商学院正日益超越基本的 AI 素养，转而教导高阶主管如何随著 AI 能力变化，与 AI 一起、对抗 AI、以及绕开 AI 来做决策。文章以布达佩斯的 Uniqa Insurance Hungary 为开端，其 NiQA 系统可分析照片、解读文件、计算损失，并核准赔付，有时甚至无需人工介入；自上月以来，它已在预先设定的门槛内自主运作，并可在 24 小时内结清理赔。Corvinus-SEED 认为，真正的问题不只是自动化，而是一种重新在人与 AI 之间分配决策权、并配套新控制与新结构的管理模式。MIT Sloan Management Review 的研究也呼应了这个更广泛的挑战，指出人类与 AI 结合后可能比单独任一方表现更好，因此，高阶主管何时信任、质疑或否决 AI 的判断，已成为商业教育的核心。
不同学校正以实务化、聚焦决策的形式回应。位于 Fontainebleau 的 Insead 使用具 AI 功能的模拟，让高阶主管在不确定情况下结合判断与机器输入；其目标是强化而非取代人类决策，参与者有时还会带著更大的策略抱负离开，例如一个 100-day 行动计划，用来重新思考商业模式。在 HEC Paris，AItelier 平台协助高阶主管把商业问题转化为 AI 使用案例；其中一位 CFO 想要的是一个即时财务风险助理，而不是每月一次的报告周期，但随之而来的却是关于资料所有权、资料品质与责任归属的管理问题。Essec Business School 强调部署之后会发生什么，提醒模型会随情势变化而退化，重新训练或扩大规模时必须在技术与商业因素之间取得平衡；其目标是培养能流畅运用 AI 的领导者，让他们能在技术团队与商业团队之间进行转译、质疑假设，并在必要时提出反对。
文章也强调，AI 不只是提供输出，还会以细微方式塑造判断；MIT Sloan Management Review 最近的研究警告，生成式 AI 可充当说服引擎。UPF Barcelona School of Management 透过工作坊模拟高风险决策，例如拒绝求职者、拒绝贷款，或重新分配医疗资源，迫使高阶主管亲自签核，并面对一个事实：最终负责的不是机器，而是人。在高阶主管教育中，AI 正越来越直接地嵌入学习流程：Insead 使用 Lexarius 等工具进行即时回馈与自适应内容，Polimi Graduate School of Management 在虚拟环境中使用 AI 驱动的化身，Esade 使用生成式工具进行黑客松式练习，而 Essca 正在试验 AI 支援的教练服务。这些做法的共同意涵是，领导者不仅必须理解 AI 是否准确，也必须理解它在实务上如何改变组织的决策权、监督与责任。
Business schools are increasingly moving beyond basic AI literacy and toward teaching executives how to make decisions with, against, and around AI as its capabilities shift. The article opens with Uniqa Insurance Hungary in Budapest, where its NiQA system can analyse photos, interpret documents, calculate losses, and authorise payouts, sometimes without human intervention; since last month it has been operating autonomously up to a predefined threshold, and it can settle claims within 24 hours. Corvinus-SEED says the real issue is not automation alone but a new management model that reassigns decision rights between humans and AI, with new controls and structures. This broader challenge is echoed by MIT Sloan Management Review research suggesting humans and AI together can outperform either alone, making executive judgment about when to trust, question, or override AI central to business education.
Different schools are responding with practical, decision-focused formats. Insead in Fontainebleau uses AI-enabled simulations where executives combine judgment with machine inputs under uncertainty; the aim is to sharpen, not replace, human decision-making, and participants sometimes leave with larger strategic ambitions such as a 100-day action plan to rethink a business model. At HEC Paris, the AItelier platform helps executives turn business problems into AI use cases, illustrated by a CFO seeking a real-time financial-risk assistant instead of monthly reporting cycles, only to confront managerial questions about data ownership, data quality, and accountability. Essec Business School emphasizes what happens after deployment, warning that models degrade as conditions change and that retraining or scaling must balance technical and commercial factors; its goal is to produce AI-fluent leaders who can translate between technical and business teams, question assumptions, and push back when needed.
The article also stresses that AI can subtly shape judgment, not just provide outputs, with recent MIT Sloan Management Review research warning that generative AI can function as a persuasion engine. UPF Barcelona School of Management uses workshops that simulate high-stakes decisions such as rejecting a job candidate, denying a loan, or reallocating medical resources, forcing executives to sign off and confront the fact that someone, not the machine, is ultimately responsible. Across executive education, AI is increasingly embedded directly into learning: Insead uses tools such as Lexarius for real-time feedback and adaptive content, Polimi Graduate School of Management uses AI-powered avatars in virtual environments, Esade uses hackathon-style exercises with generative tools, and Essca is experimenting with AI-supported coaching. The common implication is that leaders must understand not only whether AI is accurate, but also how it changes organizational decision rights, oversight, and responsibility in practice.