Jason Wei

Overview

AI researcher and co-creator of chain-of-thought (CoT) prompting — one of the most influential techniques in modern LLM reasoning. Wei et al. (2022) demonstrated that prompting large models with step-by-step reasoning examples significantly improves performance on complex reasoning tasks, establishing a foundational paradigm now used ubiquitously.

X: @_jasonwei

Career

Period	Organization	Role
Current (from mid-2026)	Meta AI (Meta Superintelligence Labs / Meta SI)	Research Scientist
~2022–2026	OpenAI	Research Scientist
Earlier	Google Brain	Research Scientist
Career note (July 2026): Wei joined Meta Superintelligence Labs (Meta SI) from OpenAI in mid-2026. The move adds a top scaling/reasoning researcher to Meta's frontier research group. (source)

Key Contributions

• Chain-of-Thought Prompting (Wei et al., 2022, Google Brain): showed that eliciting step-by-step reasoning from LLMs dramatically improves multi-step math, commonsense, and symbolic reasoning performance; sparked the "thinking" paradigm that underlies o-series, DeepThink, and all current reasoning models
• Emergent Abilities of Large Language Models (Wei et al., 2022): documented that certain capabilities appear discontinuously as model scale increases — catalyzed the scaling-law debate and informed subsequent training practices
• Research on scaling, instruction finetuning, and RLHF at OpenAI

Public Views

• AI self-improvement will be "gradual — a decade, not a fast takeoff" (X, 2026)
• Growing emphasis on safety research alongside scaling

Related

• Meta AI — current employer
• OpenAI — former employer
• Reasoning Models — CoT is foundational to all reasoning-model work
• Test-Time Compute (Inference-Time Compute Scaling) — CoT enables test-time compute scaling