前OpenAI员工Andrew Ho离职创办公司，专注为科学推理生产高质量强化学习数据集
英文摘要
Andrew Ho announced his last day at OpenAI after eight months, revealing he is starting a new company focused on producing high-quality reinforcement learning datasets. He argues that LLMs have poor generalization and that economically productive capabilities are underrepresented in existing data, predicting that frontier labs will need to spend over $100B on targeted data acquisition. The company's first products will target biology and statistical reasoning: long-horizon scientific reasoning datasets based on GeneBench-Pro, where GPT-5.6 Sol's pass rate is barely above 30%, aiming to push reliability to over 90%; and multimodal datasets covering day-to-day scientific tasks such as analyzing cell culture plates or Western blots. Beyond biology, expansion is planned into chemistry, materials science, healthcare, and white-collar office work.
中文摘要
Andrew Ho宣布在OpenAI任职八个月后离职，并创办一家新公司，专注于生产高质量的强化学习数据集。他指出大型语言模型泛化能力差，现有数据未能很好地呈现经济生产力相关的任务，并预测前沿实验室将需要投入超过1000亿美元进行精准数据采集。公司首批产品聚焦生物学和统计推理：基于GeneBench-Pro的长周期科学推理数据集（GPT-5.6 Sol通过率仅略高于30%），目标将可靠性提升至90%以上；以及覆盖日常科学任务的多模态数据集，如分析细胞培养皿或Western blot图像。后续计划拓展至化学、材料科学、医疗和白领办公领域。
关键要点
Andrew Ho is leaving OpenAI to start a new company after 8 months.
Andrew Ho在入职OpenAI八个月后离职，创办新公司。
The company will produce high-quality RL datasets, starting with biology and statistical reasoning.
新公司将生产高质量强化学习数据集，首期聚焦生物学和统计推理。
Initial products include long-horizon scientific reasoning datasets (GeneBench-Pro) targeting a reliability jump from ~30% to >90%.
首批产品包括基于GeneBench-Pro的长周期科学推理数据集，目标将可靠率从约30%提升至90%以上。
Multimodal datasets for everyday scientific tasks like analyzing cell images and Western blots are also planned.
还将提供多模态数据集，用于分析细胞图像和Western blot等日常科学任务。
Ho predicts frontier labs will need to spend >$100B on targeted data acquisition as scaling alone is insufficient.
Ho预测前沿实验室需投入超千亿美元用于精准数据采集，仅靠规模扩展不够。
Expansion planned into chemistry, materials science, healthcare, and office work.
计划拓展至化学、材料科学、医疗及白领办公领域。