The hottest Research Substack posts right now

And their main takeaways
Category
Top Health Politics Topics

Cicero in Context: Gameplaying AIs and the Future of AI-enhanced Wargaming [guest post]

Baptiste’s Substack • 0 implied HN points • 08 Aug 23
🕹 Technology Research
  1. Gameplaying AIs vary in design and capabilities for different games like Chess, Go, Poker, and Diplomacy.
  2. Understanding the complexity of AI applications in wargaming requires knowledge beyond just machine learning.
  3. AI systems like Cicero blur the line between inhuman design and human-like behavior, paving the way for innovative approaches in the future.

Science under the bell jar

Perambulations • 0 implied HN points • 16 Feb 23
🔬 Science Research
  1. The desire for purity and resistance to change are deeply human traits, influencing our evolution.
  2. Scientific journals are questioning the originality, accountability, and transparency of AI in research.
  3. Policies regarding AI use in research should be focused on scientific progress, not maintaining journal prestige.

Three thoughts on turning policy calls for ‘interpretable’ AI systems into action

buffering... • 0 implied HN points • 09 Aug 23
🕹 Technology Research
  1. The algorithms in deep learning systems are mostly unknown, making it challenging to assess their learning process and how they generate output.
  2. Firms like Anthropic are investing in making AI algorithms more interpretable, but more support is needed.
  3. To promote the development of interpretable AI systems, measures like grants, collaboration across disciplines, and improving existing techniques are crucial.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

My year 2023 in review

Martin Kleppmann’s Blog • 0 implied HN points • 04 Jan 24
🕹 Technology Research
  1. The author had an eventful year in 2023, with milestones like having a baby and starting a new job at Cambridge.
  2. Successfully collaborated with Automerge and Bluesky.
  3. Made progress on the second edition of a book and published new research papers.

Thursday Curiosity Links

Grist Potentia • 0 implied HN points • 08 Feb 24
🔬 Science Research
  1. Joseph Schumpeter popularized the term 'creative destruction' and emphasized the role of entrepreneurship in capitalism.
  2. Researchers are studying the ancient partnership between moss and fungi.
  3. There is ongoing discussion about whether science is considered a public good.

🥟 Chao-Down #266 Microsoft secures a Super Bowl ad for Copilot, Google unveils the Gemini Era with Gemini Advanced, OpenAI makes moves towards building AI agents that will automate tasks

Chaos Theory • 0 implied HN points • 08 Feb 24
🕹 Technology Research
  1. Microsoft secures Super Bowl ad for Copilot, emphasizing productivity and creativity.
  2. Google introduces Gemini Advanced, Meta deploying custom chips for AI, OpenAI automates tasks with AI agents.
  3. AI developments include ChatGPT tagging images, Waymo's car incident, and Ikea's design assistant attempts.

AI Deployed Nukes During A War Simulation

Global Community Weekly (GloCom) • 0 implied HN points • 11 Feb 24
🕹 Technology Research
  1. AI used in war simulations showed a tendency to escalate conflicts and even deploy nuclear weapons.
  2. Different AI models have varied approaches to decision-making in war scenarios, some more peaceful and predictable than others.
  3. The study indicates that as AI becomes more integrated into military operations, there may be quicker escalation of conflicts.

Book Review: "Think Again: The Power of Knowing What You Don't Know" by Adam Grant 4/5

Poczwardowski Notes • 0 implied HN points • 09 Oct 21
🚌 Education Research
  1. Rethinking what we've learned is a valuable skill, starting with ourselves and extending to influencing others and fostering collective rethinking within organizations.
  2. Changing our minds goes beyond trivial debates, impacting our careers and even saving lives. The book blends research with engaging stories and humor.
  3. It's okay to change your perspective; it shows growth and learning. Embrace moments of being wrong as opportunities for progress.

Links #3

Splitting Infinity • 0 implied HN points • 07 Sep 23
🔬 Science Research
  1. The use of genetically modified neurons to improve MRI imaging of the brain by producing protein-based contrast agents is an intriguing idea.
  2. Real hedge funds do not seem to use certain advanced algorithms for portfolio selection despite proven performance improvement.
  3. FPGA's are versatile hardware that can be programmed for various computational tasks and have applications in fields like antennas, random number generation, and hardware security.

Links #1

Splitting Infinity • 0 implied HN points • 14 Jul 23
🔬 Science Research
  1. Research is exploring deriving spacetime from simple quantum models, potentially changing how we understand gravity and space.
  2. Innovations like ultrasound for gene therapy and brain imaging are showing promise in medical advancements.
  3. Focusing on specialized models for deep learning tasks could lead to safer and more efficient AI development.

🤯 How much should I be investing in brand?

the case for brand 💼 • 0 implied HN points • 01 Mar 24
💼 Business Research
  1. Brand marketing and growth marketing are not mutually exclusive - you need both for long-term success.
  2. The optimal split for investing in brand versus demand (activation) is typically 60% brand and 40% demand.
  3. Overinvesting in lead generation is a common mistake - balancing brand building and demand generation can lead to exceptional results.

AI and the art of doing with little-Part 1

Experiments with NLP and GPT-3 • 0 implied HN points • 05 Mar 24
🕹 Technology Research
  1. In the AI field, access to large amounts of compute power and data is crucial, but it can be expensive and a barrier for many. This can lead to a reliance on funding and resources, putting a focus on money as a determinant of success.
  2. The author emphasizes the importance of simpler, more accessible experiments in AI research, drawing inspiration from V.S. Ramachandran's approach in neuroscience. Small, innovative solutions may offer promising alternatives to standard big science methods.
  3. There is a push for exploring new ways to tackle AI challenges beyond the current reliance on GPUs and deep learning models. The idea of creating open-source datasets and involving young talents from India in research signifies a shift towards more inclusive and collaborative approaches.

Equal Ventures Hiring MBA Interns For the 2020–21 School Year

Equal Ventures • 0 implied HN points • 24 Aug 20
💼 Business Research
  1. Equal Ventures is hiring MBA interns for the 2020-21 school year from October to April with the possibility of extension.
  2. Interns will assist in researching industry themes and analyzing investment opportunities for the firm's thesis-driven approach.
  3. Applicants should have experience in consulting, research, or private equity, be intellectually curious, professional, self-directed, and amicable.

OLMo: Accelerating the Science of Language Models

Gonzo ML • 0 implied HN points • 10 Mar 24
🔬 Science Research
  1. OLMo is an open language model created by Allen AI, differentiating itself by being completely open-source including logs, checkpoints, and evaluation scripts under the Apache 2.0 License.
  2. OLMo comprises three models: 1B, 7B, and 65B, demonstrating improvements in classic transformer decoders similar to GPT, such as specific tokenization for PII and non-parametric layer normalization.
  3. OLMo was trained on data from their own dataset Dolma with plans to expand beyond English, showcasing their training process with PyTorch FSDP and evaluation using their benchmark Paloma and the Catwalk framework.

Benchmark datasets for learning-to-rank

Simplicity is SOTA • 0 implied HN points • 11 Mar 24
🕹 Technology Research
  1. Benchmark datasets are crucial in ML literature, providing a standard for evaluating new methods and influencing research directions.
  2. In learning-to-rank, the Yahoo and Microsoft datasets are prominent, with Yahoo dataset being widely used in notable papers.
  3. When writing a paper using benchmark datasets, researchers must choose ML algorithms, consider user behavior, generate initial rankings, and evaluate performance with metrics like NDCG.

Open Science Hardware at GOSH 2018

Shubhi’s Substack • 0 implied HN points • 17 Nov 18
🔬 Science Research
  1. Open Science Hardware focuses on creating accessible tools for research and experimentation, promoting collaboration and knowledge sharing.
  2. Gathering for Open Science Hardware (GOSH) events bring together diverse professionals to discuss open hardware issues and solutions, aiming for global impact.
  3. GOSH 2018 resulted in action items like organizing regional open hardware movements, creating maps of makerspaces, and offering programs with local universities, all to support the open science community.

Self-Forgiveness

effectiviology • 0 implied HN points • 08 Feb 24
🏥 Health & Wellness Research
  1. Forgiving yourself for past mistakes can reduce negative feelings like guilt and shame.
  2. Self-forgiveness can help you stop problematic behaviors by accepting responsibility for your actions and avoiding rumination and self-loathing.
  3. Proper self-forgiveness involves taking full responsibility for your actions, acknowledging any harm caused, and focusing on avoiding similar mistakes in the future.

GRANTED: How we really learn and why we should all be more like Goldilocks

Granted • 0 implied HN points • 04 Jun 17
🚌 Education Research
  1. Beware of self-help books focusing on power and networking as they might create a negative self-fulfilling prophecy.
  2. Learning from behaviors that go against your values can be valuable in understanding and defending against manipulation.
  3. Engaging multiple senses in learning is more effective than relying on a single 'learning style', according to scientists.

The White House does "something" on AI

Joshua Gans' Newsletter • 0 implied HN points • 22 Jul 23
🕹 Technology Research
  1. The White House met with seven AI companies and they agreed to eight commitments on AI, such as internal and external security testing.
  2. Some of the commitments seem like things the companies are already doing, while others may not have a significant impact.
  3. There were some notable absences, like Apple, Tesla, and NVIDIA, from the meeting with the White House, which raises questions about the inclusivity and seriousness of the discussions.

What Covid can teach us about entrepreneurship?

Joshua Gans' Newsletter • 0 implied HN points • 25 May 21
💼 Business Research
  1. Covid-19 led to shifts in entrepreneurial opportunities, with some ventures thriving (like online businesses) while others struggled (like physical stores).
  2. The pandemic allowed for some individuals to reevaluate their career paths and possibly pursue entrepreneurial endeavors, especially due to the pause it enforced and the time it provided for reflection.
  3. Research indicates that discrimination in lending markets may have made it difficult for certain neighborhoods, particularly those with higher proportions of black residents, to secure loans for new business ventures.

Did exposure apps work?

Joshua Gans' Newsletter • 0 implied HN points • 14 May 21
🔬 Science Research
  1. Exposure notification apps have potential but were not very effective during the crisis due to lack of take-up and false positives.
  2. A study on the NHS app showed a 6% secondary attack rate for app-notified individuals who subsequently tested positive, similar to manually traced close contacts.
  3. The study highlighted the need for further improvements in exposure notification apps, such as tailoring notifications, specifying locations, and adjusting for the type of contacts for better efficiency.

Nudges, Identity and Covid-19

Joshua Gans' Newsletter • 0 implied HN points • 06 May 21
🏥 Health & Wellness Research
  1. An impressive study in India evaluated interventions to increase vaccination rates for children, showcasing the effectiveness of reminders, incentives, and influential individuals in promoting immunizations.
  2. Studying data and identities during pandemics is crucial for tracking and monitoring testing patterns to ensure workplace safety and address vaccine hesitancy.
  3. Conducting research to understand the impact of pausing vaccine administration on short and long-term vaccine hesitancy is essential for informing public health policies.

Science and the Policy Response

Joshua Gans' Newsletter • 0 implied HN points • 13 Jan 21
🔬 Science Research
  1. Science heavily influenced policy-making during the Covid-19 pandemic, tracking the number of cases and playing a significant role in the global response.
  2. Recent research on Covid-19 had a substantial impact on policy-making and was highly cited by both policy-makers and other scientists.
  3. International governmental organizations like the World Health Organization (WHO) played a key role in connecting policy to science, producing more impactful documents compared to national governments or think tanks.

Will work from home stick?

Joshua Gans' Newsletter • 0 implied HN points • 11 Nov 20
💼 Business Research
  1. About 20% of workdays are expected to be done from home after the pandemic, compared to only 5% before.
  2. The time saved from commuting is not entirely wasted, as it can benefit both employees and employers.
  3. Stigma around working from home is decreasing, which can lead to better outcomes for women in the workplace.

Test Week: Let's talk about Sewage

Joshua Gans' Newsletter • 0 implied HN points • 21 Aug 20
🔬 Science Research
  1. Testing sewage for the novel coronavirus can help in early detection of outbreaks before they spread widely.
  2. Analyzing sewage can provide valuable information about the presence of infectious diseases in a population, and monitoring waste patterns could lead to new public health insights.
  3. Challenges in using sewage testing for surveillance include factors like rainwater affecting the virus presence, variations in viral material survival, and the need for careful data interpretation.

Economists found something surprising and you wont believe what happened next

Joshua Gans' Newsletter • 0 implied HN points • 15 Apr 17
🔬 Science Research
  1. Uncertainty can actually increase cooperation in certain situations.
  2. To address breakdowns in cooperation in scientific studies, integration mechanisms like co-authorship can increase credibility and reliability of results.
  3. In the face of surprising experimental findings, proposing new mechanisms and experiments can lead to advancements in scientific knowledge.

Why are peer reviews private?

Joshua Gans' Newsletter • 0 implied HN points • 01 Mar 17
🔬 Science Research
  1. The current system of keeping journal article reviews private raises questions about potential benefits of making them public to improve efficiency and encourage more careful submissions.
  2. Authors might take more care in their papers if peer reviews were public, allowing for dialogue and counter arguments.
  3. While there are concerns about anonymity and potential pressure on reviewers, testing out the publication of reviews could help reduce costs and system delays in academic publishing.

Ken Arrow: The Greatest

Joshua Gans' Newsletter • 0 implied HN points • 22 Feb 17
🚌 Education Research
  1. Ken Arrow made significant contributions to economics and won the Nobel Prize in 1972, being one of the youngest recipients.
  2. Ken Arrow was known for his exceptional intelligence and quick thinking, leaving a lasting impact on economic science.
  3. Despite his brilliance, Ken Arrow remained humble, approachable, and had a humorous side, making him a unique and revered figure in the field.

Lepores missing cites

Joshua Gans' Newsletter • 0 implied HN points • 23 May 16
💼 Business Research
  1. The author's book on disruption was not anchored with Jill Lepore's criticisms because many others had already tackled those points previously.
  2. The author aimed to rise above issues of personality and focus on ideas and evidence in the book rather than getting involved in other matters like personalities.
  3. The author brought together various academic literature on disruption to create the book.

What does a Skunk Works do?

Joshua Gans' Newsletter • 0 implied HN points • 20 May 16
💼 Business Research
  1. Skunk works are organizational spaces for innovative projects away from the main operations, often initiated by established firms to explore radical ideas.
  2. Skunk works come in different models, like focusing on science and engineering, building research communities, acting as lead users, and aiming for audacious goals like a moon shot.
  3. Despite their innovative success, skunk works face challenges integrating their achievements into regular operations, requiring careful planning and management to ensure long-term success.

A Taste of Disruption

Joshua Gans' Newsletter • 0 implied HN points • 24 Feb 16
💼 Business Research
  1. Firms facing demand-side disruption have options to react and defend against the threat.
  2. Supply-side disruption requires firms to be more integrative in how they conduct their business.
  3. The articles mentioned provide a 'lite' treatment of the disruption dilemma for those seeking a quick overview.