【徐瑾】在探讨人工智能对经济的影响时，John Calhoun 于1972年进行的“25号宇宙”实验经常被引用，以说明绝对富足的潜在危险。该实验在资源丰富的环境中从4对小鼠（8只个体）开始，最终达到2200只的种群峰值。然而，极度拥挤导致了社会崩溃；在第600天左右，新生幼鼠的存活率骤降至0%，引发了以严重行为异常为特征的负增长阶段。该种群最终在第1780天左右面临彻底灭绝，这引发了当代的担忧，即由人工智能驱动的完全物质富足可能会消除生物适应力并导致社会衰退。

尽管“25号宇宙”的叙事引人注目，但将其直接应用于人类人工智能经济学在科学上是有缺陷的，它将高密度压力与资源富足混为一谈，并忽略了关键的跨物种变量。在科技领域，Elon Musk 预测在10到20年内，人工智能和机器人技术将产生近乎无限的智能和劳动力，使得人类就业成为可选项，货币变得过时。相反，经济学家强调，这种预期的技术富足并不能解决分配和控制的结构性权力动态，这反映了 Karl Marx 对阶级结构的观察；人工智能资源将不可避免地受到地位、财富和系统所有权方面现有差异的支配。

与实验小鼠不同，人类将无聊作为寻找意义和重组社会的催化剂，正如在以往历史性的生产力飞跃中所证明的那样。此外，绝对富足在社会经济上是不可能的，因为位置性商品——例如排他性的教育机会或优先的算法延迟——完全依赖于相对稀缺性和排名系统。因此，未来的人工智能经济的运作方式可能不像一个技术乌托邦，而更像中世纪晚期的欧洲，在那里，行会和修道院等分散的、相互竞争的实体提供了针对中央集权的结构性制衡，在日益优化的世界中确保了系统的稳健性和个体自由。

In discussions regarding the economic impact of artificial intelligence, John Calhoun’s 1972 "Universe 25" experiment is frequently cited to illustrate the potential dangers of absolute abundance. The experiment began with 4 pairs of mice (8 individuals) in a resource-rich environment, eventually reaching a peak population of 2,200. However, extreme density led to societal collapse; by approximately day 600, the newborn survival rate plummeted to 0%, initiating a negative growth phase characterized by severe behavioral abnormalities. The population ultimately faced total extinction around day 1,780, prompting contemporary fears that complete material abundance driven by AI could eliminate biological resilience and lead to societal decay.

Despite the compelling narrative of Universe 25, its direct application to human AI economics is scientifically flawed, conflating high-density stress with resource abundance and ignoring critical interspecies variables. In the technological sector, Elon Musk projects that within 10 to 20 years, AI and robotics will generate near-infinite intelligence and labor, rendering human employment optional and currency obsolete. Conversely, economists highlight that this projected technological abundance does not resolve the structural power dynamics of distribution and control, mirroring Karl Marx's observations on class structures; AI resources will inevitably be governed by existing disparities in status, wealth, and systemic ownership.

Unlike the experimental mice, humans leverage boredom as a catalyst to search for meaning and reorganize society, as demonstrated during previous historical leaps in productivity. Furthermore, absolute abundance is a socioeconomic impossibility because positional goods—such as exclusive educational access or priority algorithmic latency—rely entirely on relative scarcity and ranking systems. Consequently, the future AI economy may function less like a technological utopia and more like late medieval Europe, where fragmented, competing entities like guilds and monasteries provided structural checks and balances against centralized power, ensuring systemic robustness and individual freedom in an increasingly optimized world.