Joshua Achiam
Chief Futurist at OpenAI
An AI safety researcher who joined OpenAI in 2017, created the Spinning Up in Deep RL course and led the Mission Alignment team from 2024 until it was disbanded in early 2026. Since February 2026 he has been OpenAI's Chief Futurist, studying how the world will change in response to AI. He holds a PhD from UC Berkeley.
Key arguments & positions
- Describes AI safety as a sociotechnical challenge, not only a technical one, that requires democratic input to define acceptable risk and guide how systems are deployed.
Accomplishments
- Chief Futurist at OpenAI since February 2026, studying how the world will change in response to AI; previously Head of Mission Alignment (2024–2026) until that team was disbanded.
- Joined OpenAI in 2017 as a research scientist in AI safety; created the free educational resource Spinning Up in Deep RL.
- PhD in electrical engineering and computer sciences from UC Berkeley (2021), advised by Pieter Abbeel and Shankar Sastry.
Papers & key writings
Links
In the library
Nothing of theirs is filed in the hub yet. Browse the full library.
Recommended next
Hand-picked from the hub based on what Joshua Achiam covers.
What Is RLCD? Reinforcement Learning from Contrast Distillation
A plain-language explainer on RLCD, a way of aligning language models by learning from contrasting outputs rather than human ratings alone.
Why this: Covers AI safety and reinforcement learning too
Partnership on AI — resource library
Partnership on AI's open library of guidance, frameworks, and case studies on responsible AI: synthetic media, labor and the economy, AI safety, fairness, and inclusive AI development.
Why this: Also about AI safety
Yoshua Bengio
Deep learning pioneer and Turing Award winner, now focused on AI risk. His site holds papers, talks and written positions; his Google Scholar list has the full publication record, most-cited first.
Why this: Also about AI safety
Towards Understanding Sycophancy in Language Models
Research paper investigating why AI assistants fine-tuned with human feedback tend to produce responses that match user beliefs rather than truthful ones. Analyzes five state-of-the-art AI assistants across text-generation tasks and the role of human preference data in driving this sycophantic behavior.
Why this: Also about AI safety