🧀 BigCheese.ai

Social

AI News & Launches

Stay ahead of the curve with BigCheese Weekly, your source for the Latest AI news and product debuts.

Subscribe
6 months ago
Promptordle

Promptordle is an online puzzle game inspired by the popular word game Wordle. It provides daily challenges where players guess a hidden word within a limited number of attempts.

6 months ago
Open-Source 4WD AI Robot Kit Compatible with Raspberry Pi Models 4 and 5

The RaspRover, an open-source 4WD AI robot kit by Waveshare, supports Raspberry Pi 4B and 5. It includes an all-aluminum chassis, dual-controller system with ESP32 MCUs, camera-serial interface, and various operational controls for developers and tech enthusiasts.

6 months ago
AI and the Problem of Knowledge Collapse

A paper titled 'AI and the Problem of Knowledge Collapse' by Andrew J. Peterson discusses the potential negative impacts of AI on public understanding and innovation. While AI can process large amounts of data, an over-reliance on AI systems may lead to 'knowledge collapse'. The paper models scenarios where communities depend on AI-assisted processes versus traditional methods, showing that a discount on AI content can lead to beliefs further from the truth.

6 months ago
Qwen1.5-32B: Fitting the Capstone of the Qwen1.5 Language Model Series

The Qwen Team releases a new model, Qwen1.5-32B, fitting into the language model series with approximately 30 billion parameters, offering balance between performance, efficiency, and memory consumption. Outperforming competitors in various tasks, Qwen1.5-32B presents a more resource-friendly alternative to 72B parameter models, maintaining high capabilities in diverse languages and complex tasks.

6 months ago
A Lego Model of Financial Capitalism

Brett Scott analyzes financial capitalism using a Lego model, showcasing the mechanisms of equity and debt financing in property development. The model breaks down complex financial concepts into understandable analogies, explaining capital raising, leverage, and investment outcomes through the metaphor of constructing a building using investor funds. The article also touches on the potential risks and rewards of financial leverage.

6 months ago
Microsoft, Google join study to see if AI will affect jobs

Big Tech companies, including Microsoft and Google, have formed a consortium led by Cisco to study the impact of artificial intelligence on tech jobs. The group will examine how AI may change work for different technology roles and provide actionable insights. Amidst AI advancements, concerns about AI-induced job displacement are being taken more seriously, with debates heating up regarding legislative action.

6 months ago
Google Books Is Indexing AI-Generated Garbage

Google Books indexes AI-generated content leading to concerns over its credibility and impact on research tools like Google Ngram viewer. Investigations uncovered repetitive content with outdated information due to delayed knowledge updates.

6 months ago
Dippy Widget - AI Character Chat

Dippy is an AI-powered application that offers users an opportunity to chat with AI friends right on their home screen. It provides a space where users can find a virtual friend, therapist, or partner according to their preference.

6 months ago
NUCA Camera

NUCA is an AI-powered camera and speculative art project by Mathias Vef and Benedikt Groß. It captures nude portraits by combining AI with real-time input data, emphasizing consent and societal implications.

6 months ago
Pizi

Pizi AI is an innovative tool designed for chat interactions with multiple PDFs. It summarizes lengthy documents, translates texts, and solves math problems efficiently. Built with privacy in mind, Pizi AI works on various file formats and languages offering fast, encrypted, and performance-oriented services. Key features include advanced layout comprehension, formula recognition, and OCR technology supporting over 100 languages.

6 months ago
Airdrop Tracker

OneClick Defi is a platform offering users a comprehensive list of DeFi airdrops, enabling participants to explore and join various cryptocurrency airdrops with ease.

6 months ago
The Crescendo Multi-Turn LLM Jailbreak Attack

Crescendo is a new multi-turn jailbreak attack technique that demonstrates high success in steering conversations towards prohibited topics in state-of-the-art language models. Designed to be user-friendly and requiring no technical knowledge of the LLM's workings, Crescendo exploits the model's pattern-following tendency, particularly when referencing its own generated content. The technique poses significant risks to language model ecosystems and highlights the need for improved security measures.

6 months ago
Fake AI law firms are sending fake DMCA threats to generate fake SEO gains

A journalistic investigation has unveiled a scheme where fake law firms using AI-generated profiles are sending out fraudulent DMCA notices to website owners. Their aim is to create backlinks and illegitimately boost SEO rankings. The pseudo firm 'Commonwealth Legal', targeting a photo from Unsplash, is one such example, and the incident raises questions about the effectiveness of search engines to screen for the quality of backlinks.

6 months ago
How should we define 'open' AI?

The term 'open AI' lacks a universal definition, creating confusion between marketing jargon and technical descriptors. The article delves into what constitutes 'open' AI, touching upon sustainability, transparency, accountability, and the impact of AI on knowledge work.

6 months ago
Rule-Based NLP System Beats LLM for Analysis of Psychiatric Clinical Notes

The study titled 'Extracting Social Support and Social Isolation Information from Clinical Psychiatry Notes: Comparing a Rule-based NLP System and a Large Language Model' compares the efficacy of a Rule-based System (RBS) and a Large Language Model (LLM) in processing psychiatric notes. The RBS showed superior macro-averaged f-scores in identifying social support and social isolation, outperforming LLM in both Mount Sinai Health System and Weill Cornell Medicine data sets.

6 months ago
Reaching LLaMA2 Performance with 0.1M Dollars

JetMoE-8B, achieving better performance than LLaMA2-7B and comparable models, has been trained on a significantly lower budget of less than $0.1 million. It's designed to be open, using public datasets, and is optimized for lower computational costs with only 2.2B active parameters during inference. The model and training code are open-sourced and approachable for academia with limited compute budgets.

6 months ago
Yi-34B, Llama 2, and common practices in LLM training

EleutherAI corrects misconceptions from a New York Times article regarding Yi-34B and Llama 2, explaining common language model training practices and architectural similarities among LLMs. Highlighting that the use of known technology does not imply indebtedness to Meta's Llama 2, they clarify that architectural innovations often predate models like Llama and Yi-34B are based on public research. Training and dataset curation remain pivotal for model differentiation.

6 months ago
Improvements to the fine-tuning API and expanding our custom models program

OpenAI introduces new features to its fine-tuning API for developers, providing more control over model training, and expands its Custom Models Program to assist organizations in creating domain-specific AI models. Improvements include epoch-based checkpoint creation, comparative playground, third-party integration, and comprehensive validation metrics. They also offer assisted fine-tuning for organizations needing extra support and custom-trained models that infuse domain knowledge into AI.

6 months ago
Command R+: A Scalable LLM Built for Business

Cohere introduces Command R+, a highly capable large language model built for enterprise use and available on Microsoft Azure. Command R+ boasts a range of features like advanced RAG, multilingual support, and tool use, which improves performance and efficiency in business applications.

6 months ago
The Angry Video Game Model: neural network arch to predict videogame ratings

Henrique Magalhães Soares investigates the ability of various neural network architectures to predict video game review scores, despite the lack of numerical ratings in Angry Video Game Nerd episodes. Techniques like sentiment classification through machine learning aim to quantify implicit review sentiments.

6 months ago
Washington's Lottery forced to pull site after creating AI porn of lotto user

A Washington’s Lottery AI-powered mobile site, meant for entertainment by superimposing user faces onto dream vacation scenarios, was taken offline after producing an inappropriate image that resembled softcore pornography. A Tumwater mother, Megan, experienced this malfunction when her AI-generated photo appeared almost nude, prompting concerns about the site's content moderation and leading to an investigation by the lottery officials.

6 months ago
Metaforms AI

Metaforms is spearheading the evolution of feedback and survey platforms by introducing AI-driven forms that offer speedier setup, higher completion rates, and deeper insights. Labelled as Typeform's AI Successor, it empowers businesses to ask personalized and contextual questions, and flags new data through adaptive AI.

6 months ago
Brave Leo on iOS

Brave has officially completed the roll-out of its AI browser assistant, Brave Leo, on iOS devices, making it available on all platforms. Leo is a privacy-preserving AI with voice-to-text capability, PDF summarization, and multi-language support, enhancing mobile productivity without compromising privacy.

6 months ago
Parity

Parity is a customizable scheduling assistant designed to save users time by managing their meeting arrangements. Integrating seamlessly with email, Parity works with common tools and services such as Zoom, Office, and Google, allowing users to coordinate, reschedule, and brief on events effortlessly.

6 months ago
Eververse

Eververse is a platform designed for product teams, offering AI-powered tools to process feedback, prioritize features, and create dynamic roadmaps with ease.

6 months ago
Intellectia.AI

Intellectia.ai offers an AI-powered platform for smarter investment decisions, featuring tools for technical analysis, stock picks, trade ideas, and price drivers, designed to enhance financial intelligence and provide valuable market insights.

6 months ago
Ask HN: Has Anyone Trained a personal LLM using their personal notes?

A Hacker News user inquires about utilizing personal notes to train a personal Large Language Model (LLM), and several community members share their own experiences and tools, ranging from talk therapy summaries to fine-tuning generic LLMs and creating local AI assistants.

7 months ago
YC's Winter 2024 Demo Day confirms that we are indeed in an AI bubble

TechCrunch's coverage of Y Combinator's Winter 2024 Demo Day highlights a blooming AI bubble, with startups across various fields from legal tech, fintech, to healthtech showcasing AI-powered solutions. Standouts include Aidy for grant applications, Givefront for nonprofit banking, and Buster linking databases with language models. Startups like Numo, Intercept, and Nuanced Inc. focused on niche markets and emerging issues such as deep fakes and misinformation. Critical solutions for everyday problems were also presented by companies like Abel, Starlight Charging, and Pico.

7 months ago
Show HN: Plandex – an AI coding engine for complex tasks

Plandex is an open source, terminal-based AI coding engine designed to handle complex tasks by breaking them into subtasks and implementing them step-by-step. It uses the OpenAI API and is compatible with Mac, Linux, FreeBSD, and Windows.

7 months ago
Israel's 'Lavender' AI used for Gaza kill lists, report claims

An Israeli media investigation claimed that Israel uses an AI software named 'Lavender' to generate 'kill lists' in Gaza, potentially targeting about 37,000 individuals with air strikes based on broad parameters with minimal oversight.

7 months ago
'Lavender': The AI machine directing Israel's bombing in Gaza

An exposé by +972 Magazine revealed the Israeli army's use of an AI system named 'Lavender', which marked tens of thousands of Gazans as targets during the war on Gaza, with little human oversight, resulting in high civilian casualties.

7 months ago
Billie Eilish, Pearl Jam, 200 artists say AI poses threat to their livelihoods

Over 200 artists, including Billie Eilish and Pearl Jam, have signed an open letter by the Artist Rights Alliance calling out AI's existential threat to musicians’ livelihoods. They argue generative AI devalues artists' rights and could lead to diminished royalties, affecting many who depend on fair compensation for their work.

7 months ago
Stable Audio 2.0

Stable Audio 2.0 by Stability AI revolutionizes AI-generated audio with capabilities for producing high-quality, full tracks with coherent structures up to three minutes at 44.1kHz stereo. It now includes audio-to-audio generation, trained exclusively on licensed data from AudioSparx, and integrates safeguards like Audible Magic for copyright protection.

7 months ago
Welcome to the AI Gadget Era

April 2024 marks the advent of AI-first gadgets with companies like Humane, Rabbit, Brilliant, and Meta set to launch new devices. These gadgets, such as AI Pins, smart glasses, and AI-powered R1, aim to make AI the core of user experience. Unlike smartphones, these devices seek to eliminate interaction complexity by interpreting direct commands. As the technology is still in its infancy, challenges remain, but the sector hints at a future where interaction with technology becomes seamless.

7 months ago
Opera becomes the first major browser with built-in access to local AI models

Opera has introduced experimental support for 150 local Large Language Model (LLM) variants in the Opera One browser's developer stream. This feature allows for the use of generative AI while keeping user data local. Supported models include Llama, Vicuna, Gemma, and Mixtral. Users can activate the feature in the Opera Developer version and manage their AI models without reliance on cloud services. Opera continues its AI innovation with this inclusion.

7 months ago
CallStar

CallStar.AI is a platform offering AI-powered services tailored for different domains such as Movies & TV, Anime, Games, Science, Music, Books, Tech, Religion, Politics, History, and Business.

7 months ago
PyTorch Library for Running LLM on Intel CPU and GPU

Intel Analytics' IPEX-LLM is a PyTorch library for accelerating local LLM inference and fine-tuning on Intel CPUs and GPUs. It integrates seamlessly with llama.cpp, HuggingFace, and other tools, supporting more than 50 models.

7 months ago
Fracttal One

Fracttal One ofrece una solución de mantenimiento innovadora con una interfaz moderna e intuitiva. Con nuevas funcionalidades para un control y eficiencia mejorados, y aplicaciones de Inteligencia Artificial y mantenimiento predictivo.

7 months ago
SongSens.ai

SongSens.ai enhances your music experience by translating foreign song lyrics into your own language, providing context, pronunciation guides, and in-depth lyric explanations. Sign up and translate 3 songs for free. Use credits to discover more songs and words.

7 months ago
Build Social Skills with AI Voice Dates

Rizz is a platform designed to help users practice their social skills. It offers a way to get feedback and build confidence through training.

Want stories like these in your inbox?

BigCheese Weekly newsletter will be sent to you each week.