🤖 Last Week's AI Highlights from HN #10

Discover the latest breakthroughs and trends in artificial intelligence

Author

Vasco
July 28, 2023

Welcome back to HN AI Highlights, your weekly digest of all things AI!

This week's newsletter delves into significant advancements and controversial shifts within the AI industry. Our spotlight article for the week addresses a contentious incident where a Shopify employee highlights the company's move to replace laid-off employees with AI. This controversial action sparks a public backlash, casting a global spotlight on the ethical implications of substituting human labour with AI.
On a more exciting note, we also explore the latest in open AI research with the debut of Llama 2. This open-source project sees a variety of developments, making it an intriguing topic of discussion among AI enthusiasts and developers alike


Shopify employee breaks NDA to reveal firm replacing laid off workers with AI
🤬 Shopify employee breaks NDA to reveal firm's plan to replace laid off workers with AI, leading to increased workloads and customer satisfaction issues. CEO Tobi Lutke's attempt to cut costs and please shareholders has been met with backlash from the public.
https://thedeepdive.ca/shopify-employee-breaks-nda-to-reveal-firm-quietly-replacing-laid-off-workers-with-ai/
https://news.ycombinator.com/item?id=36828409

Llama2.c: Inference llama 2 in one file of pure C
💻 Summary: This project provides an inference Llama 2 platform in one file of pure C code. It includes a license, branches, tags, codespaces, a clone option, and more.
https://github.com/karpathy/llama2.c
https://news.ycombinator.com/item?id=36838051

Guide to running Llama 2 locally
🤩 Get step-by-step instructions on running Llama 2 locally on Mac, Windows, Linux, and even your phone! Learn how to use open-source tools like Llama.cpp, Ollama, and MLC LLM to run and fine-tune Llama 2 on your device and follow them to make a website about why llamas are cool.
https://replicate.com/blog/run-llama-locally
https://news.ycombinator.com/item?id=36865495

OpenAI shuts down its AI Classifier due to poor accuracy
😔 OpenAI's AI Classifier tool, which was released in January to detect content created using AI tools, has been decommissioned due to its low accuracy rate. The link to OpenAI's classifier is no longer available and the company is researching more effective ways to detect AI-generated content.
https://decrypt.co/149826/openai-quietly-shutters-its-ai-detection-tool
https://news.ycombinator.com/item?id=36862850

No one wants to talk to your chatbot
😐 A decade ago, web developers scrambled to create commercial websites. They dreamt of users directly interacting with their URLs, but the search engines became the entry point for users. With the launch of the iPhone, apps were developed, and users launched them directly from their screens. However, apps are limited and rely on the operating system for more advanced functions. Over the past few years, virtual personal assistants like Siri have become available, yet they are mostly used for simple tasks. The article suggests that companies should integrate more deeply with virtual personal assistants and their hosted apps.
https://lucas-mcgregor.medium.com/no-one-wants-to-talk-to-your-chatbot-9d8bb1a70266
https://news.ycombinator.com/item?id=36899656

‘World of Warcraft’ players trick AI-scraping website into publishing nonsense
🤣 World of Warcraft players tricked an AI-scraping games website into publishing nonsense by creating a thread discussing a non-existent feature. The AI-generated post was eventually taken down, but not before it caused a lot of amusement!
https://www.forbes.com/sites/paultassi/2023/07/21/world-of-warcraft-players-trick-ai-scraping-games-website-into-publishing-nonsense/
https://news.ycombinator.com/item?id=36815744

Stable Diffusion XL 1.0
🤩 Stability AI releases its latest image-generating model, Stable Diffusion XL 1.0, featuring more vibrant colors, better contrast, and advanced text-generation capabilities.
https://techcrunch.com/2023/07/26/stability-ai-releases-its-latest-image-generating-model-stable-diffusion-xl-1-0/
https://news.ycombinator.com/item?id=36880756

LLaMA2 Chat 70B outperformed ChatGPT
🤔 Summary of AlpacaEval: An LLM-based automatic evaluator that is fast, cheap, and reliable, based on the AlpacaFarm evaluation set to test models' ability to follow instructions and their comparison against GPT-4/Claude auto-annotators, with community contributions welcome.
https://tatsu-lab.github.io/alpaca_eval/
https://news.ycombinator.com/item?id=36895300

Show HN: Continue – Open-source coding autopilot
🤖️ Open-source autopilot for software development bringing the power of ChatGPT to your IDE.
https://github.com/continuedev/continue
https://news.ycombinator.com/item?id=36882146

From Python to Elixir Machine Learning
🤔 Summary of moving from Python to Elixir Machine Learning: Exploring the reasons why Elixir's ML ecosystem is growing, and how to port existing ML code from Python to Elixir.
https://www.thestackcanary.com/from-python-pytorch-to-elixir-nx/
https://news.ycombinator.com/item?id=36859785

Which GPU(s) to Get for Deep Learning
🤔 In-depth analysis of the best GPUs for deep learning in 2023, covering topics such as RAM, cores, tensor cores, caches, and more to understand how to choose the right GPU for deep learning.
https://timdettmers.com/2023/01/30/which-gpu-for-deep-learning/
https://news.ycombinator.com/item?id=36872514

Mali Government takes back .ml domain, brings down one of largest Lemmy servers
https://very.bignutty.xyz/notes/9hfv05qcs5xf7irr
https://news.ycombinator.com/item?id=36817179

A simple guide to fine-tuning Llama 2
🤔 A guide to fine-tuning Llama 2 to be a dialog summarizer, covering Huggingface libraries, finetuning on custom data, downloading the model & converting it to Huggingface format, running the fine-tuning notebook & running inference on the fine-tuned model.
https://brev.dev/blog/fine-tuning-llama-2
https://news.ycombinator.com/item?id=36852971

Llama and ChatGPT Are Not Open-Source
🤔 Summary: Meta's release of their open-source LLM, Llama 2, is not as open as other LLMs, and users are not provided with access to the model's training and architecture.
https://spectrum.ieee.org/openai-not-open
https://news.ycombinator.com/item?id=36900388

As Actors Strike for AI Protections, Netflix Lists $900k AI Job
🤬 As actors strike, Netflix lists a $900,000 AI job, while Hollywood executives claim it is "unrealistic" to pay actors more and decline to provide specifics on their AI investments.
https://theintercept.com/2023/07/25/strike-hollywood-ai-disney-netflix/
https://news.ycombinator.com/item?id=36878749

As AI porn generators get better, the stakes get higher
🤔As AI porn generators improve, the ethical implications of their use become increasingly complex.
https://techcrunch.com/2023/07/21/as-ai-porn-generators-get-better-the-stakes-raise/
https://news.ycombinator.com/item?id=36817738

Google Med-Palm M: Towards Generalist Biomedical AI
🤓 Summary of paper 'Towards Generalist Biomedical AI' by Tao Tu and 31 other authors, describing a large multimodal generative model and its potential applications and achievements.
https://arxiv.org/abs/2307.14334
https://news.ycombinator.com/item?id=36888948

Chidori – Declarative framework for AI agents (Rust, Python, and Node.js)
🚀 A reactive runtime for building durable AI agents
https://github.com/ThousandBirdsInc/chidori
https://news.ycombinator.com/item?id=36887412

How to scale LLMs better with an alternative to transformers
🤔 Exploring an alternative to Transformers to improve efficiency and performance, Monarch Mixer BERT (M2-BERT) is a sub-quadratic architecture with 25% fewer parameters/FLOPs than BERT and matches its quality.
https://hazyresearch.stanford.edu/blog/2023-07-25-m2-bert
https://news.ycombinator.com/item?id=36890036

ONNX runtime: Cross-platform accelerated machine learning
🤩 Accelerate machine learning process with ONNX Runtime supporting a variety of frameworks, operating systems, hardware platforms and APIs.
https://onnxruntime.ai/
https://news.ycombinator.com/item?id=36863522

ChatGPT for Android – Pre-orders
🤩 AI App that provides Instant Answers, Tailored Advice, Creative Inspiration, Professional Input and Learning Opportunities - ChatGPT.
https://play.google.com/store/apps/details?id=com.openai.chatgpt&hl=en_US
https://news.ycombinator.com/item?id=36819119

JetBrains IDE update previews “deeply integrated” AI Assistant
https://devclass.com/2023/07/25/jetbrains-ide-update-previews-deeply-integrated-ai-assistant/
https://news.ycombinator.com/item?id=36887702

Rewind.ai now available for iPhone
🤩Discover Rewind: an AI-driven app that provides personalized search, copy & paste capabilities, summarizations, and a visually appealing browsing layout for everything you’ve seen on your phone!
https://apps.apple.com/us/app/rewind-truly-personalized-ai/id6449795815
https://news.ycombinator.com/item?id=36877000

Why did Meta open-source Llama 2?
🤔 Meta open-sourced "Llama 2" for various reasons, such as reducing competitors’ moats, marketing, go-to-market strategies, and gaining goodwill.
https://matt-rickard.com/why-did-meta-open-source-llama
https://news.ycombinator.com/item?id=36817938

Publishers want billions, not millions, from AI
🤔 Publishers are pushing for billions instead of millions from AI companies for training data, but tech companies believe the payments should be much lower.
https://www.semafor.com/article/07/23/2023/publishers-want-billions-not-millions-from-ai
https://news.ycombinator.com/item?id=36849649

Tuning and Testing Llama 2, Flan-T5, and GPT-J with LoRA, Sematic, and Gradio
🤔 Exploring tools to summarize information with LoRA, Sematic, and Gradio for improved performance in a given sub-domain.
https://www.sematic.dev/blog/tuning-and-testing-llama-2-flan-t5-and-gpt-j-with-lora-sematic-and-gradio
https://news.ycombinator.com/item?id=36880149

New AI model translates vision and language into robotic actions
🤖 Summary: Google DeepMind's RT-2 is a vision-language-action model for robotics, which translates vision and language into robotic actions.
https://blog.google/technology/ai/google-deepmind-rt2-robotics-vla-model/
https://news.ycombinator.com/item?id=36905076

Job applicants are battling AI résumé filters with ‘white fonting’
🤔 Jobseekers are using a hack of "white fonting" their résumés to try and bypass AI and other tech filters when applying for jobs.
https://www.washingtonpost.com/technology/2023/07/24/white-font-resume-tip-keywords/
https://news.ycombinator.com/item?id=36857909

Actors decry 'existential crisis' over AI-generated ‘synthetic’ actors
🤯Actors fearing for their roles due to the rise of AI-generated ‘synthetic’ actors are decrying an 'existential crisis' as Hollywood studios and performers debate the use of AI in film and television.
https://www.reuters.com/technology/actors-decry-existential-crisis-over-ai-generated-synthetic-actors-2023-07-21/
https://news.ycombinator.com/item?id=3681861

How Will Language Modelers Like ChatGPT Affect Occupations and Industries?
🤔 Investigating the impact of advanced language modeling technologies on various occupations and industries.
https://arxiv.org/abs/2303.01157
https://news.ycombinator.com/item?id=36841211

Warcraft fans trick AI article bot with Glorbo hoax
🤔 Warcraft fans outsmart AI article bot with Glorbo hoax, sparking discussion on AI reliability.
https://www.bbc.com/news/newsbeat-66266344
https://news.ycombinator.com/item?id=36815902

Show HN: Litellm – Simple library to standardize OpenAI, Cohere, Azure LLM I/O
🤖Lightweight package to simplify LLM API calls - Azure, OpenAI, Cohere, Anthropic. Manages input/output translation.
https://github.com/BerriAI/litellm
https://news.ycombinator.com/item?id=36887711

JPMorgan warns that an AI bubble is brewing
📉 JPMorgan stays bearish on the stock market, warning of an AI bubble brewing amid record concentration in mega-cap stocks.
https://markets.businessinsider.com/news/stocks/stock-market-outlook-jpmorgan-bearish-ai-bubble-mega-cap-tech-2023-7
https://news.ycombinator.com/item?id=36869983

Stability AI Announces Stable Diffusion XL 1.0
🤩Stability AI releases Stable Diffusion XL 1.0, featured on Amazon Bedrock, providing the AI community with the most powerful text-to-image models.
https://stability.ai/press-articles/stable-diffusion-xl-1-featured-amazon-aws-bedrock
https://news.ycombinator.com/item?id=36881425

LLaMA 2 – Every Resource you need
🤔 A comprehensive guide to the open large language model LLaMA 2, covering its features, research, benchmarks, how to prompt it, how to train it, and other resources.
https://www.philschmid.de/llama-2
https://news.ycombinator.com/item?id=36824265

Transformers: The Google scientists who pioneered an AI revolution
🤔 Summary of article about Google scientists pioneering AI revolution: Discover how Google scientists have advanced the artificial intelligence field https://www.ft.com/content/37bb01af-ee46-4483-982f-ef3921436a50
https://news.ycombinator.com/item?id=36832797

Show HN: Pebble Finance – Personalize Your Investments Today
😃 Create your own personal ETF to diversify your investments.
https://pebble.finance/themes
https://news.ycombinator.com/item?id=36898503

How Stuff Works replaced writers with GPT-generated content and laid off editors
https://mastodon.social/@kristenhg/110714469552688499
https://news.ycombinator.com/item?id=36871498

Show HN: Shell AI – My Aggressively Minimal Open Source Assistant
🤖 Exciting AI shell assistant for minimal yet powerful automation!
https://github.com/ibigio/shell-ai
https://news.ycombinator.com/item?id=36861507

Apple is already using its chatbot for internal work
🤔 Apple is already using its GPT chatbot internally for prototyping and summarizing tasks, but is yet to find a suitable use for customer-facing support due to the potential of false information.
https://www.theverge.com/2023/7/23/23804825/apple-gpt-chatbot-apple-care-siri-chatgpt
https://news.ycombinator.com/item?id=36838178

Meta can call Llama open source as much as it likes, but that doesn't mean it is
https://www.theregister.com/2023/07/21/llama_is_not_open_source/
https://news.ycombinator.com/item?id=36817571

A History of Neural Networks
https://royalsocietypublishing.org/doi/10.1098/rsos.221454
https://news.ycombinator.com/item?id=36833208

Show HN: Borgo – a Rusty language that compiles to Go
🤖 Introduction to Borgo Programming Language, a high-level, garbage collected language for application development that transpiles to Go and is fully compatible with existing Go packages. Features Union Types and Pattern Matching, Error handling with ? operator, and Compatible with Go packages.
https://borgo-lang.github.io/
https://news.ycombinator.com/item?id=36847594

Thanks, and see you next week!