How OpenAI Shapes Its Research And What's Next – EP 46 Mark Chen
Mark Chen is one of the most influential people in artificial intelligence, and this episode is a rare look inside his world. As the Chief Research Officer at OpenAI, Chen helps decide which ideas get compute, which research threads scale, and how a 500-person lab navigates constant recruiting raids, model races, and the pressure to invent the next paradigm. We talk about the soup-delivering recruiting wars, how OpenAI thinks about staying ahead of Meta and Google, and what it actually feels like to rank 300 internal research projects when everyone believes theirs is the one that matters.
We also get into Chen’s path from math competitions and high-frequency trading to the early days of OpenAI, and how he ended up helping lead the company through its most chaotic moments. He talks about the breakthrough in reasoning models, the renewed focus on pre-training, the scientific discoveries GPT-5 Pro is already making, and why he thinks fully autonomous AI research is closer than people realize. This is a candid, high-bandwidth conversation with one of the key people shaping the direction of modern AI.
Subscribe to watch more Core Memory videos https://www.youtube.com/@CoreMemoryVideos/videos
Check out our podcasts https://www.youtube.com/@CoreMemoryPodcast
https://www.corememory.com/podcast
Find Ashlee Vance at
www.corememory.com
https://www.instagram.com/ashlee.vance/
https://x.com/ashleevance
Chapters
00:00:00 Intro
00:05:02 How OpenAI Allocates Compute & Prioritizes Research
00:08:36 Ignoring Competitors to Find the Next AI Paradigm
00:14:14 Research vs. Engineering: A False Dichotomy?
00:16:25 Reacting to Google Gemini 3 & The "42" Math Problem
00:20:07 AI vs. Humans: Competitive Coding & The "Move 37" Moment
00:28:42 Will ChatGPT Replace Technical Interviews?
00:35:58 Leaving Wall Street High-Frequency Trading for AI
00:44:26 Joining OpenAI in 2018: Big Vision, Small Team
00:48:08 Inside the OpenAI "Coup": Rallying the Research Team
00:53:46 Brain Drain: Does Losing Top Researchers Matter?
01:00:10 Is Scaling Dead? Supercharging Pre-training
01:09:29 OpenAI for Science: Automating Nobel Prize Discoveries
01:13:43 AGI Timeline: Automating Research Within 2 Years
01:15:35 The Insatiable Demand for Compute & GPUs
01:18:23 Jony Ive & The Future of AI Hardware
01:27:50 Reaction to China's DeepSeek & Open Source AI
01:34:03 AI Alignment: Preventing "Scheming" Models
Mark Chen is one of the most influential people in artificial intelligence, and this episode is a rare look inside his world. As the Chief Research Officer at OpenAI, Chen helps decide which ideas get compute, which research threads scale, and how a 500-person lab navigates constant recruiting raids, model races, and the pressure to invent the next paradigm. We talk about the soup-delivering recruiting wars, how OpenAI thinks about staying ahead of Meta and Google, and what it actually feels like to rank 300 internal research projects when everyone believes theirs is the one that matters.
We also get into Chen’s path from math competitions and high-frequency trading to the early days of OpenAI, and how he ended up helping lead the company through its most chaotic moments. He talks about the breakthrough in reasoning models, the renewed focus on pre-training, the scientific discoveries GPT-5 Pro is already making, and why he thinks fully autonomous AI research is closer than people realize. This is a candid, high-bandwidth conversation with one of the key people shaping the direction of modern AI.
Subscribe to watch more Core Memory videos https://www.youtube.com/@CoreMemoryVideos/videos
Check out our podcasts https://www.youtube.com/@CoreMemoryPodcast
https://www.corememory.com/podcast
Find Ashlee Vance at
www.corememory.com
https://www.instagram.com/ashlee.vance/
https://x.com/ashleevance
Chapters
00:00:00 Intro
00:05:02 How OpenAI Allocates Compute & Prioritizes Research
00:08:36 Ignoring Competitors to Find the Next AI Paradigm
00:14:14 Research vs. Engineering: A False Dichotomy?
00:16:25 Reacting to Google Gemini 3 & The “42” Math Problem
00:20:07 AI vs. Humans: Competitive Coding & The “Move 37” Moment
00:28:42 Will ChatGPT Replace Technical Interviews?
00:35:58 Leaving Wall Street High-Frequency Trading for AI
00:44:26 Joining OpenAI in 2018: Big Vision, Small Team
00:48:08 Inside the OpenAI “Coup”: Rallying the Research Team
00:53:46 Brain Drain: Does Losing Top Researchers Matter?
01:00:10 Is Scaling Dead? Supercharging Pre-training
01:09:29 OpenAI for Science: Automating Nobel Prize Discoveries
01:13:43 AGI Timeline: Automating Research Within 2 Years
01:15:35 The Insatiable Demand for Compute & GPUs
01:18:23 Jony Ive & The Future of AI Hardware
01:27:50 Reaction to China’s DeepSeek & Open Source AI
01:34:03 AI Alignment: Preventing “Scheming” Models
