🥟 Chao-Down #4: Anthropic releases a competitor to ChatGPT, Adobe says customer data not used in AI Tools

Welcome to the fourth edition of the “Chao”-Downs newsletter, a bite-sized digest of stories at the intersection of Tech, AI, Business and Life. It is the companion to Chaos Theory, the weekly newsletter featuring longform, original content from your resident Chaos Coordinator.

Competition is healthy and today’s stories feature two takes on some of the latest developments happening in AI. One discusses how the entrants of open-source tools like OpenAI Triton have started to ween away mindshare from NVIDIA’s CUDA ecosystem by presenting an alternative to writing software on non-NVIDIA hardware. The other highlights a rival to ChatGPT released by the AI research lab Anthropic that anecdotally appears to improve in several areas that ChatGPT lacks.

It remains to be seen who will emerge as the overall winner in this space, but one thing is for sure. There is never a boring day in AI.

-Alex

What happened in AI? 📰

How Nvidia’s CUDA Monopoly In Machine Learning Is Breaking - OpenAI Triton And PyTorch 2.0 (semianalysis.com)

Tesla video promoting self-driving was staged, engineer testifies (Reuters)

Adobe Says AI Tools Not Being Trained With Customer Data (Bloomberg)

Comparing Anthropic’s competitor to ChatGPT Claude with OpenAI’s ChatGPT (Scale AI Blog)

Always be Learnin’ 📕 📖

• The Illustrated Stable Diffusion – Jay Alammar – Visualizing machine learning one concept at a time (Link)

• WebML for Web Developers and Creatives with Tensorflow.js (Link)

Projects to Keep an Eye On 🛠

Midjourney Prompt Helper - Refine prompts that you put into Midjourney by assigning weight to particular areas of focus and seeing example output (link)

PatternedAI - Seamless Pattern Maker with Artificial Intelligence (Link)

Using OpenAI’s Whisper and ECAPA-TDNN model from SpeechBrain to transcribe YouTube videos and classify who is speaking (Link)

The Latest In AI Research 💡

Grounding Language Models In-Context: Improving Text Generation and Attribution for Off-the-Shelf LMs (AI21 Blog)

Transformers as Algorithms: Generalization and Implicit Model Selection in In-context Learning (Arxiv)

Compositional Visual Generation with Composable Diffusion Models (energy-based-model.github.io)

The World Outside of AI 🌎

‘We’re all vulnerable’: One in 10 people will end up with long COVID (Link)

Inside Elon Musk’s “extremely hardcore” Twitter - (The Verge)

The Taliban started to make use of Twitter Blue’s verification system, what Twitter’s own trust and safety team predicted months ago (BBC)

Apple, under pressure to divest from China, has few alternative paths of doing so (Financial Times)

Microsoft, Amazon and other tech companies have laid off more than 60,000 employees in the last year (MSN)

One Last Bite 😋