The Grey Matter

The Grey Matter is a Substack that delves into a wide range of subjects, including the intricacies of voting systems, travel reflections, advancements in artificial intelligence, ethical dilemmas, innovation, and cultural analysis. It examines both current technological trends and historical perspectives, offering critical insights into societal impacts, policy, and the human condition.

Voting Systems Travel and Culture Artificial Intelligence Ethics and Morality Innovation and Regulation Societal Impacts Technology and Society Culture and Identity Environmental Issues Law and Society Science and Technology Economics and Industry

The hottest Substack posts of The Grey Matter

And their main takeaways

Can Morality Be Quantified?

19 implied HN points • 04 Jan 24
📖 Philosophy Morality Ethics Decision-making Values
  1. Quantification in morality is a useful framework for decision-making, not a literal truth.
  2. Money is a valuable tool for quantification, but it has limitations in capturing all aspects of value.
  3. Quantification, like valuing a statistical human life, helps systemize decision-making, but it has practical limits and should be used as a tool wisely.

Assorted Thoughts on Tunisia

39 implied HN points • 04 Oct 23
🏝 Travel Sightseeing Local culture Cuisine
  1. Carthage has ancient ruins with beautiful mosaics and a rich history of wars and civilizations.
  2. Public opinion in Tunisia shows a mix of views on democracy, jobs, corruption, and economic growth since the revolution.
  3. The visit to Tunisia highlighted aspects of lesser economic standing, decay in some areas, and challenges in tourism infrastructure.

Debunking Dunning-Kruger

19 implied HN points • 01 Aug 23
🔬 Science Psychology Research Data Analysis Simulation
  1. The Dunning-Kruger effect is likely a statistical artifact, not a genuine psychological phenomenon
  2. The popular interpretation of the Dunning-Kruger effect as 'the dumbest people think they're the smartest' is a distortion
  3. Replication of the Dunning-Kruger effect through simulation suggests it may not be a real psychological finding

The Repugnant Conclusion Sounds Rather Nice

19 implied HN points • 17 Jun 23
📖 Philosophy Ethics Population Welfare Conclusion
  1. The Repugnant Conclusion in population ethics presents a scenario where a larger population with slightly lower welfare can be deemed better than a smaller population with very high welfare.
  2. The comparison between different universes demonstrates how the preference shifts from smaller perfect welfare populations to larger populations with slightly lower welfare, leading to a continuously increasing total happiness.
  3. Ultimately, in exploring extreme scenarios, the Repugnant Conclusion challenges our perception with mind-boggling numbers and presents a world where an astronomical number of people would find joy in their existence, highlighting the complexity of ethical dilemmas.

Approval Voting

78 implied HN points • 24 Oct 21
  1. Approval voting allows voters to choose every candidate they approve of.
  2. Approval voting prevents election spoilers and ensures the best candidate wins.
  3. One drawback of approval voting is that voters may struggle with the non-intuitive nature and limited expression.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

PaLMistry

39 implied HN points • 06 Jul 22
  1. Recent advancements in large language models like GPT-3 and PaLM show significant progress, but lack a theoretical foundation.
  2. The performance of LLMs like PaLM can vary greatly based on their size, with larger models showing much better results.
  3. Approaching true AGI with LLMs may lead to transformative societal impacts, but it's hard to predict the true capabilities or the timeline of such advancements.

The Freak, Tortured

19 implied HN points • 25 Oct 22
  1. The development of AIs that can suffer would have moral implications.
  2. Current advanced AIs, like LLMs, show no signs of consciousness or self-awareness.
  3. Conscious AI would face challenges living in a physical world designed for slower beings.

Judgment Day

19 implied HN points • 26 Aug 22
  1. We are not ready for superintelligent AI
  2. AGI may surpass human intelligence and capabilities rapidly
  3. The morality of artificial superintelligence is uncertain and could differ drastically from human morality

GPT-3 Doesn’t Understand Me

19 implied HN points • 02 May 22
  1. GPT-3 can mimic understanding but doesn't actually comprehend like humans do.
  2. GPT-3's responses appear varied and inconsistent, showing a lack of true understanding.
  3. It's crucial not to rely on GPT-3 for accurate information or advice due to its inability to understand and its lack of human attributes like compassion.

Is Innovation Illegal?

19 implied HN points • 02 Sep 21
  1. Innovation can be restricted by government regulations in certain industries.
  2. Not all products require continuous innovation; some everyday items may not change much over time.
  3. The pace of innovation in certain fields may be hindered by technological limitations rather than government regulations.

You Say "AI" Like It Means Something

19 implied HN points • 09 Jun 21
  1. The term 'AI' is broad and can be poorly defined, leading to confusion.
  2. AI can range from simple tasks like linear regression to complex applications like self-driving cars.
  3. Machine learning, a dominant subfield of AI, involves algorithms learning directly from data.

A Change

3 HN points • 14 Dec 22
  1. Personal change and growth can happen unexpectedly, such as through experiences like the COVID-19 pandemic.
  2. Questioning institutional reliability and seeking information from diverse sources is crucial for a well-rounded perspective.
  3. While acknowledging the importance of institutions, it is also important to be critical and hold them accountable for their actions.

Is H. L. Mencken Watching Me Undress?

19 implied HN points • 26 Jul 20
  1. For every complex problem, there is a clear, simple, and wrong answer.
  2. Well-known solutions to human problems are often neat, plausible, and wrong.
  3. Practical politics thrive on keeping the populace alarmed with imaginary threats.

Letter to a TSA Agent

0 implied HN points • 03 Oct 19
  1. The letter reflects on a TSA agent taking away peanut butter, questioning the impact on safety.
  2. The author wonders if the TSA agent questioned their role and purpose.
  3. The letter poses deeper questions about the motives behind security measures.

The Intersection that Broke Humanity

0 implied HN points • 05 Jul 19
  1. Bad designs can lead to negative human experiences.
  2. Small tweaks in design can have significant positive impacts.
  3. We have the power to design systems that bring out the best in people.

The Drug Dealers

0 implied HN points • 01 May 19
  1. The media being funded by drug companies can skew news coverage and downplay important issues.
  2. Prescription opioids like OxyContin were aggressively marketed, leading to addiction and overdose deaths.
  3. The drug fentanyl is now causing a new wave of overdose deaths, highlighting the influence of drug companies on news coverage.

Sexy Problems

0 implied HN points • 04 Apr 19
  1. Sexy problems are fun to talk about but may not be the most important issues.
  2. Deep fakes and the trolley problem are examples of sexy problems.
  3. We should prioritize attention on the most crucial problems, whether they are sexy or not.

The Agency Problem

0 implied HN points • 24 Mar 19
  1. AI lacks a sense of agency and consciousness, inhibiting progress towards human-like AI.
  2. Motivation is a key missing component in AI, as current methods do not truly instill desires or will.
  3. True understanding, beyond language and image processing, is a fundamental challenge that requires new paradigms in AI development.

Animals Alone in their Families

0 implied HN points • 21 Jun 17
  1. Some animals are the only extant members of their families.
  2. Certain animals are the only remaining members of their suborder.
  3. There are animals that stand alone as the only extant members of their order.

New Words in 5th Edition of the American Heritage Dictionary

0 implied HN points • 29 Apr 17
  1. New words provide insight into national consciousness and culture.
  2. New words can be scientific or technological concepts, historical items, or popular terms.
  3. The 5th Edition of the American Heritage Dictionary introduced various new terms such as 'brain freeze', 'carbon footprint', and 'fashionista'.

Invasive Species

0 implied HN points • 22 May 17
  1. Invasive species can cause significant harm to ecosystems and native species.
  2. Natural selection may not prepare species fast enough for new invasive predators.
  3. Humans often unintentionally introduce invasive species, causing detrimental effects.

Serving Our AI Masters

0 implied HN points • 05 Mar 17
  1. AI optimization is changing how content is created, focusing on pleasing machines over humans.
  2. SEO tactics on websites are often tailored to attract algorithms rather than human readers.
  3. Financial reports are being altered to appeal to sentiment analysis algorithms, prioritizing AI understanding over human comprehension.

Balancing the Scales of Justice

0 implied HN points • 18 Nov 23
🔬 Science
  1. The study discusses the importance of considering tradeoffs when incorporating medically irrelevant information in forensic analysis.
  2. Bayesian statistics can help improve decision-making in forensic analysis by considering uncertainties and probabilities.
  3. By integrating background information with medical examiner data through Bayesian reasoning, one can enhance the accuracy of identifying guilt or innocence in forensic cases.

The Climate Narrative

0 implied HN points • 20 Feb 23
🌞 Climate & Environment Climate change Global impact Environmental Activism
  1. Climate news often emphasizes negative impacts, ignoring potential benefits like improved crop growth from increased CO2 levels.
  2. The media tends to focus on portraying climate change as always getting worse, disregarding instances where it may have positive effects or remain stable.
  3. Scientific studies and media reports may frame climate change in a misleading manner, emphasizing alarming perspectives to provoke action.

Natural Neural Networks Are Full of Shit

0 implied HN points • 27 Jan 23
  1. Artificial neural networks are not actually black boxes; they are transparent and their inner workings can be scrutinized.
  2. Humans tend to provide explanations that feel solid but often have gaps when probed deeper, similar to how natural neural networks operate.
  3. Natural neural networks are full of rationalization and can be unreliable; this phenomenon is also seen in psychology research and decision-making.

An Overview of AI Explainability

0 implied HN points • 14 Jan 23
  1. AI explainability is a field dedicated to making AI systems understandable to humans
  2. Early AI techniques relied on clear rules, while modern machine learning methods can be more complex
  3. SHAP values are useful for explaining tabular data in AI models, and saliency maps are helpful for understanding computer vision models

Hurricanes and Climate Change

0 implied HN points • 28 Aug 22
  1. Hurricanes don't show clear changes due to human influences, based on research.
  2. Precipitation from hurricanes has increased due to human-caused climate change, impacting damage and deaths.
  3. Future projections suggest a modest increase in intensity and precipitation, but not catastrophic changes in hurricanes.

That Damn 97% Number

0 implied HN points • 05 Mar 22
  1. The 97% number refers to scientists endorsing the consensus that humans are causing global warming.
  2. Different interpretations of the 97% statistic exist based on political leanings.
  3. Debates about the validity of the 97% statistic continue, even though it suggests strong scientific agreement.

Assorted Thoughts on Don't Look Up

0 implied HN points • 10 Jan 22
  1. The movie could have made the comet impact probability lower to show real-world scenarios
  2. The film addressed existential threats like climate change, pandemics, and AI risk
  3. The Starship coda was criticized for being unrealistic in a movie about incompetence and corruption

Why I'm Not a Socialist

0 implied HN points • 02 Jan 22
  1. Locks on the toilet paper rolls highlight the tragedy of the commons.
  2. Build systems for people as they are, not as we want them to be.
  3. Culture, education, and institutions can be changed, but DNA cannot.

Does Anyone Take Utilitarianism Seriously?

0 implied HN points • 29 Nov 21
  1. Utilitarianism involves doing the greatest good for the greatest number of people.
  2. Medical knowledge could be gained from unethical studies, but ethics prevent their implementation.
  3. Although utilitarianism is an intriguing idea, human behavior often contradicts its principles.

Closed Problem Spaces

0 implied HN points • 31 Oct 21
  1. Closed problem spaces prevent thorough evaluation of solutions.
  2. Debates on problem and solution spaces should be kept separate.
  3. Closing off problem spaces can lead to missed solutions and futile searches for answers.

The Case Against Ranked-Choice Voting: Part II

0 implied HN points • 17 Oct 21
  1. Ranked-Choice Voting encourages strategic voting
  2. Ranked-Choice Voting doesn't always result in the best head-to-head candidate
  3. Ranked-Choice Voting allows for election spoilers and unexpected outcomes

The Case Against Ranked-Choice Voting: Part I

0 implied HN points • 17 Oct 21
  1. Ranked-choice voting can lead to strategic voting where voters choose different candidates to get a preferred outcome
  2. In ranked-choice voting, the candidate who wins may not be the best head-to-head choice against other candidates
  3. Ranked-choice voting can result in election spoilers and situations where candidates with higher rankings can end up with worse outcomes

Misleading Figure in 2017 Climate Science Special Report

0 implied HN points • 14 Oct 21
  1. The claims about extreme temperatures in the 2017 Climate Science Special Report may be misleading due to the way they are presented.
  2. Using simulations to understand the impact of temperature reporting methods can reveal potential pitfalls in data presentation.
  3. The analysis shows how certain graphing techniques can create the appearance of trends that may not exist in the actual data.

Interview with GPT-3 On Modern Architecture: Part II

0 implied HN points • 10 Oct 21
  1. The shift to minimal and monochromatic clothing in the 20th century was related to modern architecture and the rise of mass media.
  2. There was a movement towards minimalism in various art forms as a reaction to the ornate Victorian era.
  3. It's important to be mindful of the subject matter when writing poetry, considering the context of global suffering.

The Nuclear Industry is Overregulated

0 implied HN points • 18 Aug 21
  1. The Nuclear Regulatory Commission limits radiation exposure far below safe levels, hindering the industry's progress.
  2. Research shows that nuclear energy risks may be no greater than those from common environmental and lifestyle factors.
  3. Regulations on nuclear energy are influenced by the status quo bias and can impact the industry's growth.

A "Nudge" at the Airport

0 implied HN points • 13 Aug 21
  1. People at airports need nudges to behave better.
  2. Simple design elements like differently colored tiles can guide behavior effectively.
  3. Small changes can make a big impact on people's actions.

Do Not Anthropomorphize AI

0 implied HN points • 02 Jul 21
  1. Don't anthropomorphize AI - it won't be humanlike.
  2. AI development is not about mimicking human progress.
  3. Advanced AI will differ significantly from humans due to consciousness, sentience, and biological limitations.

What Important Truth Do Very Few People Agree With You On?

0 implied HN points • 21 Jun 21
  1. Technological progress follows a natural boom and bust cycle independent of policies.
  2. Discovery in the hard sciences like chemistry and physics affecting human life may have plateaued.
  3. There are more political moderates than commonly perceived, even though they may not be vocal on platforms like Twitter.