As Rocks May Think
A developer explores using AI coding agents like Claude to automate software development, research, and experimentation, focusing on implementing AlphaGo from scratch.
Eric Jang is an AI researcher and engineer writing about machine learning, robotics, and generalization. His blog covers topics from LLMs and reinforcement learning to the future of intelligent robots and AI research careers.
10 articles from this blog
A developer explores using AI coding agents like Claude to automate software development, research, and experimentation, focusing on implementing AlphaGo from scratch.
A robotics engineer reflects on leaving humanoid robotics company 1X, discussing the company's growth and the 'magical objects' driving AI progress.
Explores the concept of an 'intelligence frequency spectrum,' comparing AI's 1-2Hz conversational speed to faster/slower biological decision-making.
Explains how motor inertia and gearing affect robot safety and agility, using physics to show why lightweight, high-torque motors are key for advanced robots.
An analysis of the ARC Prize AI benchmark, questioning if human-level intelligence can be achieved solely through deep learning and transformers.
A robotics AI lead reflects on the field's future, discussing scaling robot autonomy with neural networks and parallels to large language models.
A summary and analysis of DeepMind's RoboCat paper, a self-improving foundation agent for robotic manipulation using Transformer models.
Explores the Reflexion technique where LLMs like GPT-4 can critique and self-correct their own outputs, a potential new tool in prompt engineering.
A roboticist argues for scaling robotics research like generative AI, focusing on data quality and iteration over algorithms for better generalization.
A deep learning researcher shares insights on the 2022 ML job market, comparing career options like FAANG, startups, and robotics, after joining Halodi Robotics.