技术在指数级加速，但岗位机会没有同步增长。 为什么会这样？机会流向了哪里？在这个转折点上，什么样的人能抓住新机会？ 这篇文章，我们从这堂课出发，回答四个问题： 第一节 | 不是岗位少了，是方向变了 2025 年的毕业生，正面对一个规则变了的求职市场。 2025 年 11 月，美国失业率升至 4.6 %，创近四年新高；中国城镇失业率为5.1%，青年失业率（16-24 岁，不含在校生）持续高位。同时，应届毕业生规模创纪录：2025 届 1222 万，2026 届预计 1270 万。 但与过去不同，这次不是岗位总量在减少，而是机会的流向变了。 就在昨天（12月17日），吴恩达一个月前在斯坦福的内部讲座视频才公开。11 月 18 日的 AI 课堂上， 他用两组数据解释了这个变化： 2025年，AI 让写程序变得前所未有地快。但这并不意味着工程师更吃香了，恰恰相反，许多人的工作 变得更容易被替代。 吴恩达在课堂上说了一句话： 模型可以帮你写出正确的代码，但它不会告诉你，这段代码要去解决什么问题。 这句话点出了关键：现在大多数岗位的分工逻辑已经变了。 不是谁更懂技术，而是谁先把问题定义清楚。 以往，一个产品从想法落地，需 ...

Core Insights - A $38 billion deal between OpenAI and AWS is reshaping the AI cloud computing landscape, marking a shift from OpenAI's long-term reliance on Azure to a diversified partnership with AWS [1][6] - Jeff Bezos emphasizes that AI opportunities rely on trial and error rather than predictions, focusing on unchanging customer needs [1][4][12] Group 1: Key Principles from Bezos - The core principle is to build strategies around what does not change, rather than around predictions of change [4] - Long-term decisions should be based on constant customer demands, such as the need for faster and more reliable services [5][10] - The AWS and OpenAI partnership bets on three unchanging factors: the demand for stable computing power, customers wanting to pay for results rather than efficiency, and the importance of system reliability and security [6][7][8] Group 2: Decision-Making and Experimentation - After identifying constant demands, the next step is to experiment quickly, relying on intuition and feedback rather than solely on data [13][16] - Bezos advocates for a trial-and-error approach, where organizations should act quickly and learn from mistakes, as most decisions are reversible [17][18] - The concept of "two-way doors" is introduced, suggesting that most decisions can be revisited, allowing for agile experimentation [18] Group 3: Organizational Adaptation in the AI Era - AI will impact every industry, increasing productivity, but organizations must adapt to these changes [20][25] - Recent layoffs at Amazon, affecting around 14,000 white-collar jobs, are attributed to efficiency improvements rather than AI-induced job losses [22][23] - The ability to quickly adjust and experiment will determine which organizations thrive in the fast-changing landscape, with startups having an advantage over larger, slower organizations [25][26][27] Group 4: Conclusion and Future Outlook - The essence of successful AI projects lies in understanding unchanging needs, engaging in iterative experimentation, and fostering organizational agility [29][30] - Organizations that rely on intuition and quick trials will be better positioned to seize opportunities in the AI era [31][32]