what an artificial intelligence thinks data is
what an artificial intelligence thinks data is


Economists Warn That AI Like ChatGPT Will Increase Inequality (gizmodo.com) | [💬 discussion] - AI threatens to shift even more power from workers to the wealthy, shrinking tax bases and weakening governments, writes economist Yingying Lu. { #society }

‘Godfather of AI’ quits Google with regrets and fears about his life’s work - The Verge (theverge.com) | [💬 discussion] - Geoffrey Hinton who won the ‘Nobel Prize of computing’ for his trailblazing work on neural networks is now free to speak about the risks of AI. { #risk }

IBM to Pause Hiring for ‘Back-Office’ Jobs That AI Could Kill - Bloomberg (bloomberg.com) | [💬 discussion] - Roughly 7,800 IBM jobs could be replaced by AI, automation { #job }

OpenAI Threatens Popular GitHub Project With Lawsuit Over API Use | Tom’s Hardware (tomshardware.com) | [💬 discussion] - GPT4Free uses other sites' connections to OpenAI. { #gpt4 }

Samsung bans use of AI like ChatGPT for staff after misuse of chatbot (cnbc.com) | [💬 discussion] - Samsung is restricting the use of so-called generative artificial intelligence tools such as ChatGPT for employees after the company discovered such services were being misused. { #chatgpt #ban }

AI makes non-invasive mind-reading possible by turning thoughts into text | Artificial intelligence (AI) | The Guardian (theguardian.com) - Advance raises prospect of new ways to restore speech in those struggling to communicate due to stroke or motor neurone disease { #mind }

Google, Microsoft CEOs called to AI meeting at White House | Reuters (reuters.com) | [💬 discussion] - The chief executives of Alphabet Inc’s Google (GOOGL.O), Microsoft (MSFT.O), OpenAI and Anthropic will meet with Vice President Kamala Harris and top administration officials to discuss key artificial intelligence (AI) issues on Thursday, said a White House official.

Google DeepMind CEO Says Some Form of AGI Possible in a Few Years (archive.ph) | [💬 discussion] - Demis Hassabis cites need to develop artificial general intelligence responsibly { #deepmind }

Amnesty International criticised for using AI-generated images | Colombia | The Guardian (theguardian.com) | [💬 discussion] - Group has removed AI images used to promote their reports on social media, including fake photos of Colombia’s 2021 protests { #fake }

Nearly 25% of jobs are set to be disrupted in the next five years: WEF (cnbc.com) | [💬 discussion] - The world of work is set to go through major changes in the coming years — with almost a quarter of jobs changing in the next five years, according to a new report from the World Economic Forum. { #job }

GitLab and Google Cloud Partner to Expand AI-Assisted Capabilities with Customizable Gen AI Foundation Models - May 2, 2023 (googlecloudpresscorner.com) | [💬 discussion] - GitLab AI-powered security feature to leverage Google Cloud’s foundation models to empower customers to identify and remediate vulnerabilities more efficiently while maintaining privacy standards { #devop }

This company adopted AI. Here’s what happened to its human workers : Planet Money : NPR (npr.org) | [💬 discussion] - The study is about an AI support system (a “recent version of GPT”, sources link to GPT4, fine-tuned on internal company data) supporting customer support agents at a software company that provides business process software. It increased productivity (measured in number of chats resolved by employee) by 13.8%, the lower the performance before, the higher the impact. The performance increase is mostly based on being able to handle more chats in parallel and handling each chat quicker. For reference, an agent resolved about 2.12 chats/h before and this increased by 0.47 chats/h. { #job #study }

OpenAI’s Losses Doubled to $540 Million as It Developed ChatGPT — The Information (theinformation.com) - OpenAI’s losses roughly doubled to around $540 million last year as it developed ChatGPT and hired key employees from Google, according to three people with knowledge of the startup’s financials. The previously unreported figure reflects the steep costs of training its machine-learning models during the period before it started selling access to the chatbot. { #chatgpt }

Amazon plans to give Alexa ChatGPT-like capabilities - The Verge (theverge.com) | [💬 discussion] - Amazon’s plan to make more money from Alexa could lean into more generalized large language models like the ones behind ChatGPT and Bard. { #chatgpt }

Researcher Meredith Whittaker says AI’s biggest risk isn’t ‘consciousness’—it’s the corporations that control them (fastcompany.com) | [💬 discussion] - The former Googler and current Signal president on why she thinks Geoffrey Hinton’s alarmism is a distraction from more pressing threats. { #risk }

DEF CON to set thousands of hackers loose on LLMs • The Register (theregister.com) | [💬 discussion] - Can’t wait to see how these AI models hold up against a weekend of red-teaming by infosec’s village people { #defcon }


ChatGPT for YouTube/Google|Chrome Extension - Glarity Summary (glarity.app) - Glarity Summary is a ChatGPT for YouTube/Google extension that can summarize YouTube videos and Google searches, also supports Yahoo! ニュース, PubMed, PMC, NewsPicks, Github, Nikkei, Bing, Google Patents and any page summary. { #summary }

DB Pilot - A beautiful database GUI. Enhanced by AI. (dbpilot.io) | [💬 discussion] - A beautiful database GUI. Enhanced by AI. { #database }

guidde・Magically create video documentation with AI (guidde.com) - guidde is the generative AI platform for business that helps your team create video documentation 11x faster. { #video #doc }

GitHub - irgolic/AutoPR: Fix issues with AI-generated pull requests, powered by ChatGPT (github.com) - Fix issues with AI-generated pull requests, powered by ChatGPT { #github }

GitHub - yasyf/compress-gpt: Self-extracting GPT prompts for ~70% token savings (github.com) - Self-extracting GPT prompts for ~70% token savings { #token #python }

GitHub - jerryjliu/llama_index: LlamaIndex (GPT Index) is a project that provides a central interface to connect your LLM’s with external data. (github.com) - LlamaIndex (GPT Index) is a project that provides a central interface to connect your LLM’s with external data. { #llm #index }

CraftAI (ai.craftable.pro) | [💬 discussion] - GPT-Powered Admin Panel Generator { #gpt }

Mojo 🔥: Programming language for all of AI (modular.com) | [💬 discussion] - Mojo combines the usability of Python with the performance of C, unlocking unparalleled programmability of AI hardware and extensibility of AI models. { #language #python }

GitHub - 1rgs/jsonformer: A Bulletproof Way to Generate Structured JSON from Language Models (github.com) | [💬 discussion] - A bulletproof way to generate structured output from LLMs { #llm #python }

GitHub - piercefreeman/gpt-json: Structured and typehinted GPT responses in Python (github.com) | [💬 discussion] - Structured and typehinted GPT responses in Python { #python #json }

GitHub - nadermx/backgroundremover: Background Remover lets you Remove Background from images and video using AI with a simple command line interface that is free and open source. (github.com) | [💬 discussion] - Background Remover lets you Remove Background from images and video using AI with a simple command line interface that is free and open source. { #image #oss }

GitHub - openai/shap-e: Generate 3D objects conditioned on text or images (github.com) | [💬 discussion] - Generate 3D objects conditioned on text or images { #3d }

ReRender AI - Photorealistic Renders for Architectures. Enhance your AutoCAD/Blender/Sketchup results! (rerenderai.com) | [💬 discussion] - Upload a picture of your project, discover 20+ unique design styles, and enjoy photorealistic renders in just seconds! { #image }


GitHub - mlc-ai/mlc-llm: Enable everyone to develop, optimize and deploy AI models natively on everyone’s devices. (github.com) | [💬 discussion] - Enable everyone to develop, optimize and deploy AI models natively on everyone’s devices. { #llm }

GitHub - georgia-tech-db/eva: AI-Relational Database System | SQL meets Deep Learning (github.com) | [💬 discussion] - AI-Relational Database System | SQL meets Deep Learning { #database }

Timeline of Transformer Models / Large Language Models (AI / ML / LLM) (ai.v-gar.de) | [💬 discussion] - This is a collection of important papers in the area of Large Language Models and Transformer Models. It focuses on recent development, especially from mid-2022 onwards, and in no way claims to be exhaustive. It is actively updated. { #llm }

🐦 view tweet (twitter.com) | [💬 discussion] - Let this sink in: AI now performs better at Theory of Mind tests than actual humans. GPT-4 scores 100% on 2-shot chain-of-thought and step-by-step reasoning benchmarks. { #gpt-4 }

New Layer of Abstraction, Not the End of Software Engineering (adamcohenhillel.substack.com) - An optimistic view on how Software Engineering will progress in the upcoming years { #coding }

Expanding ChatGPT Code Interpreter with Python packages, Deno and Lua | Simon Willison’s TILs (til.simonwillison.net) - The ChatGPT Code Interpreter alpha remains incredibly interesting. I wrote about how I was using it for Python and SQLite benchmarking a few weeks ago. Today I found a neat pattern for expanding its capabilities with custom binaries. { #chatgpt #coding }

Why Chatbots Are Not the Future of Interfaces (wattenberger.com) | [💬 discussion] - Why Chatbots Are Not the Future { #prompting #llm }

🎬 watch video | [📝 text-summary] (youtube.com) | [💬 discussion] - Sal Khan, the founder and CEO of Khan Academy, thinks artificial intelligence could spark the greatest positive transformation education has ever seen. He shares the opportunities he sees for students and educators to collaborate with AI tools – including the potential of a personal AI tutor for every student and an AI teaching assistant for every teacher – and demos some exciting new features for their educational chatbot, Khanmigo. { #learning #chatgpt }

AI Is Tearing Wikipedia Apart (vice.com) | [💬 discussion] - Volunteers who maintain the digital encyclopedia are divided on how to deal with the rise of AI-generated content and misinformation. { #wiki }

GPT makes learning fun again (vipshek.com) | [💬 discussion] - Recently, I tried learning about a new subject using two approaches: 1) Google searching, 2) talking to GPT about it. I found, for the eighth time this month, that talking to GPT was far better than searching. The contrast was so striking that I thought it’s worth dissecting the differences to see why the GPT experience was so much better. { #gpt #learning }

🎬 watch video | [📝 text-summary] (youtube.com) - An AI artist explains his workflow. How it works — and why it takes a surprisingly long time to make something good. { #art }

Bad Actors Are Joining the AI Revolution: Here’s What We’ve Found in the Wild | HackerNoon (hackernoon.com) | [💬 discussion] - What security researcher Carlos Fernández and I have recently found on open-source registries tells a different story: bad actors are favoring simplicity, effectiveness, and user-centered thinking. And to take their malicious code to the next level, they’re also adding new features assisted by ChatGPT. { #hack #chatgpt }

Chatbot ‘journalists’ found running almost 50 AI-generated content farms | Artificial intelligence (AI) | The Guardian (theguardian.com) | [💬 discussion] - Websites churn out content, often advancing false narratives, to saturate with adverts, says anti-misinformation firm { #spam }

Avoiding hallucinations in LLM-powered Applications - Vectara (vectara.com) | [💬 discussion] - LLM hallucinations and how to avoid them with Grounded Generation { #llm #hallucination }

Beware of AI pseudoscience and snake oil – Baldur Bjarnason (baldurbjarnason.com) | [💬 discussion] - AI research is poor. Many of their claims will be proven wrong. AI software vendors have financial incentive to exaggerate the capabilities of their tools and make them hard to disprove. This undermines attempts at scientific rigour. Many of AI’s promises are snake oil. { #scam }

GitHub - openlm-research/open_llama (github.com) | [💬 discussion] - In this repo, we release a permissively licensed open source reproduction of Meta AI’s LLaMA large language model. In this release, we’re releasing a public preview of the 7B OpenLLaMA model that has been trained with 200 billion tokens. We provide PyTorch and Jax weights of pre-trained OpenLLaMA models, as well as evaluation results and comparison against the original LLaMA models. Stay tuned for our updates. { #llm #oss }

AI-generated beer commercial contains joyful monstrosities, goes viral | Ars Technica (arstechnica.com) | [💬 discussion] - 30-second spot set to “All Star” may inspire awe—or nightmares. { #viral #video }

Prompt injection explained, with video, slides, and a transcript (simonwillison.net) - I participated in a webinar this morning about prompt injection, organized by LangChain and hosted by Harrison Chase, with Willem Pienaar, Kojin Oshiba (Robust Intelligence), and Jonathan Cohen and Christopher Parisien (Nvidia Research). { #prompting }

The Full Story of Large Language Models and RLHF (assemblyai.com) | [💬 discussion] - Large Language Models have been in the limelight since the release of ChatGPT, with new models being announced seemingly every week. This guide walks through the essential ideas of how these models came to be. { #llm }

GitHub - IST-DASLab/sparsegpt: Code for the paper “SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shot”. (github.com) | [💬 discussion] - Code for the paper “SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shot”. { #gpt }

Chatbot Arena: Benchmarking LLMs in the Wild with Elo Ratings | LMSYS Org (lmsys.org) | [💬 discussion] - We present Chatbot Arena, a benchmark platform for large language models (LLMs) that features anonymous, randomized battles in a crowdsourced manner. In this blog post, we are releasing our initial results and a leaderboard based on the Elo rating system, which is a widely-used rating system in chess and other competitive games. We invite the entire community to join this effort by contributing new models and evaluating them by asking questions and voting for your favorite answer. { #llm #rating }

Training a SOTA Code LLM in 1 week and Quantifying the Vibes — with Reza Shabani of Replit (latent.space) | [💬 discussion] - Ep. 10: The YOLO training run, live AmjadEval demo, and the future of Replit Ghostwriter! Plus: the journey from Finance to Data, from Data to AI, and from Kaplan to Chinchilla and beyond { #podcast #llm }

Google “We Have No Moat, And Neither Does OpenAI” (semianalysis.com) | [💬 discussion] - Leaked Internal Google Document Claims Open Source AI Will Outcompete Google and OpenAI { #oss }

Augmenting LLMs Beyond Basic Text Completion and Transformation - Deepgram Blog ⚡️ (blog.deepgram.com) | [💬 discussion] - Large Language Models (LLMs) are like an iceberg; maybe we only see the tip grazing above the surface. However, the further we dive below the surface, the bigger the iceberg becomes. The sheer amount of data they’re trained on is invisible to us and their emergent capabilities become evident as they keep scaling in size. LLMs are, by innate architectural property, very dynamic black boxes. { #llm }

Introducing Slack GPT, the future of AI in Slack | Slack (slack.com) | [💬 discussion] - Bringing the power of generative AI in Slack to help you work smarter, learn faster, and communicate better { #gpt }

Releasing 3B and 7B RedPajama-INCITE family of models including base, instruction-tuned & chat models — TOGETHER (together.xyz) | [💬 discussion] - The RedPajama project aims to create a set of leading open-source models and to rigorously understand the ingredients that yield good performance. A few weeks ago we released the RedPajama base dataset based on the LLaMA paper, which has galvanized the open-source community. The 5 terabyte dataset has been downloaded hundreds of times and used to train models like MPT, OpenLLaMA, OpenAlpaca. Today we are excited to release RedPajama-INCITE models, including instruct-tuned and chat versions. { #llm #oss }

It looks like GPT-4-32k is rolling out - API - OpenAI API Community Forum (community.openai.com) | [💬 discussion] - Just saw 32k pop up in the Playground, here is a quick hello world API call. { #chatgpt }

VardaGPT/STORY.md at master · ixaxaar/VardaGPT · GitHub (github.com) | [💬 discussion] - Using ChatGPT to generate a GPT project end-to-end. { #project }

⬡ ⬡ ⬡ ⏣ ⬡ ⬡ ⬡