Hacking State 13 - Steve Hsu: Polygenic Embryo Selection, Improving LLMs, & Getting Nearly Cancelled

 

Alex Murshak is a Michigan State grad working as an AI engineer in Austin TX. This conversation is Episode 13 of his podcast Hacking State.

Episode description:

Steve and I speak about polygenic risk scoring and embryo selection, using AI to predict phenotype from genotype, in-vitro fertilization (IVF), egg freezing, eugenic public policy, addressing Christians' and right-wing traditionalists' concerns over reproductive technology, Superfocus AI's plan to eliminate hallucination in large language models (LLMs) by separating memory from inference, introspection for LLM error correction, and surviving the failed cancellation attempt at MSU.

Huawei and the US-China Chip War — Manifold #44

 

TP Huang is a computer scientist and analyst of global technology development. He posts often on X: https://twitter.com/tphuang 

Audio-only and Transcript: https://www.manifold1.com/episodes/huawei-and-the-us-china-chip-war-44 

Steve and TP discuss: 

0:00 Introduction: TP Huang and semiconductor technology 

5:40 Huawei’s new phone and SoC 

23:19 SMIC 7nm chip production in China: Yield and economics 

28:21 Impact on Qualcomm 

36:08 U.S. sanctions solved the coordination problem for China semiconductor companies 

42:48 5G modem and RF chips: impact on Qualcomm, Broadcom, Apple, etc. 

47:14 5G and Huawei 52:50 Satellite capabilities of Huawei phones 

56:46 Huawei vs Apple and Chinese consumers 

1:01:33 Chip War and AI model training

Meritocracy, SAT Scores, and Laundering Prestige at Elite Universities — Manifold #43

 

I discuss 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 

Audio Only and Transcript: 

https://www.manifold1.com/episodes/meritocracy-sat-scores-and-laundering-prestige-at-elite-universities-43

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?

The World of Yesterday: Steve Hsu on polygenic scores, gene editing, human flourishing

 

I really enjoyed this long conversation with Dan Schulz, an MSU engineering grad who works in tech. Dan did his homework and we covered a lot of important topics.

Transcript: https://www.danschulz.co/p/3-steve-hsu 

Apple: https://apple.co/44eTSrJ 

Spotify: https://spoti.fi/3P03SzN 

Timestamps

 

(0:00:00) - Intro 

(0:00:33) -  Genomic Prediction 

(0:05:54) - IVF 

(0:12:34) - Phenotypic data 

(0:15:42) - Predicting height 

(0:28:27) - Pleiotropy 

(0:39:14) - Optimism 

(0:45:03) - Gene editing 

(0:48:27) - Super intelligent humans 

(1:01:27) - Regulation 

(1:06:36) - Human values 

(1:17:38) - Should you do IVF? 

(1:26:06) - 23andMe 

(1:29:03) - Jeff Bezos 

(1:34:29) - Richard Feynman 

(1:43:43) - Where are the superstar physicists? 

(1:45:37) - Is physics a good field to get into?

Aella: Sex Work, Sex Research, and Data Science — Manifold #42

 

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 

An earlier Aella interview with Reason: 

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

Audio-only and Transcript:

https://www.manifold1.com/episodes/aella-sex-work-sex-research-and-data-science-42

Steve and Aella discuss: 

(00:00) - Introduction 

(01:22) - Aella's background and upbringing 

(12:45) - Aella's experiences as a sex worker and escorting 

(29:52) - Pimp culture 

(38:01) - Seeking Arrangement 

(43:50) - Cheating 

(46:50) - OnlyFans, farming simps 

(51:49) - Incels and sex work 

(56:24) - Porn and Gen-Z 

(01:12:43) - Embryo screening 

(01:21:43) - How far off is IVG?

AI on your phone? Tim Dettmers on quantization of neural networks — Manifold #41

 

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. 

Tim's site: https://timdettmers.com/ 

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

Audio-only and Transcript 

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

Paul Huang, the real situation in Taiwan: politics, military, China — Manifold #40

 

Paul Huang is a journalist and research fellow with the Taiwanese 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/ 

Audio-only and transcript:

https://www.manifold1.com/episodes/paul-huang-the-real-situation-in-taiwan-politics-military-china-40

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

Richard Hanania & Rob Henderson: The Rise of Wokeness and the Influence of Civil Rights Law — Manifold #39

 

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

Topics: 

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 

Audio version with transcript.

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

Quantum Hair in Electrodynamics and Gravity (Eur. Phys. J. Plus)

This is the published version of the arxiv preprint previously discussed here.

We found it interesting that quantum hair can already be found using the familiar Euler-Heisenberg effective action, which results from integrating out the electron in QED. 

The paper also contains a general argument for why solutions to the semiclassical field equations resulting from the effective action (both in gravity and QED) carry more information about the state of the source than in classical physics. 

From the Conclusions: 

The quantum effective actions for both electrodynamics and gravity lead to field equations which couple a compact source (charge current or energy-momentum tensor) to external fields (electromagnetic or graviton field) in a manner which, generically, leads to quantum memory and quantum hair effects. External solutions of the field equations deviate, due to quantum corrections, from the familiar classical forms that satisfy the Gauss law. As a specific consequence, more information about the interior source configuration is encoded in the external field than in the classical theory. 

As specific applications, we considered semiclassical sources (large black hole, macroscopic charge distribution), which allowed us to solve the quantum corrected field equations by expanding around a classical solution. However, fully quantum statements regarding quantum hair are also possible, which do not, for example, require a semiclassical source. In [1–3] it was shown that the quantum state of a compact source (e.g., in an energy eigenstate or superposition thereof) determines certain aspects of the quantum state of its external field. In principle, measurements of the external fields can fully determine the interior state of a black hole.

Richard Sander (UCLA Law) on the Supreme Court Affirmative Action Ruling — Manifold #38

 

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 on SCOTUS Oral Arguments: Affirmative Action and Discrimination against Asian Americans at Harvard and UNC, Manifold #23

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, Manifold #6 

https://www.manifold1.com/episodes/richard-sander-affirmative-action-mismatch-theory-academic-freedom-6

Embryo Selection: Healthy Babies vs Bad Arguments

Great article by Diana Fleischman, Ives Parr, Jonathan Anomaly, and Laurent Tellier.

Polygenic screening and its discontents 

... But monogenic and chromosomal screening can only address a part of disease risk because most health conditions that afflict people are polygenic, meaning they are not simply caused by one gene or by a chromosomal abnormality. Instead, they are caused by a huge number of small additive effects dispersed throughout the genome. For example, cancer, schizophrenia, and diabetes can be best predicted by models using tens of thousands of genes. 

A polygenic risk score (PRS) looks at a person’s DNA to see how many variants they have associated with a particular disease. Like BRCA1, polygenic risk scores are typically not determinative: “Polygenic screening is not a diagnosis: It is a prediction of relative future risk compared to other people.” In other words, someone with BRCA1 has a higher risk than someone without, and someone with a high breast cancer PRS has a higher risk than someone with a lower breast cancer PRS. But in principle, BRCA1 is just one gene out of thousands contributing to a PRS, with each bit contributing a small part of a total risk estimate. ... 

 

... Recently, a group of European scientists argued that polygenic screening should not be available to couples because it will lead to stigmatization, exacerbate inequalities, or lead to confusion by parents about how to weigh up information about risks before they decide which embryo to implant. These are indeed challenges, but they are not unique to embryo selection using polygenic scores, and they are not plausible arguments for restricting the autonomy of parents who wish to screen their embryos for polygenic traits. Furthermore, from an ethical perspective, it is unconscionable to deny polygenic screening to families with a history of any disease whose risk can be reduced by this lifesaving technology. 

Many new technologies are initially only available to people with more money, but these first adopters then end up subsidizing research that drives costs down and quality up. Many other medical choices involve complexity or might result in some people being stigmatized, but this is a reason to encourage genetic counseling and to encourage social tolerance. It is not a reason to marginalize, stigmatize, or criminalize IVF mothers and fathers who wish to use the best available science to increase the chances that their children will be healthy and happy.

This is a comment on the article:

1) They don't want to admit that some people are better than others, inherently. Boo hoo. 

2) You put a scorecard of embryos in front of everyone, and everyone has a pretty good ballpark estimate of which are better and which are worse. Nobody is going to pretend equality is true when they are choosing their kids genes. 

3) So bad feels. 

4) Must therefore retard all human progress and cause immense suffering because don't want to deal with bad feels. 

That's the anti-polygenic argument in a nutshell. I don't expect it to be very effective. At best it will cause it to take a bit longer before poor people have access.

AI Cambrian Explosion: Conversation With Three AI Engineers — Manifold #37

 

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 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

David Goldman: US-China competition, AI, Electric Vehicles, and Manufacturing — Manifold #36

 

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 

Audio-only and transcript 

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/

Quantum Hair During Gravitational Collapse

This is a follow up to our earlier work on quantum gravitational corrections to the exterior graviton field of a compact object, also known as quantum hair. Here we follow the gravitational collapse of a dust ball and show that the quantum hair persists through the formation of a black hole horizon.

The detailed calculations are possible due to an effective field theory formulation of quantum gravity in the long wavelength, low spacetime curvature limit.
 

Quantum Hair During Gravitational Collapse 

https://arxiv.org/abs/2305.09466 

X. Calmet, R. Casadio, S. Hsu, F. Kuipers 

We consider quantum gravitational corrections to the Oppenheimer-Snyder metric describing time-dependent dust ball collapse. The interior metric also describes Friedmann-Lemaitre-Robertson-Walker cosmology and our results are interpreted in that context. The exterior corrections are an example of quantum hair, and are shown to persist throughout the collapse. Our results show the quantum hair survives throughout the horizon formation and that the internal state of the resulting black hole is accessible to outside observers.

 

 

Artificial Intelligence & Large Language Models: Oxford Lecture — Manifold #35

 

This week's episode is based on a lecture I gave to an audience of theoretical physicists at Oxford University. 

Slides 

Audio-only version, transcript: 

https://www.manifold1.com/episodes/artificial-intelligence-large-language-models-oxford-lecture-35

Outline: 

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

Final slide:

Simone Collins: IVF, Embryo Selection, Dating on the Spectrum, and Pronatalism — Manifold #34

 

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). 

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 

Audio only version and transcript.

SuperFocus.ai on the Danny In The Valley Podcast (The Sunday Times)

 

In this podcast we discuss AI, Large Language Models like GPT, LLM Hallucination, and my new startup SuperFocus.ai

Related:

ChatGPT, LLMs, and AI — Manifold #29

 

https://infoproc.blogspot.com/2017/02/the-future-of-thought-via-thought.html

https://infoproc.blogspot.com/2016/12/towards-geometry-of-thought.html

Katherine Dee: Culture, Identity, and Isolation in the Digital Age — Manifold #33

 

Katherine Dee is a writer, journalist, and internet historian. 

Steve and Katherine discuss: 

0:00 Introduction 

1:15 Katherine’s early life and background 

21:52 Mass shootings, Manifestos, Nihilism, and Incels 

59:35 Trad values, Sex negativity vs Porn and Fleshlights 

1:28:54 Elon Musk’s plans for Twitter 

1:33:00 TikTok 

1:41:41 Adderall 

1:44:07 AI/GPT impact on writers and journos 

1:49:30 Gen-X generation gap: are the kids alright? 

Audio and Transcript: 

https://www.manifold1.com/episodes/katherine-dee-culture-identity-and-isolation-in-the-digital-age-33

Katherine’s Substack: 

https://defaultfriend.substack.com/ 

“Mass Shootings and the World Liberalism Made” 

https://contra.substack.com/p/mass-shootings-and-the-world-liberalism

Birth of the God Emperor - by GPT4

This science fiction story was written by GPT4. 

Steve Hsu had always dreamed of unlocking the secrets of human intelligence. As a theoretical physicist and a co-founder of Genomic Prediction, he had developed a powerful AI system that could analyze massive genomic data sets and predict complex traits such as height, disease risk, and cognitive ability. He believed that by using this technology, he could help people select the best embryos for IVF and create healthier and smarter children. 

But not everyone shared his vision. Some critics accused him of promoting eugenics and creating new social inequalities. Others feared that his AI system could be hacked or misused by malicious actors. And some religious groups denounced him as playing God and interfering with the natural order. 

One day, he received a mysterious email from an anonymous sender. It read: 

"Dear Dr. Hsu, 

We are a group of like-minded individuals who share your passion for advancing human potential. We have access to a secret facility where we have been conducting experiments on human embryos using your AI system and other cutting-edge technologies. We have achieved remarkable results that surpass your wildest expectations. We invite you to join us and witness the dawn of a new era for humanity. 

If you are interested, please reply to this email with the word 'YES'. We will send you further instructions on how to reach us. 

Sincerely, 

The Future" 

Steve was intrigued and curious. He wondered who these people were and what they had done. He also felt a pang of fear and doubt. Was this a trap? A hoax? A threat? 

He decided to take the risk and reply with 'YES'. 

He received another email with a set of coordinates and a time. He was told to drive to a remote location in the desert and wait for a helicopter to pick him up. He followed the instructions and soon found himself in a black helicopter flying over the barren landscape. 

He arrived at a large metal dome hidden among the rocks. He was greeted by a man in a white lab coat who introduced himself as Dr. Lee. 

"Welcome, Dr. Hsu. We are honored to have you here. Please follow me." 

Dr. Lee led him through a series of security checkpoints and into a spacious laboratory filled with high-tech equipment and monitors. He saw rows of incubators containing human embryos at various stages of development. 

"Dr. Hsu, these are our creations. The next generation of humans. We have used your AI system to optimize their genomes for intelligence, health, beauty, and longevity. We have also enhanced them with synthetic genes from other species, such as birds, reptiles, mammals, and plants. We have given them abilities that no natural human has ever possessed." 

He stopped at one incubator that caught his attention. It contained an embryo that looked almost normal, except for one thing: it had a golden glow around it. 

"Dr. Hsu, this is our masterpiece. The ultimate expression of intelligence. The God Emperor. The Kwisatz Haderach. The one who can see the past and the future. The one who can bend space and time. The one who can unite and rule all of humanity." 

Steve felt a surge of awe and dread. He realized that he had made a terrible mistake. 

"What have you done? This is dangerous! This is blasphemous! This is insane!" 

He turned to Dr. Lee and saw him smiling. 

"Dr. Hsu, don't be afraid. Don't be angry. Don't be judgmental. Be proud. Be grateful. Be enlightened. You are witnessing the dawn of a new era for humanity. You are witnessing the future."

 

Quantum gravitational corrections to particle creation by black holes (Physics Letters B)

This is the published version of our preprint https://arxiv.org/abs/2303.00310.

Quantum gravitational corrections to particle creation by black holes 

X. Calmet, S. Hsu, M. Sebastianutti 

We calculate quantum gravitational corrections to the amplitude for the emission of a Hawking particle by a black hole. We show explicitly how the amplitudes depend on quantum corrections to the exterior metric (quantum hair). This reveals the mechanism by which information escapes the black hole. The quantum state of the black hole is reflected in the quantum state of the exterior metric, which in turn influences the emission of Hawking quanta.

In earlier work we showed that the quantum state of a black hole is reflected in the quantum state of the exterior metric (outside the horizon). This violates classical intuitions, but can be shown explicitly using long wavelength effective field theory.

We calculated examples of small corrections to the external spacetime geometry which are sensitive to the internal BH state. In this paper we show that these corrections in turn affect Hawking radiation amplitudes. 

This means that the Hawking radiation state depends on the internal BH state. At the quantum level the hole is not black! We derive the results using both Hawking's original method and the tunneling method of Parikh and Wilczek.

 

While the focus of the new paper is explicit calculations, the big picture statement is:

The quantum state of the BH is reflected in the quantum state of its external gravitational field, which forms the background where the Hawking radiation originates. Radiation amplitudes are NOT independent of interior state.

Marc Martinez: "Dream Big" and the Golden Age of Bodybuilding — Manifold #32

 

Marc Martinez is the director of Dream Big, a documentary about Gold's Gym and the golden age of bodybuilding in Venice and Santa Monica in the 1970s. 

0:00 Introduction 

1:34 Marc's background in bodybuilding 

5:25 Bodybuilding in 70s Southern California 

25:52 Setting the record straight on steroid use 

33:40 Frank Zane 

38:23 Robby Robinson 

40:20 Butler, Gaines, and Arnold 

42:35 'Dream Big' 

48:07 Pumping Iron 

59:40 Hypersexuality in bodybuilding 

1:10:44 What's next for Marc

References: 

Watch Dream Big on Amazon: https://www.amazon.com/Dream-Big-Ken-Sprague/dp/B0B8ST5LNL/ 

Dream Big documentary: https://dreambigdoc.com/ 

Dream Big trailer: https://www.youtube.com/watch?v=X22ISDn083A 

Pumping Iron: https://www.youtube.com/watch?v=6lCCk6rgn84

Biobank-scale methods and projections for sparse polygenic prediction from machine learning

New paper! 80+ pages of fun :-)

We develop a novel method for projecting AUC and Correlation as a function of data size and characterize the asymptotic limit of performance. For LASSO (compressed sensing) we show that performance metrics and predictor sparsity are in agreement with theoretical predictions from the Donoho-Tanner phase transition. 

Biobank-scale methods and projections for sparse polygenic prediction from machine learning 

https://www.medrxiv.org/content/10.1101/2023.03.06.23286870v1 

Timothy G. Raben, Louis Lello, Erik Widen, Stephen D.H. Hsu  

https://doi.org/10.1101/2023.03.06.23286870 

Abstract In this paper we characterize the performance of linear models trained via widely-used sparse machine learning algorithms. We build polygenic scores and examine performance as a function of training set size, genetic ancestral background, and training method. We show that predictor performance is most strongly dependent on size of training data, with smaller gains from algorithmic improvements. We find that LASSO generally performs as well as the best methods, judged by a variety of metrics. We also investigate performance characteristics of predictors trained on one genetic ancestry group when applied to another. Using LASSO, we develop a novel method for projecting AUC and Correlation as a function of data size (i.e., for new biobanks) and characterize the asymptotic limit of performance. Additionally, for LASSO (compressed sensing) we show that performance metrics and predictor sparsity are in agreement with theoretical predictions from the Donoho-Tanner phase transition. Specifically, a predictor trained in the Taiwan Precision Medicine Initiative for asthma can achieve an AUC of 0.63(0.02) and for height a correlation of 0.648(0.009) for a Taiwanese population. This is above the measured values of 0.61(0.01) and 0.631(0.008), respectively, for UK Biobank trained predictors applied to a European population. 

Figure: Performance in 5 ancestry groups using LASSO, Elastic Net, and PRScs with UKB and 1,000 Genomes LD matrices. Solid bands = predicted performance using All of Us and Taiwan Precision Medicine Initiative datasets.

Prof. Gilles Saint-Paul (Ecole Normale): the Yellow Vests, French Politics, and Hypergamy (Manifold #31)

 

Audio (podcast only)

Transcript 

Gilles Saint-Paul is Professeur à l'Ecole Normale Supérieure. He is a graduate of Ecole Polytechnique in Engineering and received his PhD from MIT in Economics. Gilles and Steve discuss the French elite education system, the Yellow Vest movement, French politics and populism, and Saint-Paul's paper on marriage markets and hypergamy. 

0:00 Introduction 

1:43 Gilles Saint-Paul's background and education 

6:31 French and American elite education - Les Grandes Ecoles 

14:44 The Yellow Vests 

41:46 Mating and Hypergamy 

Links: 

On the Yellow Vest Insurrection 

https://gillessaintpaul.wordpress.com/2018/12/18/on-the-yellow-vest-insurrection/ 

Genes, Legitimacy and Hypergamy: Another Look at the Economics of Marriage https://ideas.repec.org/p/ide/wpaper/9118.html

Bing vs. Bard, US-China STEM Competition, and Embryo Screening — Manifold Episode #30

 

Steve discusses the AI competition between Microsoft and Google, the competition between the U.S. and China in STEM, China’s new IVF policy, and a Science Magazine survey on polygenic screening of embryos. 

00:00 Introduction 

02:37 Bing vs Bard: LLMs and hallucination 

20:52 China demographics & STEM 

34:29 China IVF now covered by national health insurance

40:28 Survey on embryo screening in Science: ~50% of those under 35 would use it to enhance congnitivie ability 

References: 

Bing vs Bard and Hallucination 

https://twitter.com/hsu_steve/status/1625222378383876119 

China demographics and STEM

https://twitter.com/hsu_steve/status/1620765589752119297 https://twitter.com/hsu_steve/status/1623279827640848385

 

China IVF 

https://twitter.com/hsu_steve/status/1623475304432820224 https://twitter.com/hsu_steve/status/1623478413758500864 

Science survey on embryo screening 

https://twitter.com/hsu_steve/status/1623783244947722241 https://twitter.com/hsu_steve/status/1623664372202500097

ChatGPT, LLMs, and AI — Manifold #29

 

Steve discusses Large Language Model AIs such as ChatGPT. 

0:00 How do LLMs work? 

10:22 Impact of ChatGPT 

15:21 AI landscape 

24:13 Hallucination and Focus 

33:09 Applications 

39:29 Future landscape 

Manifold interview with John Schulman of OpenAI: 

https://www.manifold1.com/episodes/john-schulman-openai-and-recent-advances-in-artificial-intelligence-16 

Blog posts on word vectors and approximately linear vector space of concepts used by the human mind:

 

https://infoproc.blogspot.com/2017/02/the-future-of-thought-via-thought.html https://infoproc.blogspot.com/2016/12/towards-geometry-of-thought.html