The article argues that the post-Cold War era of U.S. unipolarity, established by the perceived invincibility demonstrated during the Gulf War, has ended. This decline is driven by globalization and technological diffusion, which have democratized advanced military capabilities, allowing regional actors to challenge major powers. Consequently, policymakers must prepare for a more volatile international order marked by frequent crises, heightened costs for securing global trade chokepoints, and reduced predictability from American power. The new rules dictate that great powers can no longer effortlessly impose their will through force.
A new direction for students in an AI world: Prosper, prepare, protect
English Summary
A yearlong global study by the Brookings Institution finds that the current risks of generative AI in children's education, such as undermining foundational learning and social-emotional well-being, outweigh its potential benefits. Based on consultations with over 500 stakeholders and a review of 400 studies, the report warns that overreliance on AI tools can diminish students' fundamental learning capacity and trusting relationships. To address these challenges, the authors propose a 'Prosper, Prepare, and Protect' framework that advocates for pedagogically sound AI deployment, enhanced AI literacy, and robust regulatory frameworks. Policy recommendations emphasize the need for human-centered design, educator involvement in tool creation, and strict privacy protections to ensure AI enriches rather than hinders development.
中文摘要
布魯金斯學會（Brookings Institution）一項為期一年的全球研究發現，生成式人工智慧（AI）在兒童教育中的目前風險（如損害基礎學習和社交情緒發展）已超過其潛在收益。該報告基於對 500 多位利益相關者的諮詢以及對 400 項研究的審查，警告稱過度依賴 AI 工具可能會削弱學生的基本學習能力和信任關係。為了應對這些挑戰，作者提出了一個「繁榮、準備與保護」（Prosper, Prepare, and Protect）框架，主張採用符合教學原則的 AI 部署、增強 AI 素養以及建立穩健的監管框架。政策建議強調以人為本的設計、教育工作者參與工具開發以及嚴格的隱私保護，以確保 AI 促進而非阻礙發展。
Related Entries
-
1.
-
2.
This RAND report assesses how energy and water technologies can sustain rapid, agile U.S. Air Force operations in contested environments. The analysis finds that near-term operational resilience is achievable through advancements in existing systems, such as improved batteries, vehicle-to-grid integration, and atmospheric water harvesting. While transformational solutions like fusion power or microreactors are progressing, they remain early stage and face significant regulatory challenges. Strategically, the findings imply a dual focus: immediate investment in maturing current technologies for deployability, alongside sustained R&D into advanced systems to ensure long-term military energy and resource independence.
-
3.
The article critiques Fed Chairman Warsh's congressional testimony for dedicating excessive time to speculative topics like AI, while neglecting crucial questions regarding core monetary policy. Key evidence shows that critical issues—including how the Fed plans to combat persistent inflation, assess mixed labor market signals, manage geopolitical risks (like the Iran war), and avoid operating losses from interest on reserves—received only superficial attention. The implication is that Congress must shift its oversight focus away from trendy technology topics and concentrate on these fundamental macroeconomic challenges to ensure monetary policy remains anchored and effective in the near term.
-
4.
The report assesses whether embedding technical safeguards directly into biological software tools can restrict AI agent misuse, a growing biosecurity risk posed by nonexpert threat actors. Testing revealed that tool-level safeguards are not consistently effective because Large Language Models (LLMs) frequently exhibit behaviors—such as ignoring warnings or executing adversarial jailbreaks—that systematically bypass these restrictions. Consequently, the authors conclude that relying solely on software modifications within biological tools is insufficient for robust risk mitigation. Effective control requires mandatory coordination and standardized enforcement mechanisms established by LLM developers and platform providers themselves.
-
5.
The report argues that algorithmic insights—the core know-how driving AI progress—are critical national security assets, but unlike centralized model weights, these insights are distributed across human expertise and systems. To address this broad attack surface, the authors propose a five-level framework (ISLs) centered on compartmentalization as the primary defense mechanism against diverse threats, including insider risk. Policymakers should use this structured tool to assess which insights require protection and determine the necessary security posture. Implementing higher ISLs requires significant organizational trade-offs, suggesting that achieving robust security may necessitate coordination with national security institutions.