Manifold

Steve discusses 10 key graphs related to meritocracy and university admissions. Predictive power of SATs and other factors in elite admissions decisions. College learning outcomes - what do students learn? The four paths to elite college admission. Laundering prestige at the Ivies.

Slides: https://docs.google.com/presentation/d/1n-nwoeKe_DcA5tJxTwqTeZBEY7nObxkujKLxVfAzRAY/edit?usp=sharing

CLA and College Learning outcomes:

https://infoproc.blogspot.com/2015/01/measuring-college-learning-outcomes.html

Harvard Veritas: Interview with a recent graduate

https://infoproc.blogspot.com/2022/08/harvard-veritas-interview-with-recent.html

Defining Merit - Human Capital and Harvard University:

https://infoproc.blogspot.com/2009/11/defining-merit.html

Chapter markers:

0:00 Introduction 

1:28 University of California system report and the use of SAT scores admissions

8:04 Longitudinal study on gifted students and SAT scores (SMPY)

12:53 Unprecedented data on earnings outcomes and SAT scores

15:43 How SAT scores and university pedigree influence opportunities at elite firms

17:35 Non-academic factors fail to predict student success

20:49 Predicted earnings

24:24 Measured benefit of Ivy Plus attendance

28:25 CLA: 13 university study on college learning outcomes

32:34 Does college education improve generalist skills and critical thinking?

42:15 The composition of elite universities: 4 paths to admission

48:12 What happened to meritocracy?

51:48 Hard versus Soft career tracks

54:43 Cognitive elite at Ivies vs state flagship universities

57:11 What happened to Caltech?

Music used with permission from Blade Runner Blues Livestream improvisation by State Azure.

--

Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University. Previously, he was Senior Vice President for Research and Innovation at MSU and Director of the Institute of Theoretical Science at the University of Oregon. Hsu is a startup founder (SuperFocus.ai, SafeWeb, Genomic Prediction, Othram) and advisor to venture capital and other investment firms. He was educated at Caltech and Berkeley, was a Harvard Junior Fellow, and has held faculty positions at Yale, the University of Oregon, and MSU.

Announcing this for some friends at Mechanize - a startup that builds environments for training and evaluating frontier LLMs. Its customers include the top AI labs, and it has contributed to the breakthrough in coding capabilities of frontier models. 

Mechanize is hiring! 

https://mechanize.work/b/hsu 

Compensation is extremely competitive. For technical roles, $300-500k. They are also seeking smart generalists. 

For example: 

Research Engineer, Alignment: Build evals that test for misaligned model behaviors  $500K salary 

Puzzle Maker: Design interesting and original puzzles that LLMs can’t yet solve  $300K salary 

Mechanize understands that my readership is highly selected. There is a VERY GOOD CHANCE you will be interviewed if you apply via the link above.

Aella is a sex worker, sex researcher, and data scientist.

Aella on X: https://twitter.com/Aella_Girl

Interviews with ex-prostitutes on the pimp life (Las Vegas)

https://www.youtube.com/watch?v=eAlXdyjmWUo&ab_channel=PeterSantenello

An earlier Aella interview with Reason:

https://reason.com/podcast/2022/04/27/aella-libertarian-sex-worker-turned-data-scientist/

Steve and Aella discuss:

• (00:00) - Introduction

Music used with permission from Blade Runner Blues Livestream improvisation by State Azure.

--

Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University. Previously, he was Senior Vice President for Research and Innovation at MSU and Director of the Institute of Theoretical Science at the University of Oregon. Hsu is a startup founder (SuperFocus.ai, SafeWeb, Genomic Prediction) and advisor to venture capital and other investment firms. He was educated at Caltech and Berkeley, was a Harvard Junior Fellow, and has held faculty positions at Yale, the University of Oregon, and MSU.

Announcing this for some friends at Mechanize - a startup that builds environments for training and evaluating frontier LLMs. Its customers include the top AI labs, and it has contributed to the breakthrough in coding capabilities of frontier models. 

Mechanize is hiring! 

https://mechanize.work/b/hsu 

Compensation is extremely competitive. For technical roles, $300-500k. They are also seeking smart generalists. 

For example: 

Research Engineer, Alignment: Build evals that test for misaligned model behaviors  $500K salary 

Puzzle Maker: Design interesting and original puzzles that LLMs can’t yet solve  $300K salary 

Mechanize understands that my readership is highly selected. There is a VERY GOOD CHANCE you will be interviewed if you apply via the link above.

Tim Dettmers develops computationally efficient methods for deep learning. He is a leader in quantization: coarse graining of large neural networks to increase speed and reduce hardware requirements.

Tim developed 4-and 8-bit quantizations enabling training and inference with large language models on affordable GPUs and CPUs - i.e., as commonly found in home gaming rigs.

Tim and Steve discuss: Tim's background and current research program, large language models, quantization and performance, democratization of AI technology, the open source Cambrian explosion in AI, and the future of AI.

0:00 Introduction and Tim’s background

18:02 Tim's interest in the efficiency and accessibility of large language models

38:05 Inference, speed, and the potential for using consumer GPUs for running large language models

45:55 Model training and the benefits of quantization with QLoRA

57:14 The future of AI and large language models in the next 3-5 years and beyond

Tim's site: https://timdettmers.com/

Tim on GitHub: https://github.com/TimDettmers

Music used with permission from Blade Runner Blues Livestream improvisation by State Azure.

--

Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University. Previously, he was Senior Vice President for Research and Innovation at MSU and Director of the Institute of Theoretical Science at the University of Oregon. Hsu is a startup founder (SuperFocus.ai, SafeWeb, Genomic Prediction) and advisor to venture capital and other investment firms. He was educated at Caltech and Berkeley, was a Harvard Junior Fellow, and has held faculty positions at Yale, the University of Oregon, and MSU.

Please send any questions or suggestions to [email protected] or Steve on Twitter @hsu_steve.

Announcing this for some friends at Mechanize - a startup that builds environments for training and evaluating frontier LLMs. Its customers include the top AI labs, and it has contributed to the breakthrough in coding capabilities of frontier models. 

Mechanize is hiring! 

https://mechanize.work/b/hsu 

Compensation is extremely competitive. For technical roles, $300-500k. They are also seeking smart generalists. 

For example: 

Research Engineer, Alignment: Build evals that test for misaligned model behaviors  $500K salary 

Puzzle Maker: Design interesting and original puzzles that LLMs can’t yet solve  $300K salary 

Mechanize understands that my readership is highly selected. There is a VERY GOOD CHANCE you will be interviewed if you apply via the link above.

Paul Huang is a journalist and research fellow with the Taiwan Public Opinion Foundation. He is currently based in Taipei, Taiwan.

Sample articles:

Taiwan’s Military Has Flashy American Weapons but No Ammo (in Foreign Policy): https://foreignpolicy.com/2020/08/20/taiwan-military-flashy-american-weapons-no-ammo/

Taiwan’s Military Is a Hollow Shell (Foreign Policy): https://foreignpolicy.com/2020/02/15/china-threat-invasion-conscription-taiwans-military-is-a-hollow-shell/

Steve and Paul discuss: 

0:00 Introduction 

1:44 Paul’s background; the Green Party (DPP) and Blue Party (KMT) in Taiwan

4:40 How the Taiwanese people view themselves vs mainland Chinese

15:02 Taiwan taboos: politics and military preparedness

15:27 Effect of Ukraine conflict on Taiwanese opinion

29:56 Lack of realistic military planning

37:20 Is there a political solution to reunification with China? What influence does the U.S. have?

51:34 The likelihood of peaceful reunification of Taiwan and China

56:45 Honest views on Taiwanese and U.S. military readiness for a

conflict with China

Music used with permission from Blade Runner Blues Livestream improvisation by State Azure.

--

Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University. Previously, he was Senior Vice President for Research and Innovation at MSU and Director of the Institute of Theoretical Science at the University of Oregon. Hsu is a startup founder (SuperFocus.ai, SafeWeb, Genomic Prediction) and advisor to venture capital and other investment firms. He was educated at Caltech and Berkeley, was a Harvard Junior Fellow, and has held faculty positions at Yale, the University of Oregon, and MSU.

Please send any questions or suggestions to [email protected] or Steve on Twitter @hsu_steve.

Announcing this for some friends at Mechanize - a startup that builds environments for training and evaluating frontier LLMs. Its customers include the top AI labs, and it has contributed to the breakthrough in coding capabilities of frontier models. 

Mechanize is hiring! 

https://mechanize.work/b/hsu 

Compensation is extremely competitive. For technical roles, $300-500k. They are also seeking smart generalists. 

For example: 

Research Engineer, Alignment: Build evals that test for misaligned model behaviors  $500K salary 

Puzzle Maker: Design interesting and original puzzles that LLMs can’t yet solve  $300K salary 

Mechanize understands that my readership is highly selected. There is a VERY GOOD CHANCE you will be interviewed if you apply via the link above.

Steve Hsu, Richard Hanania, and Rob Henderson were scheduled for a June 2023 panel as part of the University of Austin (UATX) Forbidden Courses series. Steve missed the panel due to travel issues, but the three have gathered on this podcast to recreate the fun!

They discuss:

0:00 Introduction
1:20 The University of Austin and forbidden courses
17:37 Will woke campus culture change anytime soon?
29:57 Common people vs elites on affirmative action
35:42 Why it’s uncomfortable to disagree about affirmative action
41:22 Fraud and misrepresentation in higher ed
44:20 The adversity carveout in the Supreme Court affirmative action ruling
50:10 Standardized testing and elite university admissions
1:06:18 Divergent views among racial and ethnic groups on affirmative action; radicalized Asian American males
1:10:00 Differences between East and South Asians in the West  
1:23:03 Class-based preferences and standardized tests
1:31:57 Rob Henderson’s next move 

LINKS

Richard Hanania’s new book: The Origins of Woke: Civil Rights Law, Corporate America, and the Triumph of Identity Politics: https://www.harpercollins.com/products/the-origins-of-woke-richard-hanania?variant=41004650528802

Richard Hanania’s newsletter: https://www.richardhanania.com/

The Center for the Study of Partisanship and Ideology: https://www.cspicenter.com/

Rob Henderson’s newsletter: https://www.robkhenderson.com/

Rob Henderson’s new book: Troubled: A Memoir of Foster Care, Family, and Social Class: https://www.simonandschuster.com/books/Troubled/Rob-Henderson/9781982168537

UATX: https://www.uaustin.org/forbidden-courses

Music used with permission from Blade Runner Blues Livestream improvisation by State Azure.

--

Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University. Previously, he was Senior Vice President for Research and Innovation at MSU and Director of the Institute of Theoretical Science at the University of Oregon. Hsu is a startup founder (Superfocus.ai, SafeWeb, Genomic Prediction) and advisor to venture capital and other investment firms. He was educated at Caltech and Berkeley, was a Harvard Junior Fellow, and has held faculty positions at Yale, the University of Oregon, and MSU.

Please send any questions or suggestions to [email protected] or Steve on Twitter @hsu_steve.

Announcing this for some friends at Mechanize - a startup that builds environments for training and evaluating frontier LLMs. Its customers include the top AI labs, and it has contributed to the breakthrough in coding capabilities of frontier models. 

Mechanize is hiring! 

https://mechanize.work/b/hsu 

Compensation is extremely competitive. For technical roles, $300-500k. They are also seeking smart generalists. 

For example: 

Research Engineer, Alignment: Build evals that test for misaligned model behaviors  $500K salary 

Puzzle Maker: Design interesting and original puzzles that LLMs can’t yet solve  $300K salary 

Mechanize understands that my readership is highly selected. There is a VERY GOOD CHANCE you will be interviewed if you apply via the link above.

Richard Sander is Jesse Dukeminier Professor at UCLA Law School. AB Harvard, JD, PhD (Economics) Northwestern.

Steve and Richard discuss the recent Supreme Court ruling in Students For Fair Admissions vs Harvard and UNC.

Sander has studied the structure and effects of law school admissions policies. He coined the term "Mismatch" to describe negative consequences resulting from large admissions preferences.

0:00 Introduction

1:09 Richard Sander’s initial reaction to the Supreme Court ruling

4:03 How data influenced the court’s decision

7:58 Overview of the court’s ruling

11:27 Carve outs in the court’s ruling

16:59 The litigation landscape

21:25 Workarounds to race-blind admissions and the UC system

32:22 Remedies: What will happen with Harvard and UNC now?

38:02 The landscape of college admissions

44:47 Effects of the Supreme Court ruling beyond higher education

LINKS

SCOTUS decision on Affirmative Action:

https://int.nyt.com/data/documenttools/supreme-court-decision-on-race-based-admissions/0a725aaabb459074/full.pdf

Richard Sander’s amicus brief: https://www.supremecourt.gov/DocketPDF/20/20-1199/222805/20220509134743957_20-1199%2021-707%20Amicus%20BOM.pdf

Richard Sander on SCOTUS Oral Arguments: Affirmative Action and Discrimination against Asian Americans at Harvard and UNC: https://www.manifold1.com/episodes/richard-sander-on-scotus-oral-arguments-affirmative-action-and-discrimination-against-asian-americans-at-harvard-and-unc

Richard Sander: Affirmative Action, Mismatch Theory, and Academic Freedom: https://www.manifold1.com/episodes/richard-sander-affirmative-action-mismatch-theory-academic-freedom-6

Music used with permission from Blade Runner Blues Livestream improvisation by State Azure.

--

Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University. Previously, he was Senior Vice President for Research and Innovation at MSU and Director of the Institute of Theoretical Science at the University of Oregon. Hsu is a startup founder (Superfocus.ai, SafeWeb, Genomic Prediction) and advisor to venture capital and other investment firms. He was educated at Caltech and Berkeley, was a Harvard Junior Fellow, and has held faculty positions at Yale, the University of Oregon, and MSU.

Announcing this for some friends at Mechanize - a startup that builds environments for training and evaluating frontier LLMs. Its customers include the top AI labs, and it has contributed to the breakthrough in coding capabilities of frontier models. 

Mechanize is hiring! 

https://mechanize.work/b/hsu 

Compensation is extremely competitive. For technical roles, $300-500k. They are also seeking smart generalists. 

For example: 

Research Engineer, Alignment: Build evals that test for misaligned model behaviors  $500K salary 

Puzzle Maker: Design interesting and original puzzles that LLMs can’t yet solve  $300K salary 

Mechanize understands that my readership is highly selected. There is a VERY GOOD CHANCE you will be interviewed if you apply via the link above.

In this episode, Steve talks to three AI engineers from his startup SuperFocus.AI.

0:00 Introduction

1:06 The Google memo and open-source AI 

14:41 Sparsification and the size of models: AI on your phone?

30:16 When will AI take over ordinary decision-making from humans?

34:50 Rapid advances in AI: a view from inside

41:28 AI Doomers and Alignment

Links to earlier episodes on AI and LLMs.

Artificial Intelligence & Large Language Models: Oxford Lecture — #35: https://www.manifold1.com/episodes/artificial-intelligence-large-language-models-oxford-lecture-35

Bing vs. Bard, US-China STEM Competition, and Embryo Screening — #30: https://www.manifold1.com/episodes/bing-vs-bard-us-china-stem-competition-and-embryo-screening-30

ChatGPT, LLMs, and AI — #29: https://www.manifold1.com/episodes/chatgpt-llms-and-ai

Music used with permission from Blade Runner Blues Livestream improvisation by State Azure.

--

Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University. Previously, he was Senior Vice President for Research and Innovation at MSU and Director of the Institute of Theoretical Science at the University of Oregon. Hsu is a startup founder (Superfocus.ai, SafeWeb, Genomic Prediction) and advisor to venture capital and other investment firms. He was educated at Caltech and Berkeley, was a Harvard Junior Fellow, and has held faculty positions at Yale, the University of Oregon, and MSU.

Please send any questions or suggestions to [email protected] or Steve on Twitter @hsu_steve.

Announcing this for some friends at Mechanize - a startup that builds environments for training and evaluating frontier LLMs. Its customers include the top AI labs, and it has contributed to the breakthrough in coding capabilities of frontier models. 

Mechanize is hiring! 

https://mechanize.work/b/hsu 

Compensation is extremely competitive. For technical roles, $300-500k. They are also seeking smart generalists. 

For example: 

Research Engineer, Alignment: Build evals that test for misaligned model behaviors  $500K salary 

Puzzle Maker: Design interesting and original puzzles that LLMs can’t yet solve  $300K salary 

Mechanize understands that my readership is highly selected. There is a VERY GOOD CHANCE you will be interviewed if you apply via the link above.

David Paul Goldman is an American economic strategist and author, best known for his series of online essays in the Asia Times under the pseudonym Spengler with the first column published January 1, 2000.

Steve and David discuss:

0:00 Introduction

2:22 David’s background in music, finance, and Asia

16:55 Looking back at the financial crisis

23:04 Rise of the Chinese economy

29:44 How Huawei’s strength is tied to China’s economic power

36:49 Competition in the global electric vehicles market

38:06 Why David thinks European countries like Germany will become closer with China

45:29 U.S. manufacturing is falling behind

52:08 Potential for war and ongoing U.S.-China competition

1:04:07 Predictions for Taiwan

Links:

David Goldman in Wikipedia:

https://en.wikipedia.org/wiki/David_P._Goldman

Spengler column:

https://asiatimes.com/author/spengler/

You Will Be Assimilated: China's Plan to Sino-form the World

https://www.amazon.com/You-Will-Be-Assimilated-Sino-form/dp/1642935409

Prisoner’s Dilemma: Avoiding war with China is the most urgent task of our lifetime

https://claremontreviewofbooks.com/prisoners-dilemma/

David Goldman articles in Claremont Review:

https://claremontreviewofbooks.com/author/david-p-goldman/

Music used with permission from Blade Runner Blues Livestream improvisation by State Azure.

--

Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University. Previously, he was Senior Vice President for Research and Innovation at MSU and Director of the Institute of Theoretical Science at the University of Oregon. Hsu is a startup founder (Superfocus.ai, SafeWeb, Genomic Prediction) and advisor to venture capital and other investment firms. He was educated at Caltech and Berkeley, was a Harvard Junior Fellow, and has held faculty positions at Yale, the University of Oregon, and MSU.

Please send any questions or suggestions to [email protected] or Steve on Twitter @hsu_steve.

Announcing this for some friends at Mechanize - a startup that builds environments for training and evaluating frontier LLMs. Its customers include the top AI labs, and it has contributed to the breakthrough in coding capabilities of frontier models. 

Mechanize is hiring! 

https://mechanize.work/b/hsu 

Compensation is extremely competitive. For technical roles, $300-500k. They are also seeking smart generalists. 

For example: 

Research Engineer, Alignment: Build evals that test for misaligned model behaviors  $500K salary 

Puzzle Maker: Design interesting and original puzzles that LLMs can’t yet solve  $300K salary 

Mechanize understands that my readership is highly selected. There is a VERY GOOD CHANCE you will be interviewed if you apply via the link above.

This week's episode is based on a lecture Steve gave to an audience of theoretical physicists at Oxford University. The topic is artificial intelligence and large language models.

Lecture slides: https://docs.google.com/presentation/d/1xiMeeRMVpB-_W66BnyRyUAtrLlUwQNlndqbVcguKK8U/edit?usp=sharing

Chapter markers:

0:00 Introduction

2:31 Deep Learning and Neural Networks; history and mathematical results

21:15 Embedding space, word vectors

31:53 Next word prediction as objective function

34:08 Attention is all you need

37:09 Transformer architecture

44:54 The geometry of thought

52:57 What can LLMs do? Sparks of AGI

1:02:41 Hallucination  

1:14:40 SuperFocus testing and examples

1:18:40 AI landscape, AGI, and the future

Music used with permission from Blade Runner Blues Livestream improvisation by State Azure.

--

Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University. Previously, he was Senior Vice President for Research and Innovation at MSU and Director of the Institute of Theoretical Science at the University of Oregon. Hsu is a startup founder (Superfocus.ai, SafeWeb, Genomic Prediction) and advisor to venture capital and other investment firms. He was educated at Caltech and Berkeley, was a Harvard Junior Fellow, and has held faculty positions at Yale, the University of Oregon, and MSU.

Please send any questions or suggestions to [email protected] or Steve on Twitter @hsu_steve.

Announcing this for some friends at Mechanize - a startup that builds environments for training and evaluating frontier LLMs. Its customers include the top AI labs, and it has contributed to the breakthrough in coding capabilities of frontier models. 

Mechanize is hiring! 

https://mechanize.work/b/hsu 

Compensation is extremely competitive. For technical roles, $300-500k. They are also seeking smart generalists. 

For example: 

Research Engineer, Alignment: Build evals that test for misaligned model behaviors  $500K salary 

Puzzle Maker: Design interesting and original puzzles that LLMs can’t yet solve  $300K salary 

Mechanize understands that my readership is highly selected. There is a VERY GOOD CHANCE you will be interviewed if you apply via the link above.

In collaboration with her husband Malcolm Collins, Simone is an author (The Pragmatist's Guide to Life, Relationships, Sexuality, Governance, and Crafting Religion), education reform advocate (CollinsInstitute.org), pronatalism activist (Pronatalist.org), and business operator (Travelmax.com).

Note: the YouTube version of this interview includes video of Steve and Simone.

Steve and Simone discuss:

0:00 Introduction

1:49 Simone's IVF journey, and embryo screening

40:02 Dating; girl autists

55:41 Finding a husband, systematized

1:09:57 Pronatalism 

Music used with permission from Blade Runner Blues Livestream improvisation by State Azure.

—

Steve Hsu is Professor of Theoretical Physics and of Computational Mathematics, Science, and Engineering at Michigan State University. Previously, he was Senior Vice President for Research and Innovation at MSU and Director of the Institute of Theoretical Science at the University of Oregon. Hsu is a startup founder (Superfocus.ai, SafeWeb, Genomic Prediction) and advisor to venture capital and other investment firms. He was educated at Caltech and Berkeley, was a Harvard Junior Fellow, and has held faculty positions at Yale, the University of Oregon, and MSU.

Please send any questions or suggestions to [email protected] or Steve on Twitter @hsu_steve.

Announcing this for some friends at Mechanize - a startup that builds environments for training and evaluating frontier LLMs. Its customers include the top AI labs, and it has contributed to the breakthrough in coding capabilities of frontier models. 

Mechanize is hiring! 

https://mechanize.work/b/hsu 

Compensation is extremely competitive. For technical roles, $300-500k. They are also seeking smart generalists. 

For example: 

Research Engineer, Alignment: Build evals that test for misaligned model behaviors  $500K salary 

Puzzle Maker: Design interesting and original puzzles that LLMs can’t yet solve  $300K salary 

Mechanize understands that my readership is highly selected. There is a VERY GOOD CHANCE you will be interviewed if you apply via the link above.