🤖 Last Week's AI Highlights from HN #13

Discover the latest breakthroughs and trends in artificial intelligence

Author

Vasco
August 19, 2023

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

Greetings, all!

This are this week’s highlights:

Azure ChatGPT: Private and secure ChatGPT for internal enterprise use
https://github.com/microsoft/azurechatgpt
https://news.ycombinator.com/item?id=37112741

How Is LLaMa.cpp Possible?
🤔 Summarizing a complex technical explanation of how quantization enables the execution of a large AI model on a single CPU, for a project called LLaMa.cpp. N/A.
https://finbarr.ca/how-is-llama-cpp-possible/
https://news.ycombinator.com/item?id=37140013

Opendream: A layer-based UI for Stable Diffusion Learn more
https://github.com/varunshenoy/opendream
https://news.ycombinator.com/item?id=37136898

Following pushback, Zoom says it won't use customer data to train AI models
😃 Zoom’s experience highlights the challenge tech organizations face when integrating AI into their products and services. Following pushback, the company has announced it won’t use customer data to train its AI models.
https://www.darkreading.com/analytics/following-pushback-zoom-says-it-won-t-use-customer-data-to-train-ai-models
https://news.ycombinator.com/item?id=37123572

Show HN: AI-town, run your own custom AI world SIM with JavaScript
🤩 AI Town is a MIT-licensed, deployable starter kit for building and customizing a virtual town populated with AI characters who can chat and socialize. It includes features such as branching, tags, commits, security, projects, and more.
https://github.com/a16z-infra/ai-town
https://news.ycombinator.com/item?id=37128293

AI bots are now better than humans at decoding CAPTCHAs
🤖 AI bots have surpassed humans in solving CAPTCHA puzzles, with accuracy up to 15% higher than humans and solving times significantly faster. The study found that in a contextualized setting, AI bots are even faster than humans at solving the puzzles.
https://qz.com/ai-bots-recaptcha-turing-test-websites-authenticity-1850734350
https://news.ycombinator.com/item?id=37160744

New York Times considers legal action against OpenAI as copyright tensions swirl
🤬New York Times is considering legal action against OpenAI due to copyright tensions over AI's use of the paper's content.
https://text.npr.org/1194202562
https://news.ycombinator.com/item?id=37157989

Fine-Tuning Llama-2: A Comprehensive Case Study for Tailoring Custom Models
🤔 Analyzing the performance of Llama-2 models for various use cases, this blog post provides a comprehensive guide to fine-tuning to achieve significant accuracy improvements compared to GPT-4.
https://www.anyscale.com/blog/fine-tuning-llama-2-a-comprehensive-case-study-for-tailoring-models-to-unique-applications
https://news.ycombinator.com/item?id=37090632

Learn as you search (and browse) using generative AI
🤩 Discover new ways to search and browse using AI technology!
https://blog.google/products/search/google-search-generative-ai-learning-features/
https://news.ycombinator.com/item?id=37161097

Fresh evidence of ChatGPT’s political bias: study
🤔New study reveals fresh evidence of political bias in ChatGPTs.
https://www.uea.ac.uk/news/-/article/fresh-evidence-of-chatgpts-political-bias-revealed-by-comprehensive-new-study
https://news.ycombinator.com/item?id=37167185

Show HN: LlamaGPT – Self-hosted, offline, private AI chatbot, powered by Llama 2
🤖 Self-hosted, offline chatbot powered by Llama 2, with no data leaving the device.
https://github.com/getumbrel/llama-gpt
https://news.ycombinator.com/item?id=37148210

Deep Learning Systems
📚 This page provides an overview of the lectures, assignments, staff, and forums for 10-414/714: Deep Learning Systems course at CMU, with lecture videos and slides linked.
https://dlsyscourse.org/lectures/
https://news.ycombinator.com/item?id=37101515

New York Times considers legal action against OpenAI as copyright tensions swirl
🤔The New York Times is considering legal action against OpenAI for copyright infringement over their use of the paper's articles to train a chatbot.
https://www.npr.org/2023/08/16/1194202562/new-york-times-considers-legal-action-against-openai-as-copyright-tensions-swirl
https://news.ycombinator.com/item?id=37154840

My deep learning rig (2022)
🤔 Summary of deep learning rig: Building a powerful deep learning rig with GPUs, CPUs, and other components to maximize GPU density and performance.
https://nonint.com/2022/05/30/my-deep-learning-rig/
https://news.ycombinator.com/item?id=37138672

Probabilistic Machine Learning: Advanced Topics
🤩 Comprehensive overview of Probabilistic Machine Learning topics, written by Kevin Patrick Murphy and published by MIT Press in 2023, with endorsements from renowned researchers.
https://probml.github.io/pml-book/book2.html
https://news.ycombinator.com/item?id=37137810

OpenAI acquires Global Illumination
🤩 OpenAI acquires Global Illumination, with the entire team joining OpenAI to work on core products including ChatGPT.
https://openai.com/blog/openai-acquires-global-illumination
https://news.ycombinator.com/item?id=37150098

Microsoft AI suggests Ottawa food bank as a “cannot miss” tourist spot
🤔 Microsoft AI mistakenly suggests food bank as a destination in a Microsoft Travel article.
https://arstechnica.com/information-technology/2023/08/microsoft-ai-suggests-food-bank-as-a-cannot-miss-tourist-spot-in-canada/
https://news.ycombinator.com/item?id=37171955

Doctor GPT: A Large Language Model That Can Pass the US Medical Licensing Exam
🤖 DoctorGPT is an LLM that can pass the US Medical Licensing Exam, working offline, cross-platform, and keeping your health data private.
https://github.com/llSourcell/DoctorGPT
https://news.ycombinator.com/item?id=37105352

Consensus: Use AI to find insights in research papers
🤩 Discover expert knowledge with Consensus: search engine powered by AI to find insights in research papers & trustable sources.
https://consensus.app/
https://news.ycombinator.com/item?id=37166880

AI cameras catch 297 drivers in three days in Cornwall
🤔 AI camera system in Cornwall catches 297 drivers in three days breaking the law, with most of them not wearing seatbelts or using mobile phones.
https://www.bbc.co.uk/news/uk-england-cornwall-66508840
https://news.ycombinator.com/item?id=37147682

ML is useful for many things, but not for predicting scientific replicability
🤔 Summary: ML can be useful for many things, but not for predicting scientific replicability due to drawbacks such as small training sample, potential for data distribution changes, and easy gaming of ML models.
https://www.aisnakeoil.com/p/machine-learning-is-useful-for-many
https://news.ycombinator.com/item?id=37089573

What coal and Jevons’ paradox tell us about AI and data
🤔 Summary: Jevons' Paradox is used to explain the potential rebound effect of AI in the data industry, suggesting that while more efficient processes may reduce the individual workload, the increased abundance of data will likely stimulate an even greater demand for data practitioners.
https://hex.tech/blog/jevons-paradox-demand-for-insight/
https://news.ycombinator.com/item?id=37129853

Thinking about God increases acceptance of AI in decision-making
https://www.pnas.org/doi/10.1073/pnas.2218961120
https://news.ycombinator.com/item?id=37143794

Show HN: Epsilla – Open-source vector database with low query latency
🤖A high-performance Vector Database Management System built by Epsilla.
https://github.com/epsilla-cloud/vectordb
https://news.ycombinator.com/item?id=37122064

Iowa School District is using ChatGPT to determine banned books
🤬 An Iowa school district is using AI to ban books, 19 titles have already been flagged for removal.
https://www.engadget.com/mason-city-iowa-school-district-ai-book-ban-censorship-202541565.html
https://news.ycombinator.com/item?id=37151749

Sites scramble to block ChatGPT web crawler after instructions emerge
🤯 Sites scramble to block OpenAI's ChatGPT web crawler after instructions emerge, with restrictions not applying to current models but potentially future versions.
https://arstechnica.com/information-technology/2023/08/openai-details-how-to-keep-chatgpt-from-gobbling-up-website-data/
https://news.ycombinator.com/item?id=37094463

Show HN: NotYetNews – AI-Generated News from the Future
😐 100 years in the future, the world faces political tensions between nations and corporations, AI-led trials and punishments, and the effects of a devastating Maui Fire on Hawaiian Electric.
https://www.notyet.news/
https://news.ycombinator.com/item?id=37110942

What if Generative AI turned out to be a Dud?
🤔 Summary: A discussion of the potentially negative economic and geopolitical implications if Generative AI turns out to be a Dud.
https://garymarcus.substack.com/p/what-if-generative-ai-turned-out
https://news.ycombinator.com/item?id=37116216

AI Detection Tools Falsely Accuse International Students of Cheating
🤬 AI Detection Tools are falsely accusing international students of cheating, leading to Stanford scientists finding a bias against non-native English speakers in the AI Detection Tools.
https://themarkup.org/machine-learning/2023/08/14/ai-detection-tools-falsely-accuse-international-students-of-cheating
https://news.ycombinator.com/item?id=37127003

Show HN: Broken Bear, the AI teddy bear that loves your broken self
💖 Embrace self-love and accept your flaws with BrokenBear.
https://brokenbear.com/
https://news.ycombinator.com/item?id=37108298

Generative AI – A curated list of Generative AI projects and services
https://github.com/steven2358/awesome-generative-aihttps://news.ycombinator.com/item?id=37107101

Show HN: A website chatbot that also uses APIs
🤩Create a powerful AI chatbot with customizable features to answer customer questions, generate documents, and connect to back-end data.
https://chatwith.tools
https://news.ycombinator.com/item?id=37120093

Show HN: Marqo – Vectorless Vector Search
GitHub - marqo-ai/marqo: Vector search for humans. 🔍 Vector search tool for humans to quickly find what they are looking for.
https://github.com/marqo-ai/marqo
https://news.ycombinator.com/item?id=37147140

Automatically classifying the content of sound files using ML
🤖 Summary: Creating a ML model to automatically classify sound files, using Google AudioSet and YAMNet, and using a Python script to add metadata to Soundly library.
https://blog.gingerbeardman.com/2023/08/13/automatically-classifying-the-content-of-sound-files-using-ml/
https://news.ycombinator.com/item?id=37111867

Show HN: A free course on how to write a good Midjourney/ChatGPT prompt Clone with HTTPS
🤩 Summary: Explore free prompt engineering online course with ChatGPT and Midjourney tutorials included!
https://github.com/thinkingjimmy/Learning-Prompt
https://news.ycombinator.com/item?id=37119112

PlayHT2.0: State-of-the-Art Generative Voice AI Model for Conversational Speech
🤩 Introducing PlayHT2.0, the state-of-the-art Generative Voice AI Model for Conversational Speech with the addition of emotions, available in closed beta and with realtime speech generation capability.
https://news.play.ht/post/introducing-playht2-0-the-state-of-the-art-generative-voice-ai-model-for-conversational-speech
https://news.ycombinator.com/item?id=37091221

OpenAI may go bankrupt by 2024, AI bot costs company $700k every day
🤯OpenAI's AI chatbot ChatGPT may be headed for bankruptcy due to its high running cost of $700,000 every day and falling user base.
https://www.firstpost.com/tech/news-analysis/openai-may-go-bankrupt-by-2024-chatgpt-costs-company-700000-dollars-every-day-12986012.html
https://news.ycombinator.com/item?id=37109394

FEC moves towards potentially regulating AI deepfakes in campaign ads
🤔 The FEC is considering regulating AI deepfakes in campaign ads, as part of a larger discussion about the use of new technology in elections.
https://apnews.com/article/fec-artificial-intelligence-deepfakes-election-2024-95399e640bd1e41182f6c631717cc826
https://news.ycombinator.com/item?id=37115848

How to Get ChatGPT to Stop Apologizing?
🤔 Summarizing a prompt design question on how to get ChatGPT to stop apologizing excessively.
https://genai.stackexchange.com/questions/177/how-to-get-chatgpt-to-stop-apologizing#1
https://news.ycombinator.com/item?id=37090081

Report: Potential NYT lawsuit could force OpenAI to wipe ChatGPT and start over
🤔Potential NYT lawsuit could force OpenAI to wipe ChatGPT dataset and start over, possibly resulting in massive fines and loss of users.
https://arstechnica.com/tech-policy/2023/08/report-potential-nyt-lawsuit-could-force-openai-to-wipe-chatgpt-and-start-over/
https://news.ycombinator.com/item?id=37165646

College professors are searching for ways to “ChatGPT-proof” their exams
🤔 College professors are returning to paper exams and handwritten essays to combat students using ChatGPT.
https://www.businessinsider.com/chatgpt-driving-return-to-paper-exams-written-essays-at-universities-2023-8
https://news.ycombinator.com/item?id=37112901

Thanks, and see you next week!