Sarah Chieng+

I turned my wall into a book with Claude Code

#ad Reading more research papers was one of my top goals for 2026, so I used Claude’s Fable 5.1 and a projector to turn arXiv PDFs into a book on my wall. I can swipe through the pages with hand gestures :) Can you guess the paper? Try Claude Code: https://clau.de/SarahChieng

Why Jeffrey Wang Chose OpenAI

Jeffrey Wang traces his path to @OpenAI from Andrew Ng’s course and early generative-model papers to the research lineage that culminated in ChatGPT. #Shorts #OpenAI #ChatGPT

Why Reinforcement Learning Is Becoming Inference

Long-horizon agents need more tokens and more time to reason. Jeffrey Wang explains why reinforcement learning is increasingly an inference challenge at @OpenAI. #Shorts #OpenAI #ReinforcementLearning

From Gemini to Android: Google’s Model Strategy

Logan Kilpatrick explains how Google DeepMind’s model strategy evolves across Gemini, Gemma, cloud products, and Android. #Shorts #GoogleDeepMind #Gemini #Android

Why Speed Will Define the Next AI Products

Logan Kilpatrick explains why speed will define the next wave of AI products—and what faster Google DeepMind models could unlock. #Shorts #GoogleDeepMind #FutureOfAI #Gemini

Does AI Training Need Separate Stages?

Do pre-training and reinforcement learning eventually merge? Jeffrey Wang explains why continual learning remains a major research direction at @OpenAI. #Shorts #OpenAI #AIResearch

Making Google DeepMind’s AI Accessible to Everyone

AI capability only matters if people can use it. Logan Kilpatrick explains how Google DeepMind is working to make advanced technology more accessible. #Shorts #GoogleDeepMind #AccessibleAI #AIForEveryone

Inside Google DeepMind’s Science-Driven Culture

Logan Kilpatrick describes the science-driven culture inside Google DeepMind—and how it shapes ambitious research and long-term progress. #Shorts #GoogleDeepMind #AIResearch #DeepMind

How Quant Trading Became an AI Power User

Jeffrey Wang explains how Hudson River Trading became an early deep-learning power user—working with hardware and algorithms similar to @OpenAI, but aimed at markets. #Shorts #OpenAI #QuantTrading

How Google DeepMind Balances Model Performance

The best model is not always the biggest. Logan Kilpatrick explains how Google DeepMind balances quality, latency, and cost across use cases. #Shorts #GoogleDeepMind #AIModels #Gemini

Google DeepMind’s Vision for AI Developers

Logan Kilpatrick shares Google DeepMind’s vision for helping developers turn frontier model capabilities into useful AI products. #Shorts #GoogleDeepMind #AIDevelopers #Innovation

Why AI Models and Chips Must Be Co-Designed

Transformers are built from matrix multiplications, making hardware efficiency critical. Jeffrey Wang explains how @OpenAI co-designs models and chips together. #Shorts #OpenAI #AIHardware

Why Gemma Is Google DeepMind’s AI Test Bed

Logan Kilpatrick explains why Gemma acts as a test bed for efficient, open-model innovation at Google DeepMind. #Shorts #GoogleDeepMind #Gemma #OpenModels

I made Claude compete against itself

The smartest model does not automatically make the best agent. I put the same Opus 4.6 model, initial prompt, and empty repo inside Claude Code and Factory Droid, then asked both to clone Excalidraw. Factory Droid: 8 minutes, 20 tool calls, $1.60 Claude Code: 19 minutes, 40 tool calls, $1.87 The only difference was the harness. Get free Factory credits: https://forms.gle/PJ7pwAGou3dbKywJ8 #AI #CodingAgents #Claude #Shorts

Build AI Apps Faster with Google AI Studio

Logan Kilpatrick explains how Google AI Studio helps developers move from idea to prototype using models from Google DeepMind. #Shorts #GoogleDeepMind #GoogleAIStudio #AIDevelopment

The OpenAI Model That Stunned This Researcher

Jeffrey Wang recalls seeing the model that would become o3—and realizing how far pre-training plus reinforcement learning could push @OpenAI’s capabilities. #Shorts #OpenAI #O3