Şub

ZXX

4

10

32

3.5K

Oliver Kim@oliverwkim·10 Şub

🚨 New 🌐Global Developments🌐 blog post 🚨 My review of Joe Studwell's How Africa Works! Link in reply.

English

5

59

285

51.5K

Basil Halperin@BasilHalperin·11 Şub

@BenSManning @sj_manning @SBenzell

QME

Andrey Fradkin@AndreyFradkin

4

39

Benjamin Manning@BenSManning·10 Şub

@sj_manning @SBenzell, I need you to start having a weird "thing" on your whiteboard that changes every episode, and I try to figure out what it is.

English

3

0

6

189

sam manning@sj_manning·10 Şub

Oh man. Excited for this one

New pod up with @BasilHalperin. We discuss the assumptions needed to allow for explosive economic growth, Basil's extrapolations from the METR graph, and his criticism of @dwarkesh_sp and @pawtrammell's blog post.

English

17

1.5K

Basil Halperin retweetledi

Stefan Schubert@StefanFSchubert·10 Şub

I love a podcast with (reasoned) disagreement. Too many podcasts lack real intellectual engagement.

Great conversation with Andrey and Seth... here's one bit I liked on blogging vs. academic publishing

English

Andrey Fradkin@AndreyFradkin

2

27

3.5K

Basil Halperin@BasilHalperin·10 Şub

Great conversation with Andrey and Seth... here's one bit I liked on blogging vs. academic publishing

New pod up with @BasilHalperin. We discuss the assumptions needed to allow for explosive economic growth, Basil's extrapolations from the METR graph, and his criticism of @dwarkesh_sp and @pawtrammell's blog post.

English

6

42

11.5K

Basil Halperin@BasilHalperin·10 Şub

@AndreyFradkin Just the bare minimum 😏. Thanks for having me on!

English

2

112

Andrey Fradkin@AndreyFradkin·10 Şub

Also, he's still investing in his 401(k)! Link here: empiricrafting.substack.com/p/basil-halper…

English

0

5

689

Basil Halperin@BasilHalperin·9 Şub

@GeorgeSelgin (indeed!!)

English

147

George Selgin@GeorgeSelgin·9 Şub

Good advice for teachers, from the well-taught @BasilHalperin.

Some teaching principles: basilhalperin.com/essays/teachin… 0. Have extreme empathy. 1. Grading should be predictable. 2. Enthusiasm matters! Show that you care – channel your enjoyment! 3. Actively solicit feedback. ...

English

0

8

2.2K

Basil Halperin@BasilHalperin·9 Şub

@JosephNWalker You made the job easy 😜

English

2

183

Joseph Noel Walker@JosephNWalker·9 Şub

Great post! Having been tutored by Basil in some growth econ for the podcast, I can attest that he practices what he preaches. Basil as tutor was especially great because: 1. He had v accurate model of me - avoided curse of knowledge 2. Never BSed me if he didn't have an answer

Some teaching principles: basilhalperin.com/essays/teachin… 0. Have extreme empathy. 1. Grading should be predictable. 2. Enthusiasm matters! Show that you care – channel your enjoyment! 3. Actively solicit feedback. ...

English

1

12

2.9K

Basil Halperin@BasilHalperin·9 Şub

@janzilinsky ✍️clearly✍️articulate✍️expectations✍️

English

3

271

Jan Zilinsky@janzilinsky·9 Şub

This is excellent. I started spending a lot more time on articulating expectations, which helped with both grading predictability and lessening anxiety

Some teaching principles: basilhalperin.com/essays/teachin… 0. Have extreme empathy. 1. Grading should be predictable. 2. Enthusiasm matters! Show that you care – channel your enjoyment! 3. Actively solicit feedback. ...

English

0

2

617

Basil Halperin@BasilHalperin·8 Şub

@40yoap Cooked my ass ngl

English

0

2

153

Siegmeyer of Catarina@40yoap·8 Şub

@BasilHalperin TaLking about teaching is a huge red flag and a bad signal for APs. 🫣

English

0

1

221

Basil Halperin@BasilHalperin·8 Şub

Some teaching principles: basilhalperin.com/essays/teachin… 0. Have extreme empathy. 1. Grading should be predictable. 2. Enthusiasm matters! Show that you care – channel your enjoyment! 3. Actively solicit feedback. ...

English

3

10

44

7.6K

Basil Halperin@BasilHalperin·8 Şub

24. Send encouraging emails at the end of the semester to relevant students. 25. Smile (see principle #2).

English

3

461

Basil Halperin@BasilHalperin·8 Şub

21. Sitting in on colleagues’ lectures for the same course is useful. 22. Teaching dialectically and showing the history of thought is useful. 23. Get student buy-in on the electronics policy.

English