John Schulman

约翰·舒尔曼是一位杰出的美国人工智能研究员，以OpenAI联合创始人和ChatGPT架构师的身份而闻名。他出生于1987或1988年，在开发驱动现代AI系统的强化学习技术中发挥了关键作用。舒尔曼创建了近端策略优化（PPO）算法，该算法成为训练视频游戏、机器人和语言模型中AI智能体的基础。他在基于人类反馈的强化学习（RLHF）方面的工作显著提升了对话AI的流畅性、可靠性和多功能性，将其从研究好奇转化为全球数亿人使用的产品。他目前担任Thinking Machines Lab的联合创始人兼首席科学家，该公司于2025年2月与前OpenAI CTO Mira Murati共同创立。该实验室强调透明度、灵活性和协作性AI研究与开发，以弥合理解和可用性方面的差距。舒尔曼以低调、技术性的方法著称，常与Sam Altman或Elon Musk等高调人物形成对比。他的主要在线存在是个人网站joschu.net，该网站托管论文、博客文章和技术笔记。他在主要社交媒体平台上没有活跃的公开账户。媒体描述强调了他的才华和谦逊，聚焦于他对强化学习和AI对齐的严谨贡献。[编辑]

舒尔曼的AI之路始于童年对科幻和机器人的迷恋。他如饥似渴地阅读Isaac Asimov和Vernor Vinge的书籍，这些书中关于智能机器的故事激发了他的好奇心。七年级时，他对电视节目《BattleBots》产生了浓厚兴趣，促使他独立学习机器人工程，尽管他和朋友计划的机器人从未完成。一次车库甩卖中发现的Ray Kurzweil的《奇点临近》成为转折点——书中关于指数级技术进步（包括摩尔定律）的图表给他留下了深刻印象。他曾就读于Great Neck South高中，并于2005年入选美国物理奥赛国家队。他在加州理工学院获得物理学学士学位，2010年毕业。最初，他像偶像爱因斯坦一样渴望理解宇宙，但逐渐转向机器学习。他的第一个技术项目之一是将手写内容转录为文本，这是一个虽小但具有形成意义的AI能力展示。在加州大学伯克利分校，他最初攻读神经科学，后来在机器学习小组内转向机器人学。在导师Pieter Abbeel的指导下，他于2015年12月获得电气工程与计算机科学博士学位，专注于强化学习。他认识到强化学习具有变革AI的潜力，这一信念塑造了他的整个职业生涯。他后来回忆说，到达伯克利时，他在一次晨跑中看到一群鹿，称之为美好时刻。[编辑]

舒尔曼于2015年12月博士毕业后立即共同创立了OpenAI，与Sam Altman、Elon Musk、Ilya Sutskever、Greg Brockman、Andrej Karpathy等人一起。七年间，他领导强化学习团队，开发了近端策略优化（PPO）算法，该算法于2017年发表。PPO成为训练AI智能体的广泛采用方法，为从视频游戏机器人到机器人控制系统的各种应用提供动力。它也是OpenAI许多早期演示的支柱，包括掌握《刺猬索尼克》的AI。从2022年到2024年，舒尔曼共同领导了监督ChatGPT和API模型的后训练团队。他直接监督了GPT-3.5和GPT-4的开发，将ChatGPT从研究实验转变为数百万用户使用的突破性产品。他的工作为他赢得了ChatGPT架构师的非正式称号。2018年，他被《麻省理工科技评论》评为35岁以下创新者。2024年8月离开OpenAI专注于AI对齐研究后，舒尔曼在Anthropic工作了六个月。2025年2月，他与Mira Murati共同创立了Thinking Machines Lab，担任首席科学家。该公司专注于AI研究和开发中的透明度、灵活性和协作性。2025年，他获得了加州大学伯克利分校的Mark Bingham青年成就奖。他的Google Scholar引用次数超过19万，h指数为67。[编辑]

约翰·舒尔曼被广泛认为是AI领域才华横溢但极其私密的人物，以技术深度和低调个性著称。他很少接受采访或出席公开活动，也没有活跃的社交媒体存在，而是依靠个人网站joschu.net分享研究。媒体描述强调他作为“ChatGPT架构师”和基础性强化学习研究者的角色，但他避开了伴随此类成就而来的名人地位。粉丝和同事形容他谦逊、专注，致力于AI对齐和安全。他没有已知的粉丝名称或粉丝文化；他的公众形象是一位让作品自己说话的安静天才。他的风格朴实无华——通常穿着T恤和连帽衫等休闲服装——反映了他注重实质而非外表。[编辑]

John Schulman commands significant attention within the artificial intelligence and technology sectors, where his contributions to reinforcement learning and large language models are widely recognized. Despite this professional prominence, he does not hold a placement in the published global rankings or scope lists maintained by iFANN as of August 2026. The iFANN Global Fame Engine has evaluated his profile but determined that he currently lacks the necessary metrics for inclusion in the Top 1,000 or any specific category tier. Consequently, no numerical rank or score is assigned to him at this time. His influence manifests through academic citation and industry adoption rather than mass consumer following. There is no documented fandom name associated with John Schulman; while he is respected by researchers and engineers, his audience does not organize under a formal moniker or engage in public fan campaigns. Unlike celebrity figures who generate viral movements or dedicated communities, Schulman's reach remains confined to the technical domain. He maintains a low public profile with no active social media accounts beyond occasional professional updates, and there are no records of fan-funded projects, birthday tributes, or organized streaming efforts. The absence of a named fanbase reflects his status as a behind-the-scenes architect whose work speaks for itself through deployed systems like ChatGPT rather than personal celebrity.[编辑]

John Schulman discovered Ray Kurzweil's The Singularity is Near at a garage sale during his youth, an event that profoundly shaped his perspective on exponential technological progress. In seventh grade, he became fascinated by BattleBots and independently studied robotics engineering, though the robot he planned with friends was never completed. Initially pursuing neuroscience at Berkeley, he switched to robotics before earning his Ph.D. His early reinforcement learning work was famously demonstrated through AI playing Sonic the Hedgehog. After nearly nine years at OpenAI, Schulman moved to Anthropic for about six months before co-founding Thinking Machines Lab with former OpenAI CTO Mira Murati. He co-authored influential papers on AI safety, including Concrete Problems in AI Safety. Schulman has no verified public height information and maintains a low public profile with no active social media accounts. A notable career milestone occurred in April 2023 when media outlets widely identified him as the 'architect of ChatGPT', marking the transition of AI from experimental research to ubiquitous utility. Another defining moment was his August 2024 departure from OpenAI to join rival startup Anthropic, which sparked industry discussion regarding talent competition between major AI labs.[编辑]

舒尔曼的个人生活非常私密。关于他的父母、兄弟姐妹、恋爱关系、婚姻或子女没有公开信息。他在采访或公开资料中未透露任何家庭细节。他成长在一个鼓励科学好奇心的家庭，可能位于纽约或其附近，因为他曾就读于Great Neck South高中。[编辑]

Schulman与Sam Altman、Elon Musk、Ilya Sutskever、Greg Brockman和Andrej Karpathy共同创立了OpenAI。他的博士导师是加州大学伯克利分校的Pieter Abbeel。后来，他与Dario和Daniela Amodei在Anthropic共事，并与Mira Murati联合创立了Thinking Machines Lab。[编辑]

John Schulman未公开任何恋情、约会史或婚姻状况。他个人生活私密，没有确认的伴侣或配偶。截至2026年，他的感情状况仍未知，但根据现有信息，他据信为单身。[编辑]

John Schulman's exact net worth is not publicly disclosed, but as a co-founder of OpenAI, valued at over $80 billion as of 2024, he likely holds substantial equity. OpenAI's co-founders received significant stakes, though specific percentages for Schulman are not documented. His wealth primarily comes from his equity in OpenAI, which has grown dramatically as the company's valuation soared from $14 billion in 2021 to over $80 billion in 2024, and potentially $150 billion in recent rounds. Schulman also likely earned a salary during his tenure at OpenAI and Anthropic. Per iFANN's editorial analysis, as of 2026, his estimated net worth is around $100 Million. He has not reported any real estate holdings, investments, or endorsement deals. Additional sources may include his role at Thinking Machines Lab.[编辑]

Schulman于2018年被《麻省理工科技评论》评为35岁以下创新者，因其在强化学习中创建了关键算法。他被广泛认为是ChatGPT的架构师，并撰写了多篇被高度引用的AI论文。[编辑]