🥟 Chao-Down #331 OpenAI reveals GPT-4o towards a more conversational voice assistant, Cruise self-driving cars are back on the road driving autonomously, How AI systems get better at tricking people
Plus, a look at the surge of AI firms selling hyper-accurate deepfake detection.
OpenAI announced GPT-4o, its latest AI model with text, vision, and audio capabilities. GPT-4o ("o" for "Omni)" will be accessible to all ChatGPT users, including those on the free version.
Compared to previous models, OpenAI says that GPT-4o is a step closer to "much more natural human-computer interaction” as it features more personality and real-time information processing and natural conversation. It’s also a single model which as I highlight below has a particularly desirable property.
Not to be outdone, a day after OpenAI makes its big announcements, Google is expected to show the world what they’ve been up to with AI at their Google I/O developer conference.
It’ll be a tough act to follow up.
-Alex, your resident Chaos Coordinator.
What happened in AI? 📰
Cruise is back driving autonomously for the first time since pedestrian-dragging incident (The Verge)
AI systems are getting better at tricking us (MIT Technology Review)
Surge of new AI firms claim to offer hyperaccurate deepfake detection (The Washington Post)
ChatGPT's new face is a black hole (TechCrunch)
Today’s AI models are impressive. Teams of them will be formidable (Economist)
The thingification of AI - The broken-gadget era is upon us. (The Atlantic)
Always be Learnin’ 📕 📖
Machine Unlearning in 2024 - Ken Ziyu Liu - (Stanford Computer Science)
How To Price A Data Asset - by Abraham Thomas - Pivotal (Substack)
The Anatomy of a Successful Team Squad - by Nicola Ballotta (hybridhacker.email)
Projects to Keep an Eye On 🛠
huggingface/lerobot: 🤗 LeRobot: State-of-the-art Machine Learning for Real-World Robotics in Pytorch (Github)
Skyvern-AI/skyvern: Automate browser-based workflows with LLMs and Computer Vision (Github)
BatsResearch/bonito: A lightweight library for generating synthetic instruction tuning datasets for your data without GPT. (Github)
The Latest in AI Research 💡
You Only Cache Once: Decoder-Decoder Architectures for Language Models (arxiv)
Iterative Reasoning Preference Optimization (arxiv)
Self-Play Preference Optimization for Language Model Alignment (arxiv)
The World Outside of AI 🌎
GameStop meme stock mania is back — and so is Roaring Kitty (qz.com)
Gen Z is struggling to land jobs and start careers (Axios)
California Beaches Are a New Gateway for Illegal Immigration (WSJ)
The changing picture of disease — living longer may not mean living healthier (ft.com)
America Wasn’t Made for Walking, and It’s Killing Us (Bloomberg)
These new managerial cities are the true winners of remote work (The Washington Post)
One Last Bite 😋
