OpenAI 总裁 Greg Brockman 随著新推出的 GPT-6 Astra 模型发表，宣称人类已正式迈入通用人工智慧（AGI）时代。尽管人类自 Alan Turing 于 70 多年前（超过 70 年）开始探索机器是否具备人类层级智慧以来已历经长期研究，但 Greg Brockman 如今将 AGI 诠释为一种主观的「氛围」或「精神概念」，并交由读者自行认定。他承认目前并不存在众人皆认同的明确时刻，并预估若在几年后回顾，此时此刻及其发表的模型将被视为 AGI 的诞生点。
业界领袖对于界定 AGI 呈现显著分歧。Nvidia 执行长 Jensen Huang 于 5 个月前曾表示多项任务已达到 AGI，但随后在 8 月指出 AGI 仅为 AI 工具生产力的代理指标而非离散门槛，视其为毫无意义的里程碑；Google DeepMind 联合创办人 Demis Hassabis 则认为 AGI 仍需数年时间。此外，微软执行长 Satya Nadella 于 2025 年 2 月将自行宣称达成 AGI 批评为无意义的基准作弊。OpenAI 亦在后续合约重新谈判中，删除了最初与微软协议中若经独立专家评审认证达成 AGI 即可终止收入分成条款的规定。
这项模糊的定调与 Greg Brockman 于 2018 年在国会听证会上的证词形成强烈对比；当时他将 AGI 定义为在开创企业、商业交易、撰写书籍乃至国际争端调解与治理国家等重要任务上超越人类的自主超人系统。他当时亦示警，AGI 迅速引发变革的核心风险，可能使人类在意识到方向前陷入不良环境，并预测后 AGI 世界与当今世界的差距，将远大于今日与 1500 年代（距今超过 500 年）之间的差异，凸显 AGI 不应仅被视为模糊的概念。
OpenAI President Greg Brockman announced that humanity has entered the AGI era alongside the release of the new GPT-6 Astra model. Although humans have explored machine intelligence since Alan Turing began research more than 70 years ago, Brockman now framed AGI as a subjective "vibe" or "spiritual concept," leaving its qualification to individual judgment. Acknowledging the absence of a universally recognized, well-defined threshold due to varying definitions, he projected that looking back in a couple of years, this specific period and model will likely mark when AGI was created.
Industry leaders show sharp division regarding AGI definitions. Nvidia CEO Jensen Huang noted 5 months prior that AGI had been achieved for many tasks, but later in August described it as a "senseless" milestone reflecting AI productivity rather than a discrete threshold; Google DeepMind co-founder Demis Hassabis suggested AGI remains years away. Furthermore, Microsoft CEO Satya Nadella dismissed self-proclaimed AGI milestones as nonsensical benchmark hacking in February 2025, and renegotiations subsequently eliminated an original contractual clause that would exit revenue-sharing terms upon verified AGI achievement via independent expert review.
This ambiguous framing sharply contrasts with Brockman's 2018 congressional testimony, where he defined AGI as an autonomous, superhuman system outperforming humans across vital tasks, including founding companies, negotiating deals, authoring books, international dispute mediation, and state governance. He warned of core dangers stemming from rapid transformation that could place humanity in undesirable conditions before realizing it, predicting that the post-AGI world would differ more from today than today diverges from the 1500s (over 500 years ago), establishing that AGI was never intended as a vague concept.