University Leaders Warn Higher Ed Faces Extinction From AI-Driven Tools

高校领导层警告：AI工具或使高等教育面临消亡危机 除非教职人员提升技能并拥抱个性化AI智能体网络，否则高等教育机构恐将被触手可及的移动学习替代方案所淘汰。 专家组探讨了人工智能如何从根本上重塑教育格局，指出高校需重新聚焦批判性思维、创造力、团队协作及AI管理等核心通识技能。与会嘉宾讨论了院校不作为带来的严峻风险，以及数字原生代学生与传统教师之间巨大的技术采纳差距。此外，讨论还强调高校必须抵御AI造假以维护科研诚信，利用多智能体LLM架构进行高性价比的内容开发，并为以个性化一对一数字辅导和加速科学发现为特征的未来做好准备。 高校在 AI 时代必须聚焦培养四大核心文理通识技能：批判性思维、宏观创造力、团队协作能力，以及与人类员工协同管理 AI 智能体的全新领导力。 (17:46) 高等教育面临的最大威胁并非 AI 采纳过快，而是体制迟缓与教师的不作为，导致学校在面对快速适应 AI 的学生群体时失去竞争力。 (19:36) 随着深度伪造与虚假学术研究泛滥，真实性的成本愈发高昂，研究型大学将作为社会权威的客观事实验证者发挥关键作用。 (24:26) StrongMind 等机构利用多智能体网络与复杂的思维链（CoT）提示词构建课程开发平台，大幅降低了数字课程的高昂制作成本并减少幻觉。 (33:47) 专家预计未来 5 到 10 年内，深度个性化学习系统、自主生成式一对一数字导师以及结合跨学科数据的科研加速将彻底重塑教育生态。 (40:29)

Academic leaders and educational technology developers argue that generative and agentic AI requires higher education to pivot toward higher-order critical reasoning, human taste, and rigorous evaluation, while warning that institutional inertia risks leaving universities displaced by personalized commercial platforms. Re-centering Liberal Arts on Human Discretion (17:46) Betsy Sinclair argues that basic algorithmic coding exercises such as 'FizzBuzz' have become obsolete benchmarks for student ability, necessitating a curricular shift toward four foundational competencies: critical thinking, creativity, collaborative interpersonal skills, and managing teams composed of both humans and error-prone AI agents. Sinclair contends that technological ubiquity does not diminish higher education, but instead reinforces the traditional core of a liberal arts education. As technical generation becomes trivialized—such as teenagers possessing the capability to create feature-length films using generative tools—Sinclair forecasts that the primary 'currency of the realm' will shift away from mechanical production and toward human taste, storytelling, and the audacity to conceptualize new ideas. Generational Divide and the Threat of Substitutes (19:36) Sinclair warns that because universities plan across centuries, their institutional inertia presents a severe hazard; failing to actively upskill faculty and students will leave legacy academies uncompetitive. Anjelina Belakovskaia highlights an acute cultural divide: an internal survey at Thunderbird School of Global Management found that typical universities saw only 20% faculty adoption of AI (with 80% resisting or waiting it out), while 80% to 90% of students engaged with AI daily. Belakovskaia cautions that students who treat AI simply as a homework automation engine fail to cultivate real skills, jeopardizing their future economic viability. Applying Michael Porter's Five Forces framework, Damian Creamer warns of the 'threat of substitutes' facing legacy education models. Creamer cites former Google CEO Eric Schmidt's prediction on Peter Diamandis's Moonshots podcast that the next multi-trillion-dollar tech giant will be a personalized learning platform delivered via smartphone, asserting that traditional institutions failing to adapt risk outright obsolescence. Institutional Truth Validation and Long-Horizon Safeguards (24:26) Amid the proliferation of deepfakes and fraudulent grant submissions to agencies like the NIH, Sinclair observes that 'truth has become really expensive,' positioning universities as essential, nonpartisan ground-truth validators. Belakovskaia contrasts the academic research environment with private industry: while commercial developers operate under short-term quarterly ROI pressures, universities maintain the mandate to pursue speculative, long-horizon R&D cycles. However, accelerating analytical speed poses distinct risks. Belakovskaia notes that while AI models can synthesize complex qualitative datasets in seconds that previously required doctoral candidates eight years of manual coding, high-stakes fields cannot tolerate probabilistic errors. She questions whether society would accept a doctor or skyscraper architect with a 1% error margin, emphasizing that human oversight remains mandatory. To mitigate these risks, Sinclair invites industry leaders to collaborate with university researchers to stress-test emerging beta tools within sandbox environments. Agentic Architectures and Scaled Educational Delivery (33:47) Creamer details how StrongMind developed 'Course Builder,' an LLM-agnostic system utilizing complex Chain-of-Thought and Tree-of-Thought prompting alongside multi-agent networks that automatically score content against rubrics. This infrastructure addresses historic educational production bottlenecks, dramatically lowering the $500,000 to over $1 million cost traditionally required to develop digital courses like Algebra 1 while minimizing hallucinations. In instructional delivery, Belakovskaia describes how ASU and Thunderbird deploy AI-generated avatars, virtual reality, and automated pipelines—such as prompting ChatGPT, employing Manus AI for agentic workflows, and generating audio via NotebookLM—to scale educational reach across global cohorts. She notes that initiatives such as Thunderbird's 100 Million Learners rely on AI translation across 40 languages to expand educational access worldwide. While panelists envision autonomous generative ecosystems acting as individualized tutors that deepen student engagement and accelerate scientific discovery, they emphasize that technology must supplement rather than displace human educators. The unresolved challenge remains whether slowly moving academic institutions can overcome internal resistance quickly enough to guide AI deployment responsibly before external substitutes outpace them.

本场专题讨论指出，人工智能的迅猛发展正在颠覆传统教育范式，大学的立校之本正从基础技能训练转向批判性思维、人际协作与智能体协同等核心博雅素养。与会学者与产业专家强调，高校面临的最大危机是行动迟缓与外部智能化替代品冲击，未来各方需依托多智能体工作流重塑课程开发、捍卫学术真实性，并借助个性化辅导体系推进全球教育公平。 核心素养重构与博雅教育价值回归 (17:46) Betsy Sinclair 指出，在 AI 已能自动化完成诸如“FizzBuzz”等简单编程代码的背景下，高校教学必须全面转向培养四项核心能力：展现深度逻辑推理的批判性思维、敢于提出颠覆性构想的创新与抱负、团队协作与人际沟通能力，以及协同管理偶发错误的 AI 智能体的新型领导力。Sinclair 认为，这场技术转型并未削弱高等教育的权威，反而再次凸显了博雅教育的根本价值，因为对复杂问题的审慎探究与创意构建是技术无法轻易替代的。 随着自动化生成工具普及，普通人创作长篇影视或复杂内容的技术门槛大幅降低，Sinclair 预测未来的竞争重心将转向个人的审美鉴赏力、品味与叙事表达。与此同时，学生如果过度依赖思维链等推理模型而将思考过程全盘“外包”，仅把 AI 当作完成作业或替代搜索的捷径，将失去构建专业底层逻辑的机会，最终沦为未来经济竞争中的落后者。 高校行动迟缓与代际技术采纳断层 (19:36) 针对学生过度依赖技术的问题，Sinclair 强调高校目前面临的最严峻威胁并非过快引入技术，而是体制上的“行动迟缓”。由于传统大学通常以百年基业的节奏运行，天生具有演进缓慢的特征；然而如果缺乏试错魄力且未能及时提升师生技能，机构将在剧烈变革中丧失竞争力。 Anjelina Belakovskaia 进一步指出了高校内部存在的严重代际脱节：成长于无计算机时代的资深教师正在向数字化原住民学生传授知识。Belakovskaia 引用 Thunderbird 全球管理学院约一年半前的一项内部调查显示，普通高校教师的 AI 采纳率平均仅约 20%（其余 80% 仍处于抵触或观望状态），即便在推广较为积极的 Thunderbird 达到 50% 教师采纳，同期也有 80% 至 90% 的学生已在日常学习中活跃使用 AI。这种应用鸿沟使得传统教学举步维艰，管理层必须为教师提供切实的落地应用案例以弥合认知差距。 替代品威胁与学术真实性维护 (24:26) Damian Creamer 借助迈克尔·波特的五力模型警示，传统教育模式正面临前所未有的“替代品威胁”。Creamer 引述前 Google 首席执行官 Eric Schmidt 在 Peter Diamandis 播客中的预测表示，下一个类似 Google 的万亿美元级巨头很可能是运行在智能手机上的个性化学习设备。传统机构若无法正视这一颠覆性替代品，极有可能面临被边缘化甚至被淘汰的结局。 与此同时，Sinclair 指出当前学术界与社会面临的深层危机在于“获取真理的成本变得极其高昂”。随着深度伪造技术的泛滥，虚假科研论文甚至伪造的 NIH 基金申请方案频频出现，高校必须承担起非党派权威事实验证者的公共职责。Belakovskaia 补充强调，技术提速决不能以牺牲严谨性为代价；正如医疗诊断或高层建筑设计绝不能接受 1% 的计算错误一样，关键领域的 AI 应用依然必须由人类专家进行绝对监管，不可因追求效率而妥协零容错底线。 智能体辅助研发与产学协同创新 (33:47) 在教学内容生产领域，Creamer 介绍了 StrongMind 借助智能体架构解决高昂开发成本的实践。以往制作一门诸如全美高中不及格率最高的基础代数（Algebra 1）数字课程，开发成本高达 50 万至 100 多万美元。为此，StrongMind 研发了支持多模型接入的内部平台“Course Builder”，通过构建极度复杂的长篇思维链（CoT）与思维树（ToT）提示词工程，并引入多智能体网络依据教学标准量规对练习和评估题目进行自动评分与校验，在大幅压缩开发成本的同时控制了模型幻觉。 在科研创新路径上，Belakovskaia 指出，相较于背负短期季度投资回报率（ROI）压力的科技企业，大学的不可替代性在于能够进行长周期、高容错的基础科学探索。以往博士研究生需花费 8 年手动编码的定性调研数据，如今通过大模型数秒即可提炼；Sinclair 则建议科技行业与高校建立沙盒合作，由企业提供前沿模型测试环境，依托高校学者善于提出深刻探究性问题的独特优势，协同评估 AI 工具的实际产出与社会价值。 自主辅导生态与跨国教育普惠愿景 (37:38) 面向未来 5 至 10 年，与会专家一致认为个性化和全球化将重构教育形态。Belakovskaia 介绍了亚利桑那州立大学（ASU）及 Thunderbird 全球管理学院利用数字人化身、AI 生成播客以及 AR/VR 技术服务跨国生源的经验（该校国际生比例达 95%）。Belakovskaia 分享了其在布拉格参加世界老年国际象棋团体锦标赛期间的个人工作流：借助 ChatGPT、Manus AI 与 NotebookLM 自动化生成多语种课件与播客，在跨时区高强度比赛期间保障了面向中国、乌克兰和美国学生的远程教学。 Creamer 预言，未来五年教育领域将出现能够提供一对一专属辅导的自主生成式学习生态，在弥补师资不足的同时帮助教师更敏锐地洞察学情。Sinclair 亦展望 AI 将大幅加速药物研发、人体微生物组学与气候应对策略等跨学科科研突破。此外，Belakovskaia 提及 Thunderbird 开展的“一亿学习者（100 Million Learners）”计划，通过将免费在线课程译为 40 种语言，借助技术手段消除语言与地理障碍，推动全球教育资源的广泛普及。 本场研讨表明，高等教育不会因 AI 的到来而消亡，但其核心价值重心已从知识单向传授转向高阶思辨与人机协同治理。尽管多智能体系统与个性化辅导展现出破解课程高成本与推动全球普惠的广阔前景，但如何平衡商业化替代品对传统院校的冲击、防范学术伪造风险，并建立起健康的产学研长期验证机制，仍是教育界面临的核心挑战。

## Background 本视频由 Ai4 频道于 2025 年 8 月 29 日发布在 YouTube 平台上。视频描述显示，本场研讨会汇聚了来自南加州大学、亚利桑那州立大学、圣路易斯华盛顿大学及 StrongMind 的专家，共同探讨 AI 时代下教育领域的变革、长期机遇与潜在挑战。 ## TL;DR 研讨会深入探讨了 AI 作为专属智能导师对传统教育模式与科研流程的深刻重塑，强调了 AI 时代重新定义教师角色与培养学生批判性思维的重要性。专家们指出，面对数据隐私、真相验证以及数字鸿沟等严峻挑战，教育机构不能因循守旧。展望未来，多智能体系统与自主生成式学习生态将推动教育的深层个性化与全球普惠，但必须通过产学研的紧密合作来确保 AI 的安全与有效性。 ## Atmosphere & style 现场讨论氛围专业且富有洞察力，主持人与三位嘉宾之间的互动敏捷而流畅，碰撞出诸多思想火花。从国际象棋的博弈思维到真实科研案例的分享，发言者们兼具宏大愿景与落地实践，整体基调既充满对技术突破的兴奋，又保持着对教育本质与人类价值的审慎关照。 ## Key takeaways - [02:35](https://www.youtube.com/watch?v=vinecMZx9bA&t=155s) AI 填补了传统数字教育在“即时指导（Instruction）”上的不足，通过结合学生上下文提供全天候的 AI 学习教练（如 StrongMind 的 Lexi）。 - [02:35](https://www.youtube.com/watch?v=vinecMZx9bA&t=155s) 大学正在利用 AI 工具（如 Cursor）大幅缩短从零到前沿科研的周期，并利用 AI 解决补办出生证明、改善患者护理等社会实际问题。 - [08:33](https://www.youtube.com/watch?v=vinecMZx9bA&t=513s) 在 AI 时代，社会将分为“AI 构建者”、“AI 使用者”与“淘汰者”三类；教育机构需要打破思维局限（The box does not exist），提前制定终局战术。 - [08:33](https://www.youtube.com/watch?v=vinecMZx9bA&t=513s) 高校必须在隐私与合规（如 HIPAA 和 FERPA）的边界内为学生提供安全的 AI 实验环境，鼓励“尝试与失败”。 - [14:36](https://www.youtube.com/watch?v=vinecMZx9bA&t=876s) 尽管 AI 擅长模式识别与算力爆发，但在纯空间逻辑推理（如“骑士巡游”测试）上仍会惨败，人类独特的逻辑、大局观与直觉依然不可替代。 - [17:46](https://www.youtube.com/watch?v=vinecMZx9bA&t=1066s) AI 时代教育的核心不在于基础技能（如 FizzBuzz 编程），而在于重新定义并培养批判性思维、创造力、协作能力以及管理 AI Agent 的领导力。 - [19:36](https://www.youtube.com/watch?v=vinecMZx9bA&t=1176s) 高校面临的最大风险不是学生的“思考外包”，而是学校本身的“因循守旧与不作为”；解决之道在于重新设计评估机制，而非一味禁止工具。 - [21:26](https://www.youtube.com/watch?v=vinecMZx9bA&t=1286s) 教育界存在“教师迟疑”与“学生滥用”的双重断层：绝大多数学生已积极使用 AI 但常误用为作业代写机，而大部分教师采纳率低且持观望态度。 - [24:26](https://www.youtube.com/watch?v=vinecMZx9bA&t=1466s) 在“获取真相成本极高”且学术造假加剧的 Deepfake 时代，大学的核心使命之一是承担起“验证真相（Validate Truth）”的社会基石作用。 - [27:03](https://www.youtube.com/watch?v=vinecMZx9bA&t=1623s) AI 代表了教育行业有史以来最大的“替代品威胁”，未来全新的颠覆性学习模式极有可能诞生于个人移动设备端。 - [29:20](https://www.youtube.com/watch?v=vinecMZx9bA&t=1760s) AI 极大地缩短了科研数据分析周期，但也带来了准确性风险（1% 的失误在关键领域是致命的），并正在隐性塑造人类的日常语言与沟通行为。 - [33:47](https://www.youtube.com/watch?v=vinecMZx9bA&t=2027s) 采用复杂 Prompt 工程与多智能体（Multi-Agentic）评估机制，可显著降低 AI 的幻觉，大幅降低高质量课程内容的开发成本。 - [35:20](https://www.youtube.com/watch?v=vinecMZx9bA&t=2120s) 大学的“超能力”在于学者能够提出关键问题与审视未来，产业界可提供沙盒工具与高校深度开展产学研合作。 - [37:38](https://www.youtube.com/watch?v=vinecMZx9bA&t=2258s) AI 驱动的数字人化身、多媒体与自动化工具链（如 ChatGPT + Manus AI + Notebook LM），赋予了教育者前所未有的标度化与自我扩展能力。 - [40:29](https://www.youtube.com/watch?v=vinecMZx9bA&t=2429s) 当生成式 AI 降低了内容创作门槛后，未来的核心竞争力将转变为品味与审美（Taste）及讲故事的能力。 - [41:51](https://www.youtube.com/watch?v=vinecMZx9bA&t=2511s) 未来五年教育科技将迎来自主生成式学习生态，同时借助 AI 跨国翻译与分发（如雷鸟“1 亿学习者计划”），将大幅缩小数字鸿沟并推动教育普惠。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=vinecMZx9bA&t=4s) 研讨会开场与嘉宾背景介绍 主持人 Nitin Kale 教授（来自南加州大学应用计算与技术专业）向现场观众致以晨间问候，并宣布 AI 时代教育研讨会第二天首场议程正式开始。随后，三位来自高等教育界与教育科技（EdTech）领域的重磅嘉宾依次进行了自我介绍： * **Anjelina Belakovskaia：** 亚利桑那州立大学雷鸟全球管理学院全球金融学副教学教授，同时担任该校 AI 与新兴技术委员会主席。 * **Betsy Sinclair：** 圣路易斯华盛顿大学（WashU）数字转型助理副教务长兼政治学系主任。 * **Damian Creamer：** 数字教育公司 StrongMind 的创始人兼 CEO。 在嘉宾介绍完毕后，主持人 Nitin Kale 补充了一个有趣的幕后细节：在前一天的晚宴上，他得知 Anjelina 实际上是一位**国际象棋特级大师（Chess Grandmaster）**。他打趣道，非常期待 Anjelina 如何将国际象棋、博弈论与 AI 教育融会贯通。随后主持人明确了本次研讨会的规则——探讨 AI 给教育带来的深刻变革，并要求每位嘉宾将单次发言控制在 1 分钟以内，以便覆盖尽可能多的核心议题。 - **主持人/嘉宾履历：** - Nitin Kale（南加州大学教授，主讲工程、数据科学、机器学习与 AI） - Anjelina Belakovskaia（ASU 雷鸟学院副教学教授，国际象棋特级大师） - Betsy Sinclair（华盛顿大学助理副教务长，政治学系主任） - Damian Creamer（StrongMind 创始人兼 CEO，深耕数字教育 25 年） --- ### [02:35](https://www.youtube.com/watch?v=vinecMZx9bA&t=155s) AI 重塑教育：从科研工具到专属智能导师 针对今年全美关于 AI 进课堂、数字公平和隐私保障的热烈讨论，主持人向三位嘉宾抛出了第一个核心问题：AI 究竟在哪些方面为教育带来了不可替代的价值，又在哪些维度构成了挑战？ Betsy Sinclair 率先从高等教育的“价值主张（Value Proposition）”切入。她指出，大众常质疑高校的价值是否在衰退，但 WashU 的实践证明 AI 正让教育变得更具社会影响力： 1. **加速科研前沿探索：** 在过去，教学生掌握 Python 等工具往往需要花费漫长的学期；而今年夏天，WashU 利用 AI 编码工具 Cursor，迅速让所有科研小组上手，极大地缩短了从零到前沿研究的周期。 2. **打破学科壁垒：** AI 正在深层次连接神经科学与 AI 领域，建立起物理学与生物学之间的桥梁。 3. **直击社会痛点服务公众：** 约 9% 的美国公民没有出生证明且补办极度繁琐。WashU 团队开发了 AI 聊天机器人来极速推进该流程，同时利用 AI 改善患者护理。 Damian Creamer 随后从 K-12 数字教育的角度进行了补充。他坦言，自己在数字教育行业摸爬滚打近 30 年，过去最大的痛点是——平台可以轻松提供课程（Curricula）和测评（Assessment），但在“即时教学指导（Instruction）”上一直非常匮乏。而 AI 的突破，在于为学生提供了兼具灵活性与实时性的“全天候学习教练”。StrongMind 打造了一款名为 **Lexi**（意为 Learning Experience）的专有 AI 助手： > "It's not really the tech that's making the change, it's the context. Lexi has the context around the student." ``` +-----------------------------------------------------------------+ | Lexi (AI 学习教练框架) | | | | +--------------------+ +--------------------+ | | | K-12 全美标准课程库 | | 专有教学测评数据 | | | +---------+----------+ +---------+----------+ | | | | | | +-----------+-----------+ | | | | | v | | +---------------------+ | | | 学生个性化上下文 | | | | (Student Context) | | | +----------+----------+ | | | | | v | | +---------------------+ | | | 全天候即时指导与答疑 | | | +---------------------+ | +-----------------------------------------------------------------+ ``` - **数字教育的三大关键支柱：** 课程（Curricula）、测评（Assessment）、指导（Instruction — AI 的介入填补了最后的空白）。 - **全美无出生证明人口比例：** 约 9% 的美国公民缺乏出生证明，可通过 AI 智能客服加速办理。 --- ### [08:33](https://www.youtube.com/watch?v=vinecMZx9bA&t=513s) 国际象棋视野下的 AI 战略：构建者、使用者与淘汰者 凭借独特的国际象棋特级大师视角，Anjelina Belakovskaia 对未来的社会结构提出了深刻而形象的预测。她认为在国际象棋中，下棋前必须精确评估全局，而当前人类正快速迈入一个由三类人构成的新时代：**AI 构建者（AI Builders）、AI 使用者（AI Users），以及淘汰者（Losers）**。 为了避免成为“淘汰者”，教育机构必须采取战略性的棋局思维：不仅要让教授、行政人员和学生全员上车，更要先在脑海中预判出理想的终局（Envision），再通过战术（Tactics）去落地执行。针对“打破常规”的说法，她给出了极富哲理的见解： > "We have to think outside the box. Well, you know what? The box does not exist. The box is only in our mind." 针对 Anjelina 的警醒，Betsy Sinclair 和 Damian Creamer 进一步展开讨论： * **容错与隐私屏障：** Betsy 强调，大学的核心使命是给学生一个可以“尝试并失败”的安全环境。为此，WashU 部署了获得 **HIPAA（医疗隐私）** 和 **FERPA（教育隐私）** 双重安全认证的专用 ChatGPT 实例，目前已有约 50% 的学生日常使用。这既保障了数据不被用于模型训练，又让学生在安全边界内掌握了前沿 AI。 * **产学研深度融合：** Damian 指出，产业界必须与高校紧密合作。高校拥有长线科研的土壤，能够验证 AI 教学的有效性（Efficacy），纠正偏误，确保 AI 真正服务于教育本质。 - **WashU AI 开放平台：** 官方网站 `awashu.edu` 整合了全校教学资源，大力推动“氛围编码（Vibe Coding）”。 - **合规标准：** 遵守 HIPAA（Health Insurance Portability and Accountability Act）与 FERPA（Family Educational Rights and Privacy Act）。 --- ### [14:36](https://www.youtube.com/watch?v=vinecMZx9bA&t=876s) 从国际象棋历史看人机关系与人类独特的直觉 在研讨会Part 1的尾声，主持人将话题再次引回国际象棋，探究象棋高手提升棋艺的过程与 AI 赋能教育之间有何异同。Anjelina 回顾了一段著名的历史：1997 年，IBM 的“深蓝”（Deep Blue）击败了当时的世界冠军加里·卡斯帕罗夫（Garry Kasparov）。那时全社会惊呼“国际象棋要终结了”，但事实证明，人类至今依然热爱下棋，甚至借助计算机大幅提升了训练效率。 有趣的反转是，当今许多顶尖的 AI 大模型，在面对 50 年前基于 Atari 开发的早期国际象棋程序时，依然会被打得溃不成军。Anjelina 剖析道，AI 极度擅长模式识别（Pattern Recognition）和算力爆发，但人类拥有无法被算法替代的**逻辑、大局观与直觉（Intuition）**。 为了验证这一点，Anjelina 分享了她对每一代最新大模型的“终极测试”——**骑士巡游（Knight's Tour）**：要求棋子“马”在 8x8（64 格）的棋盘上走遍每一个格子，且每个格子只能经过一次。 > "GPT-5 came out, I plugged it in. It failed miserably as all the previous versions. I cheered for human, and AI has a very long way to go." - **国际象棋历史节点：** 1997 年 IBM 深蓝（Deep Blue）击败卡斯帕罗夫（Garry Kasparov）。 - **骑士巡游测试（Knight's Tour）：** 8x8 棋盘、64 个方格，要求“马”走遍每格且不重复。包含 GPT-5 在内的最新大模型在此类纯空间逻辑推理题上依旧惨败。 ### [17:46](https://www.youtube.com/watch?v=vinecMZx9bA&t=1066s) 重塑教师角色：AI 时代高校的核心价值 随着讨论从学生学习体验转向教育者的角色变化，南加州大学（USC）的 Nitin Kale 教授向圣路易斯华盛顿大学（WashU）数字转型副副校长 Betsy Sinclair 提问：面对部分人担心 AI 将取代教师、而另一部分人将其视为赋能工具的现状，AI 究竟在如何重塑教育者的职责？教育者又需要具备哪些新的技能与思维方式？ Betsy Sinclair 指出，当今大学毕业生必须掌握四项关键技能，这些技能的内涵在 AI 时代已发生了深刻演变。首先，像“FizzBuzz”这种基础的编程入门测试已不再有必要，大学必须大力推动**批判性思维（Critical Thinking）**，而教育者的挑战在于设计出能让学生充分展现该能力的评估与作业。其次是**创造力（Creativity）**，人类对创新的欲望永### AI 如何重塑教育者的角色与核心技能 讨论从学生与教学体验转向了教育者的未来。面对外界关于 AI 是否会取代教师的担忧，Betsy 提出了不同的见解。她认为 AI 并非要取代教师，而是成为增强教育工作的强大工具。在新时代，学生毕业时需要具备四项被重新定义的核心能力： * **批判性思维（Critical Thinking）：** 类似 FizzBuzz 这样的基础编程测试已不再需要人力解决，重点在于评估学生对复杂问题的分析能力。 * **创新与创造力（Creativity）：** 人类的欲望与追求无止境，未来经济将高度依赖推陈出新的创想，教育需要赋予学生“大胆做梦”的能力。 * **协作与团队沟通（Collaboration）：** 孤军奋战的时代已经过去，社交能力等“软实力”在实际应用中变得至关重要且难以教授。 * **管理与领导力（Management & Leadership）：** 管理对象不仅包括人类下属，还将涵盖频繁出错的 AI Agent（智能体），这催生了全新的管理维度。 Betsy 强调，这些改变实际上将大学推向了其最核心的价值——**博雅教育（Liberal Arts Education）**。虽然教学机制在变化，但大学培养学生批判性思考与领导力的核心价值比以往任何时刻都更加重要。 --- ### [19:36](https://www.youtube.com/watch?v=vinecMZx9bA&t=1176s) 应对“思考外包”与高校的最大的风险：不作为 当被问及“是什么让你夜不能寐”时，Betsy 揭示了目前 AI 在教育领域带来的核心风险：人们担心下一代学习者将思维彻底“外包”给 AI，导致批判性思维能力的退化。然而，对于习惯于按百年尺度规划的大学而言，真正的危机在于**因循守旧、不作为**。 为了在变革中保持竞争力，高校必须敢于承担风险，全面提升教师、员工和学生的 AI 技能。当前的迫切任务在于重新设计**评估机制（Assessment）**，寻找能够让学生展示其实际思考与技能的全新方式，而不是简单禁止工具的使用。 --- ### [21:26](https://www.youtube.com/watch?v=vinecMZx9bA&t=1286s) 两代人的数字鸿沟与学生的滥用误区 Anjelina Belakovskaia 从教学实践的角度揭示了目前教育界存在的两重“断层”现象： ``` [ 教师群体：传统思维 / 迟疑 ] -------- (鸿沟) -------- [ 学生群体：数字原生代 / 广泛应用 ] • 采纳率低（约 20%-50%） • 采纳率高达 80%-90% • 担忧被取代或选择观望 • 常误用为“代做作业”工具 ``` 第一重断层是**世代之争（Battle of Generations）**：生长于无 AI、无电脑时代的教育者，正在努力教授数字原生代（Digital Natives）。尽管像亚利桑那州立大学（ASU）及雷鸟全球管理学院（Thunderbird）走在创新前沿，全美仍有高达 80% 的教师对 AI 持观望态度；相比之下，80% 至 90% 的学生已在积极使用 AI。如果教师不跟上步伐，教学将无法维系。 第二重断层在于**学生的错误使用习惯**：许多学生仅仅将 AI 当作搜索引擎或“作业代写机”。这种敷衍应对的方式不仅无法提升能力，还会让学习者在未来的竞争中沦为被淘汰者。教育者需要重新引导学生，使其出于自身技能提升的目的去负责任、高效地使用 AI。 --- ### [24:26](https://www.youtube.com/watch?v=vinecMZx9bA&t=1466s) 大学的新使命：在“真相变贵”的时代验证真实 随着技术普及，AI 带来的挑战已延伸至学术和科研的根本——**真伪辨别**。Betsy 指出，当今社会“获取真相的成本变得极其昂贵”。学术界已出现向美国国家卫生研究院（NIH）提交虚假申请书、向学术期刊投稿伪造论文等严重现象，社会极度缺乏能够深度审视并验证真实性的专业能力。 大学作为非营利、非偏见的实体，必须承担起**验证真相（Validate Truth）**的社会基石作用。面对日益精密的 Deepfake（深度伪造）技术，高校与研究机构必须加速研发验证工具，捍卫科研与知识的真实性。 --- ### [27:03](https://www.youtube.com/watch?v=vinecMZx9bA&t=1623s) 替代品威胁与教育领域的“下一个 Google” Damian Creamer 借助迈克尔·波特（Michael Porter）的“五力模型”（Five Forces）阐述了 AI 对整个教育产业的颠覆性冲击。他指出，AI 代表了教育行业有史以来最大的**替代品威胁（Threat of Substitutes）**。从 K-12 到高等教育，任何忽视这一替代品威胁的机构都面临着被迅速淘汰的风险。 同时，这其中也孕育着巨大的机遇。在引用前 Google CEO 埃里克·施密特（Eric Schmidt）的观点时，Damian 提到： > “下一个类似 Google 的巨头公司，将诞生于在手机端实现全新学习模式的设备或应用中。” 无论是正规教育还是终身学习，这种基于个人设备的智能化学习载体都将开启极其广阔的创新空间。 --- ### [29:20](https://www.youtube.com/watch?v=vinecMZx9bA&t=1760s) 科研中的机遇、潜在陷阱与人机互动的隐性重塑 在探讨科研优势时，Anjelina 指出，大学之所以能够承担 industry（工业界）无法承担的科研课题，是因为高校不需要追求短期的投资回报率（ROI），容错率更高。AI 极大地缩短了数据分析的周期——以往需要花费 8 年完成的定性数据分析博士论文，如今通过 AI 工具处理只需数秒。 但这种高效也伴随着巨大的安全隐患： * **准确性风险：** 如果 AI 的准确率是 95% 甚至 99%，你敢让这样的医生治病或住在这样的建筑师盖的高楼里吗？1% 的失误在很多领域是致命的。 * **文化与行为塑造：** 人类在塑造 AI 的同时，AI 也在重塑人类的语言和行为习惯。人们习惯于用“命令”而非“礼貌”的态度与 AI 对话，同时在日常表达中频繁套用 AI 生成的词汇（如 delve into 等）。 Anjelina 警示大家，在借助 AI 获得“超级力量”的同时，必须保持人类的核心价值观、个性与安全意识。 --- ### [33:47](https://www.youtube.com/watch?v=vinecMZx9bA&t=2027s) 打造下一代智能课程系统：多智能体与深度提示工程 Damian 分享了 StrongMind 在数字内容开发上的技术突破。过去，团队开发一门《代数 1》（Algebra 1，全美失败率最高的课程之一）需要耗费约 **50 万美元**，累计投入数百万美元尝试破解数字化代数教学的难题。 如今，他们通过开发名为 **Course Builder** 的系统重新定义了课程开发流程： ``` [ 复杂提示工程工作室 ] • 包含 Chain of Thought (思维链) / Chain of Tree (思维树) • 提示词长度堪比《战争与和平》 │ ▼ [ 多模型 / LLM 中立架构 ] │ ▼ [ 多智能体 (Multi-Agentic) 评估体系 ] • 形成性/总结性评估运行于自动化 Rubric 评分标准 ``` 该系统不依赖单一的大语言模型（LLM Agnostic），而是联合数十种模型，利用极长的复杂 Prompt（提示词）进行深度引导，并建立多智能体（Multi-Agentic）机制和评分 Rubric（标准），显著降低了幻觉（Hallucinations）并提升了教学内容的精准度。 ### [35:20](https://www.youtube.com/watch?v=vinecMZx9bA&t=2120s) 多 Agent 系统与产学研合作邀请 Damian 详细阐述了多 Agent 系统的实际应用机制：系统会审视正在开发的每一项教学内容，依据评分量表（Rubric）进行打分，并将其反馈给专用的 AI Agent；这些 Agent 会仔细研究如何进一步完善与提升该内容。他强调这种能力正在呈指数级增长，其影响将极为深远。主持人 Nitin 感谢了 Damian 分享 StrongMind 的智能体实践，随后转向 Betsy，询问大学开发了哪些成果或探索，是值得向产业界与终端用户分享的。 Betsy 借此机会向在场观众提出了合作倡议。她指出，大学的核心优势（或称“超能力”）在于聘请的教授极其擅长提出关键问题——他们不仅是提问，更是通过审视与拷问来探索未来与可能性，而这正是目前的 AI 难以企及的。大学学者具备洞察未来、甚至“隔墙看物”般的远见。针对现场拥有各种前沿工具或处于测试阶段（Beta/Sandbox）的产业界代表， Betsy 诚挚邀请大家与大学学者合作，共同测试工具并探索其产出价值，欢迎有合作意向者在会后直接与她交流。 - **多 Agent 评估机制**：AI Agent 根据既定量表对教学内容进行自动评分与迭代改进。 - **产学研沙盒合作**：邀请产业界提供测试沙盒与前沿 Agent 工具，由大学学者进行严苛测试与价值评估。 --- ### [37:38](https://www.youtube.com/watch?v=vinecMZx9bA&t=2258s) 未来5至10年颠覆教育的 AI 趋势与创新 主持人 Nitin 提出了最后一个开放性问题：展望未来 5 到 10 年，哪些 AI 技术或趋势将最彻底地改变教育？如果 5 年后大家再次同台，最令人兴奋的突破会是什么？Anjelina 首先分享了亚利桑那州立大学（ASU）及雷鸟全球管理学院的探索。她提到目前正在使用数字人化身（Avatar）进行全球教学，同时结合电影、播客、增强现实（AR）与虚拟现实（VR）等技术。 为了说明 AI 带来的扩展能力，Anjelina 分享了自己的亲身经历：作为三届美国女子国际象棋冠军，她今年 2 月代表美国队前往布拉格参加世界国际象棋锦标赛（50岁以上组），担任第一台。当时她白天要在国际赛场比赛，晚上则要为分布在中国、乌克兰和美国的学生授课。面对时差与体力的极限，她利用 ChatGPT 编写提示词驱动 Manus AI，生成脚本后再通过 Notebook LM 制作教学播客。她强调，人类需要睡眠，无法 24 小时不停运转，而 AI 赋予了教育者前所未有的标度化（Scale）与自我扩展能力。 - **数字人化身与多媒体教学**：结合 Avatar、AR/VR、播客与电影制作打造沉浸式课程。 - **Anjelina 的课程重构工作流**：使用 ChatGPT 编写 Prompt 驱动智能体 Manus AI 生成脚本，再导入 Notebook LM 生成个性化教学播客。 - **真实案例**：Anjelina 在布拉格参加世界象棋锦标赛期间，利用 AI 工具链实现了跨越中国、乌克兰和美国的多时区自动化教学支撑。 --- ### [40:29](https://www.youtube.com/watch?v=vinecMZx9bA&t=2429s) 个性化学习、人类创造力与加速科研突破 Betsy 提出了三大最让她感到兴奋的未来趋势。第一是 Damian 所提倡的个性化体验：当个性化学习被深度落实时，学生能够对所学材料建立更深层次、更扎实的理解。第二是 AI 对人类创造力的极大释放。她引用了当天早会儿议的观点——当任何 17 岁青少年都能制作电影时，决定价值的不再是“制作电影的技术”，而是“品味”与讲故事的能力，未来我们将看到更优质的电影、音乐和文学作品。 第三点也是最令科研领域激动的一点：AI 将大幅加快全人类的创新节奏。面对药物研发、人体微生物组探索以及气候变化战略等复杂难题，AI 处理海量数据和跨领域关联的能力，将帮助今天的学生以极快速度解决这些重大的全球性挑战。 - **深层个性化学习**：通过定制化教学帮助学生建立对知识点更 profound（深刻）的理解。 - **创造力门槛降低与“品味”崛起**：生成式 AI 降低了影视与内容创作门槛，创作的核心竞争力转向审美与鉴赏力（Taste）。 - **科研创新加速（Scientific Discovery）**：利用 AI 跨学科数据关联能力，加速药物研发、微生物组学及气候变化解决方案的突破。 --- ### [41:51](https://www.youtube.com/watch?v=vinecMZx9bA&t=2511s) 自主生成式学习生态与教育公平的未来展望 Damian 预判未来五年将是教育科技（EdTech）历史性变革最大的时期。过去教育科技领域常常存在“过度承诺却未能兑现”的状况，导致学校产生倦怠感。而在未来，**自主生成式学习生态系统（Autonomous Generative Learning Environments）**将真正落地，在每个学生身边配备专属的教学辅助智能（Instructional Presence），从而缩小学生与教师之间的鸿沟，并赋予教师更强的教学洞察力，而非取代教师。同时他提醒，技术在解决问题的同时也会带来风险，产业界必须与高等教育深度合作，开展长远研究，确保开发出的不只是另一个软件版本，而是真正服务于人类学习的严肃事业。 最后，Anjelina 补充了关于教育普惠与消除数字鸿沟的展望。她介绍了雷鸟全球管理学院的“1 亿学习者计划”（100 Million Learners Initiative），该计划旨在通过免费课程覆盖全球 1 亿名无法获得传统教育的人群，目前课程已被翻译为 40 种语言。AI 技术将极大降低优质教育资源的门槛，赋能整个社会，带给全人类新的希望。最后主持人 Nitin 感谢了三位嘉宾（Damian、Betsy、Anjelina）的精彩分享与观众的倾听，宣布研讨会圆满结束。 - **自主生成式学习生态系统**：构建陪伴式 AI 教学智能体，实现一对一精准辅导并赋能教师。 - **产学研长周期责任机制**：呼吁企业与高校合作进行长远研究，避免急功近利的产品迭代损害教育质量。 - **雷鸟 1 亿学习者计划（100 Million Learners Initiative）**：利用 AI 翻译与分发技术，将课程翻译为 40 种语言，免费面向全球 1 亿名学习者提供教育支援。 ## Notable quotes > [02:35](https://www.youtube.com/watch?v=vinecMZx9bA&t=155s) It's not really the tech that's making the change, it's the context. Lexi has the context around the student. — Damian Creamer > [08:33](https://www.youtube.com/watch?v=vinecMZx9bA&t=513s) We have to think outside the box. Well, you know what? The box does not exist. The box is only in our mind. — Anjelina Belakovskaia > [14:36](https://www.youtube.com/watch?v=vinecMZx9bA&t=876s) GPT-5 came out, I plugged it in. It failed miserably as all the previous versions. I cheered for human, and AI has a very long way to go. — Anjelina Belakovskaia > [27:03](https://www.youtube.com/watch?v=vinecMZx9bA&t=1623s) 下一个类似 Google 的巨头公司，将诞生于在手机端实现全新学习模式的设备或应用中。 — Damian Creamer ## Mentioned Nitin Kale, Anjelina Belakovskaia, Betsy Sinclair, Damian Creamer, University of Southern California, Arizona State University, Thunderbird School of Global Management, Washington University in St. Louis, StrongMind, Cursor, Lexi, ChatGPT, HIPAA, FERPA, IBM, Deep Blue, Garry Kasparov, Atari, GPT-5, National Institutes of Health, Michael Porter, Eric Schmidt, Google, Course Builder, Manus AI, Notebook LM, 100 Million Learners Initiative ## Video description Nitin Kale, Professor of Technology and Applied Computing Practice, University of Southern California Anjelina Belakovskaia, Associate Teaching Professor, Chair of the AI & Emerging Technologies Committee, Arizona State University, Thunderbird School of Global Management Betsy Sinclair, Assistant Vice Provost of Digital Transformation, Washington University in St. Louis Damian Creamer, Founder & CEO, StrongMind School Solutions Abstracts This session will focus on the transformational impact of AI on the education sector, exploring the long-term opportunities and challenges it presents. Panelists will discuss how AI is reshaping the way students learn, educators teach, and institutions operate, leading to a future of personalized, data-driven education. From enhancing access to learning resources to rethinking the role of educators, AI holds the potential to revolutionize education as we know it. However, this transformation also comes with risks—such as equity concerns, data privacy, and the digital divide. Join this strategic discussion to understand the future of education in the AI era and how institutions can navigate these profound changes. [Source](https://www.youtube.com/watch?v=vinecMZx9bA)

[2:35] Damian Creamer: It's not really the tech that's making the change, it's the context. Lexi has the context around the student. [8:33] Anjelina Belakovskaia: We have to think outside the box. Well, you know what? The box does not exist. The box is only in our mind. [14:36] Anjelina Belakovskaia: GPT-5 came out, I plugged it in. It failed miserably as all the previous versions. I cheered for human, and AI has a very long way to go. [27:03] Damian Creamer: 下一个类似 Google 的巨头公司，将诞生于在手机端实现全新学习模式的设备或应用中。

[17:46] Betsy Sinclair: Human greed knows no limit. And so we're going to see tremendous potential for innovation here for people to come up with new things and new ideas that people want... We need to give our students the ability to dream really big. [24:26] Betsy Sinclair: truth has become really expensive.

[0:04] [Applause] [0:07] All right, good morning and thank you [0:09] for the introduction. Um, welcome [0:12] everyone to our day two uh first uh [0:15] session. Those who are interested in [0:16] education in the age of AI uh my name is [0:20] uh Nathan Khal as was introduced I'm a [0:22] professor of technology and applied [0:25] computing at the University of Southern [0:27] California and Los Angeles. Um I'm the [0:30] moderator for this panel today and uh [0:33] joining me are uh uh three um academics [0:37] and those who are in the edtech industry [0:39] and who will do a short introduction of [0:41] themselves please. [0:43] >> Uh good morning everyone. My name is [0:45] Angelina Belikovska. I'm an associate [0:47] teaching professor in global finance at [0:50] Thunderber School of Global Management [0:52] Arizona State University and I'm also a [0:54] chair of the AI and emerging [0:56] technologies committee. [0:58] Good morning. Good morning. I'm Betsy [1:01] Sinclair. I'm an assistant vice provost [1:03] of digital transformation and the [1:05] department chair of political science at [1:07] Washington University in St. Louis. [1:10] >> Good morning everyone. My name is Damen [1:12] Kramer. I'm the chief executive officer [1:14] and founder of Strongmind, a digital [1:16] learning company. For the last 25 years, [1:19] we've been reimagining what education [1:21] looks like through technology. And the [1:23] big breakthrough at this point in time [1:25] is being able to uh put an AI uh [1:30] everpresent [1:32] learning coach next to every single [1:33] student. So that's the big breakthrough [1:35] and we're really excited to be here [1:36] today. [1:37] >> Thank you. And I'd like to add I had [1:39] dinner with Angelina last night and I [1:41] learned she's a chess grandmaster. [1:44] Um so she's going to use somehow her uh [1:47] chess and games and game theory and AI [1:50] and education um al together. So I'll [1:53] follow up with the question. Okay. Uh so [1:56] the format of this uh panel will be I [1:58] have prepared some questions as you can [2:00] see here um about uh AI and uh the [2:04] future of education in this age and I [2:07] moderate a meaningful insightful and [2:09] respectful discussion and if time [2:12] permits uh only then we'll have some [2:14] audience Q&A but welcome to the audience [2:16] for your attention today. Um and to the [2:19] panelists uh u I'll post questions and [2:22] we'll go back and forth uh and maybe [2:24] keep about your responses to be on less [2:28] than a minute. We have a lot of uh [2:30] exciting uh topics today. Okay. Uh so we [2:35] have a lot of academics or those [2:36] interested in um the future of uh of uh [2:39] teaching and academia. Uh so this year [2:42] has been uh the headlines have been full [2:44] of debates about AI and education. I'm a [2:48] professor. I teach and I also teach in [2:51] AI. I teach in engineering, data [2:52] science, uh machine learning and AI. Um [2:56] from generative AI tools like chat GPT [2:58] in the classroom to questions about [3:00] digital equity and privacy. [3:02] So how are these trends influencing your [3:05] thinking uh about where AI can generally [3:08] add value or where it might add some [3:11] challenges for your students, for [3:13] educators and institutions? Angelina [3:17] maybe we can start with you. [3:18] >> Sure absolutely. So uh what's [3:21] interesting uh you know we we have to [3:23] look at the situation and uh try to [3:26] understand it first. This is what we do [3:28] in chess. Yes. You know like before you [3:29] decide to make the moves you need to [3:31] understand what what you have in the [3:33] position and well here is a story. We [3:36] live in very dynamic times. So there's a [3:38] lot of changes and we have to be uh [3:40] aware of it and we have to be ready for [3:42] it. And you you you heard everybody [3:44] heard the expression uh we have to think [3:46] outside the box. Well, you know what we [3:48] have to think the box does not exist. So [3:51] the the box is only in our mind. It [3:54] restricts us. And [3:56] if we have this approach, we will [3:59] realize that AI opens some unbelievable [4:02] opportunities something that we don't [4:04] even aware of uh yet. And [4:08] you know the more the more we uh we act [4:11] the more we take ownership uh so we uh [4:14] we will build the future instead of [4:16] letting somebody else build the future. [4:18] >> Okay. Uh yes. So Betsy would you like [4:21] to? [4:22] >> Yes. So I think um one of the concerns [4:25] that people often have with higher [4:26] education is that our value proposition [4:28] has changed. And I just want to give [4:31] three examples of where it has changed [4:33] and we're actually really delivering I [4:35] think more for our students but also for [4:37] our communities in terms of that value [4:38] proposition. Um the first is just to [4:41] think about how we're really committed [4:43] to building you know pretty deep [4:45] responsible and human- centered AI. One [4:47] of the things we've been doing with our [4:48] students has been thinking about how we [4:50] can build collaborative teams to get [4:52] them to the frontier of research and AI [4:55] has made that possible. So, one of the [4:56] things we did this summer is we ran a [4:58] summer core where we got every team up [5:01] up on cursor really quickly. And in [5:03] olden times, it would have taken me a [5:04] long time to get every team up to speed [5:06] on Python. And now we have some AI [5:08] mobilized tools that speed up that [5:10] process where they can really get to the [5:12] frontier of research much more much more [5:13] quickly. It's pretty exciting. Um, it [5:16] also has really let us connect [5:18] disciplines. So one of the pretty [5:20] extraordinary projects that we've been [5:21] working on at WashU has been connecting [5:23] the fields of AI and neuroscience and [5:26] this has really established a pretty [5:27] deep relationship between physics and [5:30] biology. [5:31] The ability to have conversations across [5:33] those spaces and understand that [5:34] literacy. It comes from this [5:36] appreciation that the AI that we're [5:38] leveraging is really grounded in this [5:40] need to do like responsible human- [5:43] centered project-based AI. But I also [5:46] think we're really delivering services [5:48] in a different way and studying them. [5:50] So, you know, about 9% of all American [5:53] citizens don't have a birth certificate. [5:54] It's pretty amazing if you think about [5:56] it. And it's hard to get one. You've [5:58] probably all had to think about how is [5:59] it I'm going to get a copy of my birth [6:01] certificate. Well, here's an easy way to [6:03] do it. We can build a chatbot to enable [6:05] that to speed up and go really fast. Um, [6:08] it's also the case that we've really [6:09] built a number of AI tools to transform [6:11] patient care. And so I think thinking [6:13] about the research enterprise as [6:15] connecting to that student experience is [6:16] one of the ways that we're really [6:17] delivering on that value proposition for [6:19] higher education. [6:21] Thank you Betsy on your perspective on [6:23] uh AI and research. So Damian um you're [6:26] in the industry for providing uh content [6:29] and uh for education u building [6:33] curricular and digital assets. Um, what [6:36] are your thoughts about [6:38] >> what are your thoughts about education [6:40] in the age of AI? [6:43] >> Yeah, that's a great question. I think [6:45] education is definitely changing right [6:46] now, faster than I've ever seen it [6:48] change. I've been involved in digital [6:50] learning for the last 30 years. Um, one [6:54] of the things that in digital learning [6:55] that we are able to do is we're able to [6:57] provide the curricula and we're able to [6:59] provide the assessment, but the tools [7:02] around instruction have always been [7:03] waning. And that has been a uh challenge [7:06] in digital learning until now. And so [7:09] what I see with artificial intelligence [7:11] is that we're able to provide AI [7:14] instruction along with the curricula and [7:16] the and the assessment. And that allows [7:19] for the students who are choosing that [7:21] as their option for learning. One of the [7:23] biggest reasons they choose that is [7:25] because of the flexibility. But one of [7:27] the problems also is then because you [7:28] have all of this flexibility, you're not [7:31] connecting with teaching or the [7:33] instruction when you need to teach when [7:35] you need to be connected with the [7:37] instructor. So AI is allowing us in the [7:39] K12 sector to really provide that [7:41] omniresent um just in time instructors [7:44] that who's there to help students and [7:48] the AI that we are building is not just [7:50] a simple chatbot. We're building it from [7:52] the ground up. This is a uh she is her [7:57] name is Lexi stands for learning [7:59] experience. We put her together, Lexi, [8:02] and she is there. She's trained on our [8:04] proprietary content, all of the [8:05] curricula that is K12 aligned. It's [8:08] aligned to all the standards throughout [8:09] the United States. And she also [8:11] understands the student. And so, she's [8:13] able to put that together. So, it's not [8:15] really the tech, the the the tech that's [8:17] making the change, it's the context. [8:19] Lexi has the context around the student [8:21] and is helping the students persist and [8:23] be able to help them with their with [8:25] learning. So, it's it's a fascinating [8:27] time. I think it's going to be really [8:29] really uh transformational over the next [8:32] 3 years. [8:33] >> I like that idea of transformational [8:35] kind people are using the word [8:36] inflection point in society but also in [8:39] education. This morning uh Jeff Hinton [8:41] mentioned one of the uh great promises [8:43] of AI is in healthc care medicine and [8:46] education. So I'm very happy that we [8:49] having this discussion. So um following [8:51] up uh Angelina about your chess hobby or [8:55] I should say expertise and you hobnob [8:57] and you in Prague and you know winning [8:59] con competitions uh so but you are a [9:02] teaching professor also a chess [9:04] grandmaster um and so you have a unique [9:06] perspective on strategy so from the lens [9:09] of chess uh what do you think is a key [9:12] strategy for incorporating AI into [9:15] education? [9:18] >> Okay uh great question. So look from the [9:20] um viewpoint of chess uh you know I [9:24] assess the position I assess the [9:25] situation and I see the following. So we [9:27] are rapidly moving into uh the future [9:30] where we will have three types of people [9:34] AI builders [9:37] AI users and losers. [9:41] [Laughter] [9:42] So, and seriously, you know, AI is here. [9:46] AI is here to um to stay. And what what [9:52] we have, we have the situation where we [9:53] have to bring everybody on board. We [9:56] have to get professors, we have to get [9:59] staff, we have to get uh students uh you [10:03] know into using AI and it has to be very [10:06] strategic. So like in chess you have the [10:09] position and then you envision hey you [10:12] know like it would be much better to [10:13] have that position but you have to [10:15] envision it first then you use a [10:17] tactical play you know like an execution [10:20] to actually make it happen. So and I [10:22] think you know like to uh to use AI [10:24] effectively to build education around AI [10:28] and uh extract the value we have to [10:30] envision what we want to achieve but it [10:33] has to be everyone's contribution. [10:35] >> Great thank you for that. Um Betsy I [10:38] have a question for you. Um so you know [10:41] for decades we have been discussing this [10:44] idea of personalized education. Um and [10:48] there's huge promise uh about AI [10:50] delivering more personalized effective [10:52] learning experiences. So could you share [10:54] a specific example where you have seen [10:56] AI successfully improve teaching [10:58] learning and institutional operations? I [11:01] think you wear a administrative [11:02] leadership hat as well. um uh and what [11:07] made that implementation work well. [11:11] >> Thanks. Um I think this is the question [11:13] for how to answer Angelina's question of [11:16] how not to be the losers. [11:18] >> Um you know I think really the one of [11:22] the things that I think we take really [11:23] seriously at WashU is the idea that when [11:26] you're educating students they need to [11:28] be able to take risks and they need to [11:30] be able to fail. And so one of the [11:32] things that we've been thinking pretty [11:33] hard about is how is it that we can give [11:35] them access to a toolkit that is gen [11:40] AI enabled but also make sure that they [11:42] don't share data that they shouldn't [11:44] otherwise share or share you know IP [11:46] that they shouldn't otherwise share. And [11:47] so one of the things that we built at Wu [11:49] is a secure instance for chat GPT that's [11:52] HIPPA and FURPA protected. This has big [11:54] implications in terms of how we provide [11:56] patient care. It's really important for [11:58] that kind of data, but it's also really [12:00] important for those learners who are [12:01] coming in to our doors at, you know, the [12:03] ages between 18 and 22 and they need to [12:06] experiment and they need to try things [12:08] out and they need to learn how this [12:10] technology works in a way that then [12:12] their data isn't going to be as for [12:13] inferencing. And I think it's a really [12:15] good example. We're seeing about 50% of [12:17] our students on a regular basis use this [12:19] technology. We've been thinking hard [12:21] about doing some of the things that [12:22] Damian's company is working on. Thinking [12:24] about how we can, you know, create some [12:26] more personalized learning experiences, [12:28] chat bots for individualized students, [12:30] for example. Um, but I think it's really [12:32] important to acknowledge that the first [12:34] step here, the really big thing is this [12:36] investment usually from university [12:37] leadership that says, listen, we need a [12:39] place for students to take risk and for [12:41] them to learn what the wrong thing is to [12:43] do um before they go out in the world [12:45] and work at your companies. [12:48] I would just piggy back on top of what [12:50] uh what you said is what Betsy said was [12:53] um you know the the ability for the the [12:56] the place that universities have in this [12:59] in this with AI and what they're going [13:02] to bring to it is really important and [13:04] industry needs to partner with academia [13:07] because academia has this opportunity to [13:09] do these these these long horizon [13:11] research projects to make sure that uh [13:14] what they're bu what we're building in [13:16] industry and what what they can help us [13:18] with is meant to be able to to make sure [13:21] that we have efficacy in the learning [13:23] that we're making sure that we're we're [13:24] correcting anything that's taking place [13:27] around learning and that not happening. [13:29] And so I think the partnership between [13:31] industry and academia is going to be [13:33] paramount as we move forward. [13:36] >> Um Betsy, you touched upon privacy um in [13:40] your response. So how are you uh [13:42] protecting student privacy and also [13:45] teach those from the institutional and [13:47] you know faculty perspective and uh and [13:49] teaching students to use these uh tools [13:52] responsibly. [13:55] >> So actually you know I'm going to give a [13:56] shout out to a website we put together [13:58] that's awashu.edu [14:00] edu um which gives us like a gigantic [14:03] list of resources because this is [14:05] actually an enormous collaboration from [14:07] whole different pieces of campus from [14:09] the center for teaching and learning who [14:10] are really thinking about how they can [14:12] enable um faculty to to onboard new [14:15] kinds of teaching experiences but also [14:18] to try to think about am I going in and [14:20] out here there we go um to try to think [14:22] about how it is that uh that students [14:25] can actually do this kind of innovation [14:27] I think it's pretty exciting the the [14:29] moment we have vibe coding Um it's [14:31] pretty amazing to see what students can [14:33] come up with. [14:34] >> Great. Thank you. Um going back to chess [14:37] one more time. Uh so Angelina um [14:42] I have dabbled in chess and uh I think [14:45] every rudimentary uh chess program can [14:48] beat me. Um but chess players refine [14:51] their skills. Um so can you draw any [14:54] parallels between how AI accelerates [14:56] learning in education and how chess [14:59] players uh acquire skills? [15:03] >> You know um what's interesting so chess [15:07] has been around for uh for centuries uh [15:10] and um [15:13] people still play this game. So in uh [15:16] 1997 [15:18] uh this was the first time when computer [15:20] debloom actually defeated uh reigning [15:22] world champion uh Gary Kasparov and at [15:25] that time uh it was like well that's the [15:28] end of chess. [15:30] So now you know like the computer is [15:32] going to be taking over. We have very [15:33] parallel situations right now and uh [15:36] that's not the case. People are [15:37] continuing to play chess. they are using [15:39] their skills and moreover you know um [15:42] all the current AI programs they are [15:44] even afraid to play against Atari uh [15:48] chess programs that was developed 50 [15:50] years ago because it's actually they all [15:53] ch and others they are getting beaten [15:55] you know like really badly so um what [15:59] are the skills skills are analytical [16:01] abilities logic uh pattern recognition [16:05] now AI is good at pattern recognition [16:06] but there's also intuition there's [16:08] something you know where um you have to [16:12] um translate [16:14] material into uh you know intangible. [16:20] Uh so if uh if you think about everybody [16:23] can calculate how many pawns on the [16:24] board, how many uh rooks, knights, you [16:27] know like other pieces. So this is what [16:29] exactly happened in 1997. I was um [16:31] actually contracted with IBM. I was [16:33] given the commentaries on this uh on [16:36] this much and [16:38] AI uh computers AI has to be taught [16:42] you know other sense not just the [16:45] calculation but the intuition and that's [16:47] something like really human abilities [16:49] and I think you know like uh what uh [16:52] what we need to do we need to preserve [16:54] what's ours and I can tell you by the [16:55] way every time um chat GPT or other [16:58] model comes with a new uh AI model I do [17:01] the simple test I run um how many of you [17:04] are how many of you play chess? [17:08] Okay, great. So, what I do you know like [17:11] I just give give the task you probably [17:13] heard about it knight tour uh where you [17:16] put the knight on the chess board which [17:18] is uh 8 by8 64 squares and the [17:21] assignment is you have to move knight on [17:23] every square but just one time and cover [17:26] the whole board. uh chat GPT5 came out. [17:30] I plug it in. It failed miserably as all [17:34] the previous versions. So [17:36] >> I cheered for human and I think you like [17:38] AI has a very long way to go. [17:42] >> Okay. Um so uh let's talk. So we have [17:46] talked about opportunities especially [17:47] from the student teaching learning exper [17:50] perspective. Um so Betsy some uh people [17:52] fear AI could replace certain aspects of [17:55] teaching um while others see it as a [17:57] powerful tool to augment uh educators [18:00] work. So from your perspective uh how is [18:03] AI reshaping uh the role of of of [18:06] educators, teachers, professors and what [18:09] new skills or mindsets might educators [18:12] need in this new age? I think there are [18:15] four skills that we need to make sure [18:16] our students have when they finish [18:18] college and I think those have changed a [18:20] little bit. So I think we're seeing you [18:22] know no one needs to do that fsbuzz [18:24] problem anymore. Um but I think we we [18:27] need to see a real push towards critical [18:29] thinking and the challenge for educators [18:31] is to make sure we craft assessments and [18:33] assignments that let students illustrate [18:35] those skills. Um we need a real [18:38] engagement with creativity. I think what [18:40] we're going to see like human greed [18:42] knows no limit. And so we're going to [18:44] see like tremendous potential for [18:47] innovation here for people to come up [18:48] with new things and new ideas that [18:50] people want and that's going to be the [18:51] economy of the future. So we need to [18:53] give our students the ability to dream [18:55] really big. Um they need to work in [18:57] collaborative teams. None of you are [18:58] working by yourselves and I think those [19:00] are really those those those social [19:02] skills that get called soft skills. [19:04] Those are hard and important to teach. [19:06] Um and finally I think we need more [19:07] management and leadership skills. So you [19:09] know all of us are managing um some [19:12] numbers of staff. You're also going to [19:13] include in that staff some number of AI [19:15] agents now and those AI agents are going [19:17] to make mistakes and those are new kinds [19:19] of management tools that need to [19:20] develop. And so I think this makes this [19:22] really big push actually for the core [19:24] value of what a university does which is [19:26] liberal arts education. Um and I think [19:28] the mechanisms by which we deliver that [19:30] are changing but those core values are [19:32] actually are are sort of never before [19:34] more important. [19:36] >> Uh I want to dwell on this a little bit [19:38] longer. Uh Betsy um uh so recently we [19:43] started you know reading that uh people [19:45] are afraid that the next generation of [19:48] learners uh may outsource so much of [19:50] their thinking um they lose their [19:52] critical thinking skills and everything [19:54] can be solved by a chain of thought GPS [19:58] etc. So what keeps you awake at night [20:01] when it comes to AI and education. And [20:03] so I want to focus a little bit on the [20:06] kind of challenges and some um um [20:09] outcomes that we did not predict. So [20:11] what are the biggest dangers or risks [20:13] you see and how should we address them [20:16] effectively? [20:18] >> So I think the the the biggest danger [20:20] the the greatest risk is actually [20:21] inaction. I think it's very easy for [20:24] universities to to stop and say hang on [20:27] this could be dangerous and to not try. [20:30] And so I think when you have [20:31] universities or other institutions that [20:33] you expect to be around for the next 100 [20:35] years, they tend not to move very [20:36] quickly. So I think right now the [20:38] biggest challenge for most universities [20:40] is is to go ahead and and take some risk [20:43] um and usually to trust that um you know [20:46] that's the world is changing and if we [20:49] don't work on doing what Angelina said [20:51] upskilling our faculty and our staff and [20:53] our students um we're not going to be [20:54] competitive in the world. the world is [20:56] changing and so I think the risk is to [20:57] do nothing and that's the biggest threat [20:59] for universities. I think you're you're [21:02] right to think about the concern for you [21:04] know students outsourcing their thinking [21:05] and I think the challenge for all of us [21:07] is to think about well how do we help [21:09] students develop those skills and assess [21:11] them and that question of assessment is [21:14] is the really big question on the table [21:17] um to help the students understand that [21:20] they need these skills and we want to [21:21] work with them but we we need to give [21:23] them a way to illustrate what it is that [21:25] they're doing. [21:26] >> Okay, Angelina. Yes. [21:27] >> Yeah. way I add to this not from the [21:29] chess perspective but actually from the [21:30] tition perspective. So look uh what I [21:35] feel is uh very important right now and [21:38] where we have two areas of kind of [21:41] disconnect. So one uh one area we have [21:46] the battle of generations. So the [21:48] educators are older. So we grew up [21:52] without having AIs without having even [21:55] computers. Um now [21:59] we have to teach digital natives [22:03] and so what you could see you could see [22:06] huge reluctancy among many faculty to [22:10] accept and adopt uh they feel you know [22:14] oh that's going to go away that you know [22:16] like we we should not um like let's just [22:19] wait it's not going to happen and uh [22:22] about a year and a half ago we had this [22:24] survey um at Thunderbird. So at that [22:27] time I was s you I was also reading a [22:29] lot about how is everybody other [22:31] universities are doing and about [22:33] adoption rate was about 20% of faculty [22:37] uh embracing a uh AI and 80% not so [22:40] Thunderbird was doing much better at [22:42] 50/50 [22:43] uh and ASU is number one in uh [22:46] innovations for 10 years in a row. So [22:49] and Thunderbird is at for uh forefront [22:51] of um uh of the university. So um [22:56] meanwhile the same you know survey for [22:58] the students we have 80 to 90% of the [23:01] students were using AI actively. So you [23:05] can't teach in those conditions. So we [23:08] really have to bring up the faculty. We [23:12] really have to empower them. We have to [23:14] give them the skills the abilities the [23:16] understanding the use cases of hey this [23:19] is how you should use it. Don't be [23:20] afraid. Mhm. [23:21] >> So AI is not going to replace you. [23:23] >> Um and the second part which I think [23:26] extremely important is how students are [23:30] using it. And this is where we have [23:32] another huge challenge because um a lot [23:36] of times students are using it as [23:39] Google search like replacement. So as uh [23:43] hey do my homework for me. That's not [23:46] the way to uh to live. It's not the way [23:49] to develop skills and we really have to [23:52] explain to students we have to show to [23:54] students. We have to rethink our own [23:56] process of how to make sure that [23:59] students use AI responsibly effectively [24:03] that they actually use AI to build their [24:06] skills and who they are instead of [24:08] getting the grades and getting the [24:10] homework done because then if AI does [24:13] your homework well you are not needed. [24:18] Although those the third category the [24:20] losers [24:23] okay uh moving on um uh so Betsy let's [24:26] talk about um uh AI and education and [24:30] the tech piece of this so as we know [24:33] this is not only about technology it's [24:34] also about culture funding policy okay [24:38] so um in your role uh what has been the [24:41] biggest challenge you have faced or just [24:44] observed uh in trying to bring these [24:46] initiatives AI initiatives to life in [24:48] your institution and how have you worked [24:50] to overcome them? [24:54] >> So, I'm going to say something and it's [24:56] going to sound like um [24:59] I I think actually there's tremendous [25:01] hunger at universities for for people to [25:04] come up with ideas for how to solve some [25:05] of these problems. And um so I actually [25:08] think at Wu we've we've been going [25:10] pretty full speed ahead in terms of um [25:13] the innovation. the innovations that [25:15] we've been that that doing the best on [25:17] has has just been this really big [25:19] question of research and I think there's [25:21] actually like this tremendous [25:23] recognition from across the country that [25:25] all of a sudden truth has become really [25:27] expensive. It's really hard to figure [25:29] out what is true. We're seeing, you [25:32] know, fake grant proposals getting [25:33] submitted to the NIH. And we're seeing [25:36] fake research articles getting submitted [25:37] to research journals. And all of a [25:39] sudden, the world is kind of short on [25:41] people who can actually interrogate [25:44] things and say, "Okay, well, this is [25:45] actually true." And so, I think there's [25:47] a recognition that that's a place that [25:49] universities have a tremendous role is [25:51] to validate what's true. Um, you know, [25:54] they're nonpartisan entities that are [25:56] deeply grounded in that entire mission. [25:58] So, so I would say it hasn't been a [26:00] university or there there isn't a hurdle [26:02] to it. I think there's this sense of [26:04] urgency and I think the the big threat [26:07] right now is that we have to move really [26:08] quickly to try to validate truth. Um I [26:11] think the New York Times yesterday had a [26:13] new article on just the quality of deep [26:15] fakes that have come out. Um and and the [26:18] need and the tremendous need for all of [26:20] us to be innovating in tools that do you [26:22] know to validity and on this quest for [26:24] truth. So, um, you know, I think that's [26:27] the that's the component of where we're [26:29] seeing this this tremendous momentum [26:32] coming. Um, we're seeing lots of [26:34] innovation on the the medical care side [26:36] and there's all kinds of ways in which [26:38] the world is becoming easier and more [26:40] interesting. And I think that's the [26:42] thing to really celebrate. The challenge [26:44] is usually that there's, you know, some [26:45] kind of form that has to get filled out [26:47] that someone is concerned about risk. Um [26:49] but but I think there's this real need [26:50] in society for someone who can talk, you [26:53] know, a little bit longer and think a [26:56] little bit harder about how it is that [26:57] we can we can provide a a ground source [27:00] of truth to a broader community. [27:03] >> Yes, Damian. Um please. [27:05] >> So I was just going to piggyback on what [27:07] Betsy was saying. So there's many many [27:09] different areas of innovation with [27:11] artificial intelligence. It's not just [27:13] in instruction. It's not just going to [27:14] be in curriculum development. and you [27:16] start to take a look at the different as [27:18] learning in general, right? So you've [27:20] got education, the formal process of [27:22] learning, you've got your K12 sector, [27:24] you've got your post-secary sector as [27:26] well. All of this is going to go through [27:28] a massive transformation. It's not just [27:30] going to be at the university level or [27:32] at the K12 level. Every area is going to [27:34] be disrupted. And I don't want to I [27:37] don't want to underscore the profound [27:38] impact of what AI is going to have on [27:40] learning. I was listening to Eric [27:42] Schmidt a few uh weeks ago on uh [27:46] moonshots if you listen to that podcast [27:48] and they asked him what's the next big [27:49] piece of technology going to be and he [27:51] said the next Google will be someone who [27:54] comes out with a learning device right [27:56] here on your phone right here where [27:57] human beings can learn and um that will [28:00] be the next Googleiz company that's [28:02] going to come out and that's all around [28:04] learning. So there is tremendous [28:07] innovation that's happening with [28:09] artificial intelligence right now. Um [28:11] there's tremendous opportunity. Yes, we [28:14] have to be wide open wideeyed and [28:16] realize that the gaps that we can create [28:18] using artificial intelligence can also [28:20] be just as profound as the gaps that we [28:22] can close if we're not careful and [28:24] responsible about designing artificial [28:26] intelligence. But there's a huge [28:28] opportunity. I want to make sure like [28:31] where we are seeing this right now. we [28:33] see a transformation. [28:35] I remember when I go back to my my my [28:38] schooling back when I was doing my MBA [28:40] and we learned about Michael Michael [28:42] Porter's five forces, right? The five [28:44] forces that shape an industry. And the [28:47] fourth force was the threat of [28:50] substitutes. And those threat of [28:52] substitutes, that's exactly what we're [28:54] seeing right now. This is the biggest [28:55] substitute that we're ever going to see. [28:58] And if we're not careful, you know, we [29:00] will be rep there are for sure going to [29:05] be institutions that will be replaced [29:07] because if you give someone an [29:09] opportunity to do something differently [29:11] and they and you are not seeing this [29:14] threat of the substitute, this will move [29:17] past you and you you could be replaced [29:19] very easily. [29:20] >> Great. Thank you. Um again this morning [29:23] at Jeff Inton's um keynote, he touched [29:26] upon uh research. So let's talk about uh [29:28] a pillar of universities is is research. [29:31] So um universities historically have [29:34] held a unique edge in research uh [29:38] because of access to data, talent, [29:40] computing power and the mission of [29:41] research universities. So as AI [29:44] transforms this research and innovation [29:46] uh what unique role do you believe [29:48] universities can play in answering [29:51] questions or solving problems that might [29:53] be out of reach for industry? because [29:55] especially in AI we have seen that a lot [29:58] of um new um [30:01] innovation has come from uh private [30:04] industry. So maybe Angelina do you want [30:06] to say something about this? [30:08] >> Well the difference between universities [30:10] and uh private industry we don't have to [30:13] worry as much about return on investment [30:16] in a very short period of time. So when [30:19] private companies uh have have to they [30:23] don't really have a choice but they have [30:25] to perform and have the financial [30:28] results and I'm teaching global finance [30:30] so I'm really like uh into you know like [30:33] the performance and ROI uh well [30:36] university can take a longer position [30:39] and they can uh allow this research and [30:42] development to go through the different [30:44] stages trying to figure out and fail you [30:46] know and uh try again. Um, now what's uh [30:50] what's interesting however I see I see [30:52] that AI is going to provide amazing [30:55] opportunities, amazing resources, [30:57] amazing help because for example I have [30:59] a colleague uh who took eight years to [31:04] uh to work on her uh PhD uh because she [31:08] did the surveys and she need she needed [31:10] to um analyze the data and the data was [31:14] uh uh qualitative not quantitative. So [31:17] it took a lot of time to to read and [31:20] compare and to uh to make the [31:22] conclusions. Well, right now more or [31:24] less you can just plug it all all into [31:27] AI and within you know like a few [31:28] seconds you're going to have all the [31:30] different conclusions and everything [31:31] that you need to know and then you can [31:33] you know like uh provide the commentary [31:35] and you are done. Well it's also creates [31:37] enormous danger because exactly you know [31:39] like as Bessa said uh what's true what's [31:42] not true? How do you know if the [31:45] conclusion, the result, whatever you [31:46] receive, the report is actually [31:48] accurate? Think about uh if if there are [31:53] doctors who would be correct 95% of the [31:57] time or even 99% of the time, would we [32:00] really want to are we going to trust [32:02] them? [32:03] >> What about the architects who going to [32:05] you know like build all this beautiful [32:07] highrises uh but they would be wrong one [32:10] time of 1% of the time? I don't want to [32:13] live in that highrise. So, um I think [32:16] you know like it's very very important. [32:18] We have the issues and you know it's [32:20] it's it's it's another thing that I want [32:22] to mention. [32:25] We shape AI and AI shapes us. It goes [32:29] back and forth. How many of you have [32:33] started the conversation with AI saying [32:35] hi AI? [32:38] Okay, just a few. And a lot of times [32:41] it's like give us just do it fast you [32:46] know okay tweak it I don't like that [32:48] part do this part so we [32:52] you know it's like kind of if we wake up [32:54] in the morning and we don't make our bed [32:57] it's it's who we are it's shaping it's [33:00] you know like we have to stay who we are [33:03] we have to keep our values our [33:05] personality our tra traits as a human [33:09] and how many of you are now using delve [33:13] into [33:16] you know all these different expressions [33:18] that AI kind of put on us. So as I said [33:20] like it's going uh it's going back and [33:22] forth it's changing. So we have to be [33:24] very careful uh of our interaction with [33:27] AI and we can have a lot of benefits. We [33:30] can get superpower skills from the AIS [33:32] but we can also um have really bad [33:36] results. So uh you know so there are [33:39] opportunities in research but there's uh [33:41] also you know like um [33:44] warning signs so you know like safety [33:46] first. [33:47] >> Yeah Damian [33:48] >> I was just going to say you know when we [33:49] started designing our platform the first [33:51] thing we did is we started with [33:52] designing the curricula right we have [33:55] been in the business for developing uh [33:56] digital content for over 20 years. I [33:59] will tell you that in the past it took [34:01] us about a half a million dollars in [34:02] order to build an algebra course. [34:04] Algebra 1, the most failed course in the [34:07] United States, um, is Algebra 1. And, [34:10] uh, and we have spent millions of [34:12] dollars trying to crack the code on how [34:14] to effectively teach algebra 1 to [34:16] students digitally. And it's tough, but [34:19] we started with our course builder is [34:23] what we simple name course builder [34:26] with the models that we were using and [34:28] we use all of the models. We're we're [34:29] we're LLM agnostic. Um there are dozens [34:33] and dozens of models that we use. Um and [34:35] some of them do other things very [34:37] differently, but as the models progress, [34:39] they're becoming more and more powerful [34:41] and they're becoming much more accurate. [34:42] A lot of people talk about the [34:43] hallucinations and artificial [34:45] intelligence. I can tell you a lot of [34:47] that is being we're starting to see much [34:50] much better results. But what we've done [34:52] is we developed a um a prompting studio [34:56] which is using chain of thought and [34:58] chain of tree prompting. And our prompts [35:00] look like something some of them look [35:01] like the novel war and peace. I mean [35:03] these are these are comprehensive [35:06] prompts that are going in there. And in [35:08] addition to that we're taking every [35:10] single piece of content whether that's a [35:12] formative assessment or a summitative [35:14] assessment piece and we are running it [35:16] that runs it through a rubric. So our [35:18] system is multi-agentic and it's also [35:20] multi- aent. And that system right there [35:23] then will take a look at every single [35:24] piece of content that's being developed. [35:26] It's going to score it on a rubric and [35:28] it's going to send it back and it's [35:29] going to have agents who are going to be [35:31] looking the AI agents that are going to [35:32] be looking to see exactly how can we [35:35] improve that piece of content. So, it's [35:37] going to get profound where we're going [35:38] with this, but um it's very interesting [35:41] the the capabilities are increasing [35:43] exponentially. [35:45] >> Yeah, thank you Damian for for that [35:46] perspective from your strong mind and [35:48] use of agency agentic uh and AI. uh [35:51] maybe I have a follow-up actually with [35:53] the same kind of point uh maybe with [35:55] Betsy. Um what is one thing that a [35:58] university or your university has [36:00] developed or has you have come across [36:02] that you would like to share with [36:04] industry and users? [36:06] >> I'm so glad you asked that question [36:07] because I actually have a request for [36:09] the audience. So um you know one of the [36:12] things I think that uh universities have [36:14] as a superpower when we hire faculty we [36:17] hire faculty because they are really [36:19] good at asking questions. Not any [36:22] question they don't but they're asking [36:24] questions to really interrogate the [36:26] future. They're thinking hard about what [36:28] is possible. This is something that AI [36:31] doesn't do very well. Um but university [36:34] faculty are incredibly well equipped to [36:36] see a little bit into the future. Some [36:39] of them can even see through walls. I [36:40] mean, it's amazing to meet some of these [36:42] folks. And so, so they have some sense [36:44] of where the world is going and what [36:46] else is possible. And some of you in [36:49] this room have access to all kinds of [36:51] interesting tools where, you know, it [36:53] would be really great if you had a [36:54] partner who could test out something in [36:56] beta where you have a sandbox and you'd [36:58] like someone to interrogate it to see [37:00] what kinds of outputs and values it [37:03] explores. And I think there's actually a [37:05] tremendous number of opportunities for [37:07] collaboration with the kind of person I [37:08] just described to you that can you can [37:10] see where the world is going. Um, and [37:12] you'd like to have you you have [37:14] something that we also want access to [37:16] some new kind of tool, some new kind of [37:18] agent. Um, and and you'd like to [37:19] collaborate. And so if that's you, if [37:22] you're sitting in the room today and you [37:23] think I would like some kind of [37:24] university collaboration, come see me [37:27] after the panel. [37:29] >> Nice. So we'll be hanging out outside [37:32] after this because another session comes [37:33] here. If you have questions for any of [37:35] the panelists, please see us. So I think [37:38] we are almost at the end. So I'm going [37:40] to have one um kind of open-ended final [37:43] question for our panelists. Um on a [37:46] positive note, okay. Um so what trends, [37:49] technologies or AI capabilities do you [37:52] believe will most transform education in [37:54] the next 5 to 10 years? What are you [37:57] most excited about? If we met here on [37:59] this stage five years from now, [38:03] what what would excite what would you be [38:05] thrilled about? Maybe Angelina you can [38:08] start. [38:09] >> Okay. Um [38:12] well the way I see it as I said you know [38:14] like um ASU's really big in innovations [38:17] and uh we see amazing technology we are [38:21] using avatars where actually you know [38:23] like my avatar can be teaching you know [38:26] and uh presenting to to students all [38:28] over the world and we are very global uh [38:31] school we have like 95% of our students [38:34] are from all over the world um and so [38:37] have this technology where you have [38:40] movies uh podcasts uh uh you know a lot [38:44] of uh augmented and virtual reality that [38:48] is going to be part of the educational [38:50] process uh I think you know like this is [38:53] what's going to happen it's actually [38:54] happened uh right now so for example I'm [38:56] redesigning my courses and uh I'm using [39:00] um chat GPT [39:03] to ask asking make a prompt for manosai [39:07] which is aentic AI I'm asking Manus Aai [39:10] to get the script for um Notebook LM and [39:15] I'm asking Notebook LM to do the podcast [39:18] for my students. Uh and I can tell you [39:20] the reason why just briefly uh because [39:23] uh in uh February like I'm three times [39:25] US women's chess champion and that was [39:28] you know like a while ago. Now I'm back [39:29] into chess and I'm playing and in [39:31] February I was playing in Prague as a [39:33] first board um of the US uh women's 50 [39:36] plus team. So it was world chess [39:39] championship. Um and the team really [39:42] needed me. Meanwhile, you know, like I [39:44] had to teach. So I was playing chess in [39:46] the world championship during the day [39:49] and I was teaching at night. [39:51] >> And I had students in China. I had [39:53] students in Ukraine. I had students in [39:55] the US. Uh you know, I'm a human being. [39:59] I I need some time to sleep. So I can't [40:02] teach 24/7. I can't play 24/7. So AI [40:06] gives us the opportunity to uh to [40:09] actually expand oursel to uh to scale uh [40:14] you know like and this is enormous [40:16] value. It's enormous potential and I [40:18] think you know like we should envision [40:19] the future and this is where we are [40:21] heading. We're going to have a lot of [40:23] great things you know like that we can [40:25] present to the students and uh teach [40:27] them moving forward. [40:29] >> Great Betsy. What excites you most? [40:31] >> We have three things. First it's what [40:33] Damian's been talking about. It's the [40:35] ability to provide a really personalized [40:37] experience for each student. So, if done [40:40] well, they should actually be able to [40:42] learn things even more deeply. They [40:44] should have a really profound [40:45] understanding of the material that [40:46] they're exposed to. That's pretty [40:48] exciting. The second is just this is a [40:51] technology that should unleash [40:53] incredible human creativity. So, for [40:55] those of you who went to the plenary [40:57] session this morning, like the idea that [40:59] any 17-year-old can make a movie. It [41:01] means the currency of the realm is [41:02] actually not the ability to make a [41:03] movie. It's taste. It's what we should [41:05] see are better movies, friends. Um, and [41:07] I think the ability to tell stories and [41:09] read read books and like just consume [41:12] material, music, like we're going to see [41:14] more amazing things presented. But the [41:17] final thing that I'm just so excited [41:19] about for for all of us, especially the [41:21] research community, is this idea that [41:23] all of a sudden we can innovate faster. [41:25] And there's a lot of stuff the world [41:27] needs us to innovate on much faster. [41:29] Whether that's drug discovery or [41:31] understanding the human biome or, you [41:33] know, basic questions around strategies [41:36] for climate change. like this idea that [41:38] we can process data and we can actually [41:40] you know link fields together and we can [41:42] do it faster is going to be the case [41:44] that the students of today are going to [41:46] be solving the problems that we have and [41:48] they're big. [41:49] >> Thank you. And so finally Damian uh what [41:53] excites you the most what in five years [41:55] what you think would be [41:57] >> I think this is going to be the five [41:58] years of the highest change in [41:59] educational technology. I think schools [42:01] are burned out. they've been over ed [42:04] tech is overpromised and underdelivered [42:06] on so many different areas with uh with [42:08] educational technology and I think what [42:10] we're going to see is um autonomous [42:12] generative learning ecosystems start to [42:15] take place where students are going to [42:17] finally we'll be able to deliver on the [42:19] promise that we have deliver that we've [42:21] been telling everyone in education and [42:23] we're going to be able to place a a [42:26] instructional presence alongside every [42:29] student to close the gap between learner [42:32] and and and teacher and support these [42:34] teachers. I also see this as not going [42:36] to replace teachers. I think it's going [42:38] to empower teachers. It's going to give [42:40] them the ability to really understand [42:42] the needs of those students, of their [42:45] own students, and what's taking place in [42:46] that student's education, how that [42:48] student is learning, and be able to [42:50] support that student even more. So, I'm [42:52] very bullish on where I see artificial [42:54] intelligence going. I'm also wide open. [42:57] I think that we can create a lot of [42:58] problems just as easily as we can solve [43:00] problems. And we need to be responsible [43:02] and we need to partner with with with [43:04] higher ed because they have the ability [43:07] and the resources as they've mentioned [43:09] to be able to do these longtail long [43:11] horizon research projects and industry [43:13] would be very wise to partner with them [43:15] so that we can make sure that what we're [43:17] designing is not just another you know [43:20] software release. It's it's it's done [43:23] differently and that it's because this [43:25] is students learning. These are this is [43:27] about how human beings are going to to [43:29] embrace the future their own future and [43:31] we're moving into this future very very [43:33] quickly right now. No. And just one more [43:36] thing I want to mention. I think you [43:38] know the education uh expands and it's [43:41] able to reach all those people who might [43:43] not be able to get the education [43:45] otherwise because for example we have a [43:48] 100 million learners uh program [43:50] initiative at Thunderbird where we [43:52] actually trying to reach 100 million [43:54] learners all over the world for free. Uh [43:56] and our classes are translated in 40 [43:59] different languages. So that's exactly [44:01] you know like that opportunity where we [44:03] we provide for for people who can't have [44:07] the education the education for free. So [44:10] we lift the whole society up and we [44:12] getting you know like all the human [44:14] beings uh going into the future with new [44:17] hopes. [44:18] >> Wow. So with that hopeful message, I'd [44:21] like to thank my panelists here, Damian, [44:23] Betsy, Angelina for your participation [44:26] and insights on this very important [44:28] topic. And of course, thank you to the [44:30] audience for your attention. Uh we are [44:32] out of time, but uh we will be outside [44:34] if anyone wants to chitchat, discuss, [44:36] bring up some ideas. And thank you all [44:38] very much for the session.