🥟 Chao-Down #348 Anthropic shows off latest Claude 3.5 model plus Artifacts, Apple has a missing piece in China for broader AI push, Copywriters find new opportunities to make AI sound more human
Plus, is AI good at comedy? Comedians try using AI to write their routines.
Anthropic’s latest Claude 3.5 model is out and it reportedly is besting GPT-4o on various benchmarks on math reasoning and visual question answering. The model, which can analyze text and images as well as generate text, is twice as fast as their largest model while still offering good performance.
With this release, Anthropic is also introducing Artifacts, a dedicated window for users to see the code, documents, or web designs that the model generates turning what Claude from a “conversational AI to a collaborative work environment.”
With competition fast catching up, the question now on everyone’s mind is: when is OpenAI going to release GPT-5?
-Alex, your resident Chaos Coordinator.
What happened in AI? 📰
AI took their jobs. Now they get paid to make it sound human (bbc.com)
For Apple’s AI Push, China Is a Missing Piece (WSJ)
The Billion-Dollar Price Tag of Building AI (TIME)
How to spot a deepfake: the maker of a detection tool shares the key giveaways (The Guardian)
The AI Arms Race to Combat Fake Images Is Even—For Now (IEEE Spectrum)
What happened when 20 comedians got AI to write their routines (MIT Technology Review)
Always be Learnin’ 📕 📖
The Rise of Medium Code (Dagster Blog)
Python Notebooks for Fundamentals of Music Processing (audiolabs-erlangen.de)
Why we no longer use LangChain for building our AI agents (octomind.dev)
Projects to Keep an Eye On 🛠
tiangolo/typer: Typer, build great CLIs. Easy to code. Based on Python type hints. (Github)
Avaiga/taipy: Turns Data and AI algorithms into production-ready web applications in no time. (Github)
argilla-io/argilla: Argilla is a collaboration platform for AI engineers and domain experts that require high-quality outputs, full data ownership, and overall efficiency. (Github)
The Latest in AI Research 💡
Bag of Tricks: Benchmarking of Jailbreak Attacks on LLMs (arxiv)
The Prompt Report: A Systematic Survey of Prompting Techniques (arxiv)
Make It Count: Text-to-Image Generation with an Accurate Number of Objects (arxiv)
The World Outside of AI 🌎
TikTok says US ban is inevitable without a court order blocking law (Reuters)
The ‘Mad Men’ Era Is Long Over. Why Is Advertising Still So White? (nytimes.com)
Customer Experience Gets Worse. Again. (WSJ)
How Inside Out 2 Based Emotions on Science (TIME)
The rise—and fall—of the software developer (ADP Research Institute)
Tech founder Bryan Johnson says he slowed down his aging by editing his DNA (qz.com)