Show HN: Hacker News em dash user leaderboard pre-ChatGPT

The top 50 users on Hacker News who posted with em dashes before November 30, 2022, have been ranked, with derefr taking the top spot with 4,247 posts containing em dashes. The leaderboard shows the number of em dash posts, as well as the first and last post dates, for each of the top 50 users, providing insight into their posting habits on the platform.

Read more

AI models need a virtual machine

The development of AI models has led to the need for a standardized virtual machine that can embed these models into software systems, providing qualities such as security, isolation, and portability. A proposed AI Model Virtual Machine (MVM) would decouple model development from integration logic, allowing any model to "plug in" to a rich software ecosystem, similar to the Java Virtual Machine, and enable a "write once, run anywhere" execution environment for AI models while maintaining security and privacy.

Read more

Pentagon Docs: US Wants to "Suppress Dissenting Arguments" Using AI Propaganda

The US Special Operations Command is seeking to develop and use machine learning technology to create and distribute propaganda overseas, aiming to influence foreign audiences and suppress dissenting arguments. The technology, which can operate with minimal human oversight, would enable the US to control narratives and influence audiences in real-time, raising concerns about the potential for autonomous propaganda and the blurring of lines between foreign and domestic targeting.

Read more

SynthID – A tool to watermark and identify content generated through AI

Google DeepMind is a leading AI research organization that offers a range of AI models, including Gemini and Gemma, and applies AI to various fields such as biology, climate, mathematics, and physics. The organization's mission is to build AI responsibly to benefit humanity, and it provides resources for building with AI, researching AI, and exploring AI applications in science.

Read more

ETFs now hold more than $3.1T worth of just top US companies

The top 100 equity holdings of US-listed ETFs have a total market value of $3.1 trillion, with the top 10 holdings accounting for a significant portion of this value, led by NVIDIA Corp at $612.0 billion. The list of top holdings includes a diverse range of companies across various sectors, including technology, finance, healthcare, and consumer goods, with other notable holdings including Microsoft Corp, Apple Inc, and Amazon.com Inc.

Read more

A convex polyhedron without Rupert's property

A three-dimensional convex body has Rupert's property if a copy of it can fit through a hole within itself. Researchers have constructed a non-Rupert polyhedron, disproving a 2017 conjecture, and also found a polyhedron that is Rupert but not locally Rupert, contributing to the understanding of this geometric property.

ArXiv

Quantum Bayes' rule and Petz transpose map from the minimum change principle

Bayes' rule can be derived from a principle of minimum change, which updates beliefs based on new evidence while deviating minimally from the prior belief. A quantum analog of this principle has been introduced, leading to a quantum Bayes' rule that minimizes the change between two quantum input-output processes, and in some cases, recovers the Petz transpose map when maximizing fidelity.

ArXiv

Cognitive load is what matters

Cognitive load refers to the amount of mental effort required to complete a task, and in software development, high cognitive load can lead to confusion, wasted time, and money. Reducing cognitive load is crucial, and this can be achieved by simplifying code, avoiding unnecessary complexity, and using techniques such as introducing intermediate variables, early returns, and composition over inheritance.

Repo

Show HN: Sourcerer – MCP for semantic code search that reduces token waste

Sourcerer MCP is a server for semantic code search and navigation that enables AI agents to efficiently find relevant code without reading entire files, reducing token usage and cognitive load. It builds a semantic search index of a codebase using OpenAI's API and Tree-sitter, allowing agents to search conceptually and jump directly to specific functions, classes, and code chunks.

Repo

Moondream – a tiny vision language model that runs anywhere

Moondream is a highly efficient, open-source vision language model that can run on a wide range of devices and platforms, offering powerful image understanding capabilities with a small footprint. It comes in two variants, Moondream 2B and Moondream 0.5B, which can be used for tasks such as captioning, visual question answering, and object detection, and can be run locally or in the cloud.

Repo

Show HN: Keeptalking

keeptalking is a simple, pythonic interface to any OpenAI-compatible LLM server, providing a straightforward way to interact with language models through a minimalistic API of just three functions: talk, write, and vibe. The library allows for easy conversation, asynchronous batch requests, and definition of functions in natural language, with features like automatic self-throttling, parallelization, and structured outputs.

Repo

Boo-urns to all the AI widgets (declank Chrome extension, adblock for AI)

The declank Chrome extension (with Firefox support to be determined) allows users to highlight and hide specific phrases or elements on web pages based on a configurable block list. To develop and use the extension, users must install Node.js, clone the repository, install dependencies, and build the project, then load the extension in Chrome by enabling developer mode and selecting the build directory.

Repo