Unverified50% confidenceOpinionExact time
OpenAI is betting on the transferability of general intelligence to the physical world, and its accumulation in large-scale pretrained models may accelerate this process.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/18/2026
First Seen
Valid until: 1/16/2027
Sources
Related Entities
Related Claims
UnverifiedOpenAI continues to rely on NVIDIA for training while choosing to use its own in-house chips on the inference side.75% similarUnverifiedIn 2017, OpenAI published the paper 'Evolution Strategies as a Scalable Alternative to Reinforcement Learning,' demonstrating that in reinforcement learning scenarios ES can rival the most advanced deep RL methods of the time and naturally supports large-scale parallelism.74% similarUnverifiedOpenAI's research on instruction following and web-based learning systems shows that the training-layer dynamics are far more complex than they appear on the surface, and training data cannot be directly optimized tactically.74% similarUnverifiedOpenAI's training of InstructGPT and Anthropic's training of Claude both rely heavily on human-in-the-loop mechanisms.73% similarUnverifiedOpenAI、Anthropic等公司在预训练之上叠加RLHF，将输出向人类偏好中心对齐73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1057283API
curl https://kongchang.com/api/v1/knowledge/claims/1057283MCP
get_claim(id=1057283)