University Leaders Warn AI Implementation Double-Costs Threaten Budgets

大学领导警告AI实施的双重成本威胁预算 高等教育机构面临日益严峻的成本压力，供应商软件许可费用与员工重训及学术工作流重组所需的开支相互叠加。 来自多所大学的研讨嘉宾探讨了将人工智能融入高等教育所面临的战略、经济和运营挑战。对话强调了持久的人类技能与短期职业技能之间的区别、以人为本的AI治理的重要性，以及在快速技术变革中为教职员工提供支持的必要性。此外，发言者还讨论了传统机构学位与微凭证、数据控制与基础设施成本，以及地理空间AI的全球潜力。最终，他们展望了一个共生的未来，即负责任地部署海量智能以改善学习成果并激发人类潜能。 高校在引入 AI 时面临“双重付费”的经济挑战：既要支付商业软件的许可费用，还必须承担员工技能重塑、资源重用以及教学工作流重新设计的庞大隐性成本。 (0:04) 面对日新月异的技术迭代，高校应侧重培养批判性思维、适应力与元认知等持久技能；学位代表着机构对学生人格与思维的整体塑造，而非仅仅记录单项技能掌握情况的成绩单或微认证。 (0:40) 圣路易斯大学将“以人为本的 AI”和伦理审辨融入多学科教育，将其实际应用拓展至临床医学、航空起降安全评估以及基于开源图像的被拐儿童搜救等现实场景中。 (6:46) 教职员工对 AI 的抗拒往往源于被技术取代的恐慌，高校需将教师角色从单纯的信息传递者转变为高阶导师，并通过同行示范与切实的激励机制推动教学转型。 (21:13)

University leaders argue that artificial intelligence is compelling higher education to shift from passive information delivery to human-centered formation, requiring institutions to navigate hidden implementation costs, faculty apprehension, and accelerating knowledge lifecycles. AI Economics and Institutional Data Sovereignty (33:07) Adopting AI introduces financial and operational strain that contradicts vendor promises of immediate efficiency. Jason Baker notes that universities initially 'pay twice' for AI: first for proprietary software subscriptions and subsequently for the costly retooling of personnel, resources, and institutional workflows. This dual expense challenges institutional budgets before efficiency gains materialize. Lin Zhou cautions that reliance on third-party commercial vendors threatens institutional data governance. Drawing an analogy to historical consolidations in the computer hard drive market, Zhou warns that universities risk forfeiting proprietary intellectual property if vendor platforms fold or alter terms. Furthermore, scaling frontier commercial models across entire student populations presents ongoing cost hurdles, with usage rates reaching $5 to $10 per million tokens. Degrees, Durable Skills, and Institutional Formation (64:41) As technical skill requirements shift between a student's matriculation and graduation, university leaders distinguish transient technical capabilities from durable human capacities. Baker contrasts micro-credentials and course transcripts—which track modular skill mastery and can be assessed more affordably by commercial third parties—with academic degrees. In his view, a degree represents an institutional endorsement of holistic character and discipline-specific thinking that employers continue to prioritize. Zhou frames this educational lifecycle as an essential journey of cognitive development across the brain's 100 trillion synapses, building commitment, passion, and resilience. To remain relevant, institutions must integrate human competencies like conflict resolution, critical thinking, and metacognition into active coursework rather than teaching them as theoretical concepts. Faculty Adaptation and Accelerated Knowledge Cycles (21:13) Widespread AI integration within universities remains stalled partly because faculty members harbor personal anxieties about professional obsolescence. Gregory Triplett argues that overcoming this hesitation requires academic leadership to align AI projects with economic development and three operational benchmarks: novelty, wide adoption, and measurable impact. Baker adds that voluntary peer demonstrations of classroom success persuade reluctant faculty far more effectively than mandatory administrative workshops. Simultaneously, the speed of academic knowledge generation has shifted from centuries-long intervals to monthly cycles. Zhou observes that students now routinely fact-check instructors during lectures using AI tools, estimating they are accurate 70% to 80% of the time. This shift forces classrooms to move away from authoritative one-way lecturing toward continuous, collaborative knowledge creation. To facilitate this, Anjelina Belakovskaia notes that Arizona State University utilizes interdisciplinary communities of practice to clarify open issues surrounding research productivity, classroom ethics, and copyright. Interdisciplinary Applications and Legacy Leapfrogging (61:32) Rather than isolating computer science, Saint Louis University embeds AI across disparate disciplines to solve applied challenges. Examples include training 476 medical residents in an 'AI and Medicine' track, evaluating airfield landing safety in aviation, and using AI on open-source imagery to locate abducted children for law enforcement. Working alongside the National Geospatial-Intelligence Agency, researchers combine AI with temporal and spatial data to evaluate supply chain bottlenecks and global climate trends. Addressing global implementation, Baker observes that institutions and developing nations unburdened by legacy educational architecture possess a 'greenfield' advantage. Much like developing regions bypassed landline telecommunications directly for mobile networks, organizations without rigid historical structures can rapidly deploy novel, collaborative educational delivery models. The panel forecasts an increasing symbiosis between human cognitive effort and computing workflows by 2030. However, the long-term viability of this transformation hinges on unresolved questions regarding how universities will finance recurring token-based computational costs, secure institutional data against platform failure, and manage faculty displacement anxieties.

多所高校的教育管理者指出，人工智能正在深刻颠覆高等教育的传统定位；大学的核心价值不仅在于传授短期就业技能，更在于通过全人培养塑造持久能力、深化跨学科伦理审辨，并重塑师生协同创新的课堂关系。 技术转型难题与教育的双重经济成本 (0:04) 高校在引入人工智能时面临着多维度的体制压力。德克萨斯理工大学副校长兼首席信息官 Lin Zhou 表示，信息技术团队当前的核心挑战是在提升教育标准、加速科研、拓宽学生就业渠道以及树立大学品牌这四大支柱间取得平衡，而同时快速推进这些领域的技术采纳极为困难。圣路易斯大学的 Gregory Triplett 院长指出，作为一所天主教耶稣会大学，其宗旨是实现“全人教育”并服务于更高社会利益，但生成式技术对内容创作与设计的简化，加剧了在快速技术迭代与学生综合人格塑造之间寻求平衡的难度。 针对教育科技供应商宣称的降本增效，摄政大学的 Jason Baker 强调高校在实际落地中面临复杂的经济负担，往往被迫“买单两次”：机构不仅需要为软件许可证支付初期采购费用，还必须承担重新培训教职员工、调整教学资源和重构工作流程的巨额隐形成本。此外，亚利桑那州立大学雷鸟全球管理学院的 Anjelina Belakovskaia 指出，本科学制长达四年，而不同教师对 AI 的接受度差异巨大，管理者必须前瞻性地规划未来需求，而非仅盯着日常教学工具的更迭。 持久能力培养与大学学位的本质价值 (0:40) 在人工智能重构劳动力市场的背景下，Jason Baker 提出了“持久能力”与“即时技能”的区别。他认为，沟通、冲突解决、问题应对、批判性与创造性思维以及适应力属于持久能力，而针对特定入门岗位的即时技能更新极其迅速，例如其子女在大学四年间就业市场的热门技术便发生了彻底更迭。因此，高校必须将元认知思维融入日常课程，引导学生评估推导答案的思考路径及其实际应用方向，而非仅仅追求标准答案。 面对微证书和数字徽章的兴起，讨论明确划分了“成绩单”与“学位”的界限。Baker 认为，成绩单仅记录单门课程与离散技能的掌握程度，第三方商业机构能以更低成本、更快速度完成此类技能测评；而学位则代表大学对学生在某一学科内完成人格塑造与系统培养的权威认可。Lin Zhou 亦认为，尽管部分家长因成本与技能更迭对高等教育产生疑虑，但为期20至25年的受教育历程提供了开放式学习体验，能够教会学生如何投入专注与坚持履责，这种综合人格特质是应对复杂世界的关键。 化解教师焦虑与重构课堂共创关系 (21:13) 将 AI 融入教学的最大阻力之一在于教职员工对职业被取代的恐慌。Triplett 指出，部分教师担心普及 AI 后大学将不再需要传统教员，这种脆弱感阻碍了技术的迅速推广。为此，他提出推动创新的三项基准：新颖性（未尝试过的领域）、普及性与可量化的影响力。他主张将 AI 创新与具体经济发展机会挂钩，并拉近学术界与产业界的纽带，以此为教师营造安全的转型框架。 Baker 补充道，若教师将自身定位为单纯的信息传播者，遭遇恐慌在所难免，因为机械讲授并非教育人才的最佳用途。他指出，自上而下的行政培训难以激发热情，唯有通过同行之间的成功范例展示与经验互鉴，才能促使教师主动重构教案。Lin Zhou 从知识更迭周期分析，过去学术内容以百年或几十年为周期静态更新，如今数月或数天即出现颠覆性成果；在现代课堂上，学生利用数字工具对教师陈述进行实时事实核查的正确率高达70%到80%，这迫使传统的单向讲授必须转变为师生协同共创知识的互动机制。 跨学科应用实践与以人为本的伦理审辨 (6:46) 圣路易斯大学主张将人工智能贯穿于跨学科研究，而非局限于计算机系内部。Triplett 介绍，该校不仅在商学院与医学院开设联合课程，为476名医学住院医师设立“AI与医学”培训通道，还拓展至航空领域以评估跑道着陆安全。在人权保障方面，研究人员利用开源图像与 AI 技术分析失踪或被拐儿童的照片，精确定位地理环境，从而协助执法部门完成搜救。对于哲学、艺术与历史等非技术专业，教育重点在于技术应用如何契合社会安全与公共价值，而非底层代码开发。 伦理审辨被视为人机交互中的核心壁垒。Triplett 表示，企业对理解 AI 治理与伦理的毕业生需求日益迫切，学校将治理视作提出正确伦理问题的能力，而非被动遵从合规条款。Lin Zhou 引述两千年前苏格拉底反对文字发明的典故，指出如同文字最终重塑了学习一样，AI 正倒逼高等教育聚焦于人类本质特征的激发，推动大脑中100万亿个突触所蕴含的激情、专注与深度投入。Belakovskaia 则分享了亚利桑那州立大学通过每月“AI 技术讲座”组建实践社区的经验，探讨提示词优化、知识产权归属和教学伦理等现实议题。 底层架构主权与未来人机共生展望 (33:07) 在技术架构与数据资产层面，Lin Zhou 发出长期治理预警。他以20年前硬盘驱动器厂商由几十家骤减至两三家的整合历史为例，警告各高校若过度依赖第三方商业嵌入模型或闭源聊天平台，一旦服务商变更条款或倒闭，大学将面临丧失核心知识产权的严重风险。此外，顶尖前沿模型数以万亿计的参数规模带来了高昂成本，百万 Token 单价若达到5至10美元，在庞大师生群体中全面推行将造成巨大的财务负担，高校必须构建具有数据自主权与可持续算力的技术管线。 讨论还探讨了后发优势与未来演进趋势。Baker 指出，如同部分发展中国家跳过固定电话网络直接进入移动支付时代一样，缺乏传统包袱的组织或非传统教育机构具备“绿地优势”，能够以更快速度探索前沿教学模式。在总结预测中，Lin Zhou 预言碳基生物与硅基计算的共生融合将进一步加深，预计到2030年人机协同的原型方案将成为常态；Triplett 则认为，只要在部署过程中坚守伦理与责任，“充裕的智能”有望显著改善全球公共卫生并促进社会公平。 本场研讨展现了高等教育在拥抱人工智能时的前瞻思考与现实困境。尽管管理者对人机共生与充裕智能的长期红利持乐观态度，但高校在短期内仍面临双重财务重负、教师抵触情绪以及第三方商业技术对学术数据主权的潜在威胁；如何在坚守人文全人教育宗旨的同时，真正打破学科孤岛并建立可持续的技术管线，仍是全球高校悬而未决的系统性挑战。

## Background 本内容由 Ai4 于 2026 年 8 月 22 日发布在 YouTube 平台上。视频记录了一场由高校管理者与前沿教育技术领袖共同参与的圆桌讨论，探讨 AI 如何在重塑课堂教学与提升学习成效的同时，应对落地实施、技术伦理以及教育公平等核心挑战。 ## TL;DR 来自亚利桑那州立大学、德克萨斯理工大学、摄政大学和圣路易斯大学的教育领袖们深入探讨了 AI 时代高等教育的根本转型。讨论聚焦于持久能力（如元认知与批判性思维）与即时技能的平衡，剖析了学位认证与单纯技能证书的本质差异，并直面高校在双重采购成本与教师职业焦虑等方面的现实阻力。嘉宾们强调必须坚守“以人为本”与全人教育的使命，通过跨学科实践、同行赋能及打破信息孤岛，推动教育向双向共创转型，让充裕的智能真正造福全人类。 ## Atmosphere & style 圆桌讨论的基调兼具学术深思与务实犀利，各位学者与管理者既有对“苏格拉底文字之争”和碳基硅基共生的宏大哲学审视，也毫不避讳地调侃高校领导的减压威士忌以及一线教师的失业焦虑。主持人的国际象棋战略视角与嘉宾们对现实痛点的剖析形成了紧凑而默契的对话节奏，使整场研讨不仅充满理性的战略前瞻，更洋溢着深厚的人文关怀与现实温情。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=-eykunyQ55Y&t=4s) 高校推进 AI 变革面临多维现实挑战，包括技术选型与加速落地、软件采购与人员重塑技能的“双重经济成本”，以及在自动化浪潮中坚守全人培养与育人使命。 - [04:40](https://www.youtube.com/watch?v=-eykunyQ55Y&t=280s) AI 时代能力培养需平衡“持久能力”（批判性思维、沟通、适应力）与“即时技能”；持久能力无法脱离情境抽象教授，需在具体实践中通过元认知追问深入其求解逻辑。 - [04:40](https://www.youtube.com/watch?v=-eykunyQ55Y&t=280s) 企业对 AI 时代治理与伦理人才的核心诉求，并非机械的死板合规，而是对算法及其生成内容“提出正确问题”的思辨与审视能力。 - [10:32](https://www.youtube.com/watch?v=-eykunyQ55Y&t=632s) 核心管理能力聚焦于精准提问（Prompting）、评估输出偏见与实际业务落地；跨学科融合正成为关键，如计算机科学向医学、商学与航空安全等领域深度渗透。 - [14:48](https://www.youtube.com/watch?v=-eykunyQ55Y&t=888s) “成绩单”仅记录单项模块技能，而“学位文凭”代表大学的信誉背书与完整思维塑造；高校若仅在技能证明维度竞争，必将败于更快捷廉价的商业培训机构。 - [18:09](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1089s) 高等教育的长远价值超越孤立的技能掌握，其核心在于经历复杂的学业淬炼，学会如何立足、全身心承诺并坚持到底，完成步入职场前的人格构建。 - [20:34](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1234s) 赋能一线教师必须正视其职业焦虑与心理脆弱；推动 AI 教学创新需立足“新颖性、广泛采纳度与实际影响”三支柱，并紧密链接产业实际需求以提供职业安全感。 - [25:39](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1539s) 教师角色需从单纯的信息传递者转变为学习引导者；比起自上而下的行政要求，同行成果展示与配套的激励机制更能有效激发教师自我转型的内生动力。 - [27:46](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1666s) 知识创造与更新周期缩短至以月为单位，学生在课堂上的实时事实核查促使教学打破传统单向权威灌输，演变为平等的师生双向对话与“知识共创”。 - [32:51](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1971s) 高校部署 AI 底层架构需防范三大技术风险：构建私有化 Embedding 模型以防供应商倒闭锁定、平衡模型调用 Token 成本，以及确保研究与教学数据的全生命周期隐私安全。 - [37:02](https://www.youtube.com/watch?v=-eykunyQ55Y&t=2222s) AI 与地理空间大数据（GeoAI）的融合打破了资源壁垒，将算力与洞察力延伸至全球供应链优化、极端天气预警等缺乏技术话语权的领域，解决关乎人类福祉的复杂问题。 - [38:34](https://www.youtube.com/watch?v=-eykunyQ55Y&t=2314s) 警惕 AI 工具导致个体封闭自守（insular），必须依托国际合作与打破孤岛的团队协作，才能最大化激发人类群体的创新潜能。 - [39:27](https://www.youtube.com/watch?v=-eykunyQ55Y&t=2367s) “绿地开发”效应表明，缺乏传统系统与利益包袱的后发国家或新兴机构，拥有极高的创新敏捷度，能够跳过过渡阶段直接采用最新架构实现跃迁式创新。 - [40:58](https://www.youtube.com/watch?v=-eykunyQ55Y&t=2458s) 长远来看，碳基生命与硅基智能将走向深度共生融合，在合理的引导与治理下，充裕的智能将成为提升人类福祉与健康水平的关键力量。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=-eykunyQ55Y&t=4s) 圆桌论坛开场与嘉宾面临的现实挑战 论坛主持人 Anjelina Belakovskaia 宣布会议正式开始，并指出本次讨论的主题虽然是“重塑课堂”，但探讨的深度绝不仅限于“如何使用具体的 AI 工具”，而是要深入探讨 AI 对教育本质究竟意味着什么，以及整个高等教育体系正在发生的根本性变革。随后，各位受邀嘉宾依次介绍了各自在高校的职责以及当前应对 AI 浪潮时面临的最核心挑战。 来自德克萨斯州立理工大学（Texas Tech University）的信息技术副总裁兼首席信息官 Lin Zhou 率先发言。他介绍自己的团队负责全校的信息技术基础设施，致力于利用技术提高教学水准、加速学术研究、拓宽学生的职业发展路径并提升学校的教育品牌。他指出，目前学校面临最紧迫的挑战，在于如何迅速甄别并有效采纳能够同时在上述四个维度起到加速推动作用的技术。 来自摄政大学（Regent University）的副教务长 Jason Baker 随后分享了作为一所以教学为主的私立教会大学所面临的现实困境。他幽默而尖锐地指出，当前最大的挑战其实是“经济成本问题”。每当走进供应商展区，所有展商都会声称采用其 AI 工具可以节省时间和资金；从长远来看确实可能如此，但在短期内，学校实际上是在“双重买单”——不仅要为采购技术软件付费，还要投入巨资为教职员工进行再培训、技能重塑（reskilling）和业务流程重塑（reimagining）。整个高等教育界，尤其是以教学为核心的大学，都必须在 AI 时代直面这一财务与运营上的严峻考验。 圣路易斯大学（Saint Louis University, SLU）科学与工程学院院长 Gregory Triplett 接着发言。他强调作为一所天主教耶稣会大学，其核心挑战在于如何在剧烈的技术变革中“忠于育人使命”。学校秉持“为了更大的福祉”（for the greater good）这一理念，注重对“全人”（the whole person）的培养。面对 AI 这种在内容生成、设计和流程简化上具有颠覆性潜力的技术，传统教育假设受到了极大冲击。学校必须在帮助学生掌握前沿技术技能的同时，确保他们成长为具有健全人格、能够担负家庭与社会责任的完整之人。 主持人 Anjelina Belakovskaia（来自亚利桑那州立大学雷鸟全球管理学院）补充了她的观察：对很多高校而言，最棘手的挑战是如何让全校师生“步调一致”。由于不同教师对 AI 的认知和采纳水平参差不齐，学校必须重新审视课程设置的本质。考虑到大学本科是长达四年的培养周期，教育者绝不能只看眼前的日常操作，而必须着眼于未来四到五年后学生真正需要的核心能力。 ```mermaid flowchart TD accTitle: AI时代高校面临的核心挑战 subgraph Challenges["高校推进 AI 面临的关键挑战维度"] A["技术选型与加速应用<br/>(Lin Zhou / Texas Tech)"] B["双重经济成本负担<br/>(Jason Baker / Regent)"] C["全人培养与使命坚守<br/>(Gregory Triplett / SLU)"] D["师资步调协同与课程重构<br/>(Anjelina Belakovskaia / ASU)"] end ``` - **德克萨斯州立理工大学**：技术战略围绕“提高教学标准、加速科研、拓宽学生就业渠道、提升学校品牌”四大支柱展开。 - **摄政大学指出的短期经济矛盾**：不仅要支付高昂的技术采购许可费用，还要支付庞大的人员培训与流程重构成本。 - **圣路易斯大学的办学理念**：强调在技术自动化浪潮中坚守天主教耶稣会“为了更大的福祉”及“全人教育”的育人传统。 --- ### [04:40](https://www.youtube.com/watch?v=-eykunyQ55Y&t=280s) AI 时代的能力重塑：持久能力与即时技能的平衡 围绕主持人提出的核心议题——“哪些技能在 AI 时代是不可或缺的，而哪些技能目前被过度高估”，Jason Baker 展开了深入剖析。他将能力划分为“持久能力”（durable skills）与“即时技能”（immediate skills）两大类。持久能力包括沟通能力、冲突解决、问题剖析、创造性思维、批判性思维以及适应力等。这类能力具有终身价值——如果去问毕业十年的校友在大学学到了什么，很少有人能记住某个具体的数学方程式或考试题目，他们通常会说“我学会了像工程师或护士一样去思考”，或者“我获得了看待世界的全新视角”。然而，即时技能则是学生毕业当下获取工作敲门砖所必需的。他以自己三个二十多岁的孩子为例谈到，孩子们在大一时最抢手的热门技能，到他们大四毕业时往往就已经过时被淘汰了。 Jason Baker 指出，教育研究表明人类的持久能力“绝不能脱离具体情境而在抽象中教授”。要让软性思维真正生根，就必须在教授具体、有形的即时技能的过程中进行渗透。关键在于教师要深入到学生的思考流程之后，引导学生进行元认知（metacognition）练习，不仅要追问“正确答案是什么”，更要追问“你是如何得出这个答案的、它背后意味着什么、未来将走向何方”。 ```mermaid flowchart LR accTitle: 持久能力与即时技能的结合教学模式 subgraph EducationModel["AI时代能力培养架构"] A["即时技能 (Immediate Skills)<br/>工具操作 / 当下就业技能"] --> C["具体任务实践"] B["持久能力 (Durable Skills)<br/>批判思维 / 沟通 / 适应力"] --> C C --> D["元认知与过程洞察 (Metacognition)<br/>追问求解过程与底层逻辑"] end ``` Gregory Triplett 从圣路易斯大学的实践出发呼应了这一观点。他强调学校特别重视耶稣会传统中的“明辨”（discernment）能力，但他也风趣地指出明辨的潜在风险在于“如果整天都在空想明辨而不采取实际行动，也会成为问题”。因此，现代课堂的目标是让学生运用 AI 工具迅速完成繁琐枯燥的日常任务，从而释放出精力去践行核心价值观。圣路易斯大学反复强调“以人为本的 AI”（human-centered AI），确保学生在与算法交互中绝不丢失自我身份认同。关于用人单位的需求，他认为雇主虽然无法清晰用语言定义，但他们极度渴求懂得技术伦理与“治理”（governance）的人才。在他们的定义中，治理绝非单纯的合规或死板的规章制度，而是“提出正确问题的能力”。这种深度思考与发问的能力无法被内置在任何软件平台中，是求职者真正的核心竞争力。 Lin Zhou 则从深远的历史视角看待当前的 AI 焦虑。他指出人类历史上总是在重复类似的命题：2000多年前当书写技术刚刚被发明时，苏格拉底就曾面临过同样的困惑，甚至在当时对文字书写持反对态度，担忧它会破坏人类的记忆与深层思考。今天 AI 的演进正是历史的再现。在 Lin Zhou 看来，教育应当始终聚焦于“何以让人成为人”（what makes humans human）的特质上——人类大脑拥有超过100万亿个突触连接，人类拥有热情、专注、奉献与承诺等美好品质，教育的根本机遇就在于借此技术转折点去探索并确立人类自身的独特价值与命运。 - **持久能力的核心清单**：沟通、冲突化解、问题解决、创造性思维、批判性思维与高适应力。 - **企业对治理能力的实质需求**：并非机械的合规核对，而是对算法及其产出“提出正确问题”的思辨本领。 - **历史镜像案例**：苏格拉底在文字出现初期曾对书写形式对人类思考方式的侵蚀提出强烈质疑。 --- ### [10:32](https://www.youtube.com/watch?v=-eykunyQ55Y&t=632s) AI 管理能力与跨学科培养：学位的演变与交叉实践 Anjelina Belakovskaia 总结指出，AI 不仅已经到来，而且已全面融入现实，就像农业机械化彻底改变了农业生产方式一样。她自豪地提到，亚利桑那州立大学（ASU）已连续11年被评为全美最具创新力大学，在 AI 探索方面走在最前列。在雷鸟全球管理学院，他们将关键能力的落脚点放在对 AI 的管理上——包括管理 AI 工具、提出恰当精准的 Prompt 与问题、评估 AI 输出内容的准确性与偏见，以及将生成的结果有效应用到实际业务中。针对当前教育界从高度专业化知识转向通识与综合能力的趋势，她向嘉宾抛出了一个关键问题：未来的学位认证形态将会如何演化？传统的本科学位、研究生文凭是否会被微证书（micro-credentials）或技能认证所取代？ Gregory Triplett 结合圣路易斯大学科学与工程学院的创新举措作出了回应。作为全校技术交叉的“神经中枢”，工程学院正跨越传统学科壁垒，与各个专业展开深度融合。他们将计算机科学的力量外溢到其他学院：与商学院联合开设交叉课程培养商科学生；与医学院深度合作推出了“AI 与医学”专属项目，目前正在为 476 名住院医师提供系统的 AI 赋能培训；此外还联合航空学院设立了“AI 与航空”方向，专门研究利用智能技术保障交通运输安全。通过这些全方位的交叉渗透，传统的专业学院正在转变为面向各行业实际场景输出 AI 融合能力的综合孵化平台。 - **亚利桑那州立大学创新成就**：连续11年蝉联全美最具创新力大学第一名（US News）。 - **AI 时代的核心管理能力矩阵**： - 精准有效的问题定义与提问能力（Prompting & Inquiry） - 对 AI 生成内容的批判性评估与质检能力（Result Evaluation） - 结合实际场景落地应用的整合执行力（Applied Execution） - **圣路易斯大学的跨学科 AI 落地项目**： - **医学领域**：与医学院合作开展“AI与医学”项目，已覆盖培训 476 名住院医师。 - **商学领域**：计算机科学系与商学院联合共建跨学科课程。 - **航空领域**：聚焦航空交通安全分析与智能算法应用。 ### [12:43](https://www.youtube.com/watch?v=-eykunyQ55Y&t=763s) AI 的实际应用案例与跨学科赋能 作为技术中心的负责人，Gregory Triplett 阐述了如何帮助非技术背景的师生理解并使用 AI。他指出，我们没必要让每个人都去“窥探幕后”（peek behind the curtain），因为帘幕后坐着的是硬核程序员；对于哲学、艺术或历史等文科学者而言，他们更关心的是“我该如何应用它”。因此，技术专家的职责是快速帮助他们上手，并展示具体的实际应用场景。 Gregory 分享了一个极为震撼的真实案例：在儿童诱拐案件中，犯罪分子有时会将照片发布在网络上；研究人员利用开源工具结合 AI 技术快速识别照片中的地理位置，随后将情报同步给执法部门，从而成功解救被拐儿童。当非技术人员看到这样具体的案例时，思想的“灯泡”就会瞬间被点亮。此外，在航空领域，AI 也被用于评估着陆场地的环境条件是否适合飞机降落。Gregory 强调，圣路易斯大学作为一所教会背景的高校，保障儿童安全与生命安全与学校的核心价值观高度契合。他充满激情地表示，AI 在安全领域能解决无限多的现实问题，而技术专家的使命就是搭建桥梁，帮助所有人掌握利用 AI 解决问题的能力。 ```mermaid flowchart LR accTitle: AI 跨学科赋能与实践落地流程 subgraph NonTech["非技术学科应用"] A["文科/艺术/哲学师生"] --> B["聚焦实际应用需求"] end subgraph PracticalUse["具体应用落地"] B --> C["儿童搜救：图像定位与执法联动"] B --> D["航空安全：降落条件实时评估"] end subgraph CoreGoal["核心价值目标"] C --> E["契合核心价值观 / 解决社会实际痛点"] D --> E end ``` ### [14:48](https://www.youtube.com/watch?v=-eykunyQ55Y&t=888s) 成绩单与学位的本质区别：高等教育不可替代的育人价值 Regent University 的 Jason Baker 接着回应了有关认证和课程设置的问题。他极力赞同跨学科协作，指出 AI 技术具有无所不在和跨职能的特性，绝不能将其局限在某一个学院或系所中。文科生固然需要补充技术技能，但技术人员同样迫切需要文科学者的人文思辨能力。 在谈到微证书（micro-credentials）与学位时，Jason 提出了一个深刻的区分——“成绩单”（transcripts）与“学位文凭”（diplomas）截然不同： - **成绩单**：仅用于记录单门课程的成绩表现，或是证明学生掌握了某些模块化的技能。 - **学位文凭**：代表大学的信誉背书（imprimatur）。它宣告着这所大学不仅见证了学生的思考、掌握与成长，更将他们塑造成了该领域的专业人才。 Jason 直言，如果大学仅仅在“技能证明/成绩单”的维度上竞争，必败无疑。因为社会上有大量商业培训机构能够以更快、更便宜的方式完成纯技能的培训与评估。大学的使命在于塑造人格、培养人类并建设社会。尽管堆叠式证书、徽章等微证书很有价值，但它们远远不够。企业界（例如在 LinkedIn 上）不仅看重技能标签，近年来对正式学位的要求反而有所增加，因为雇主渴望寻找经过完整系统化思维训练的人才，而不仅是一张罗列技能的清单。 ### [18:09](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1089s) 培养完整的人：超越技能掌握的教育长征 Texas Tech University 的 Lin Zhou 结合自身经验分享了大型公立大学的视角。德克萨斯理工大学拥有约 13 个学院、超过 42,000 名在校生（涵盖本科到博士阶段），同时开展了大量的线上项目、微证书与认证课程。Lin 透露，自己在计算机系教授量子计算，在日常中深切感受到不少家长的犹豫与动摇——有相当大比例的家长正在重新评估大学学位的实际价值与回报。 Lin 深刻地剖析道，大众常常将学位的价值狭隘地等同于“技能掌握”，但这只是教育极小的一部分。回顾一个人接受过的 20 年甚至 25 年的高等教育历程（尤其是攻读博士），其真正的核心收获绝非几门课程的孤立技能，而在于： 1. **学会如何立足与收获（how to earn）**； 2. **学会如何作出承诺并全身心投入（how to commit）**； 3. **学会如何坚持到底、贯彻执行（how to follow through）**。 完成一整套复杂的学业组合，本身就是一次深刻的人格淬炼。Lin 强调，大学阶段很可能是年轻人在步入职场之前，所能经历的“最后一次自由形式的学习”（last free-form learning）。教育工作者的神圣职责，就是在技能和人文素养双重维度上，让学生做好应对复杂现实世界的准备。 ### [20:34](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1234s) 赋能教职员工：直面 AI 带来的脆弱感与职业焦虑 主持人 Anjelina Belakovskaia 将焦点拉回一线教师身上。在 AI 重塑课堂的过程中，教师们既要面对院长布置的新课任务，又要主动摸索如何重构课程以传授新技能。现场调查显示，在座的许多听众也都是为人父母者，正共同面临着“如何引导下一代”的巨大挑战。Anjelina 由此向 Gregory 提问：究竟该如何为高校教师提供培训与赋能？ Gregory 坦言这是“最难回答的问题”，并剖析了教师群体深层的心理脆弱与职业焦虑： - **失业恐惧**：教师们普遍担心，一旦全面拥抱 AI，大学可能就不再需要他们了。这种对失去养家糊口能力的真实恐惧，正是 AI 在高等教育界未能如预期那般迅速普及的核心阻力。 - **走出舒适区**：要促使教师涉足令他们感到极度脆弱的领域，需要极高的领导力智慧。 为了破局，Gregory 提出了推动跨学科创新的“三支柱”原则，要求教师在探索 AI 与新教学模式时必须满足以下三点： 1. **新颖性（Novelty）**：是否创造了前所未有的新事物？ 2. **广泛采纳度（Adoption）**：是否具有可被大众广泛接受和落地的可行性？ 3. **影响力（Impact）**：是否能产生切实的实际影响？ ```mermaid flowchart TD accTitle: 驱动 AI 教学与跨学科创新的三大支柱 subgraph ThreePillars["创新三支柱"] N["1. 新颖性 (Novelty)"] A["2. 广泛采纳 (Adoption)"] I["3. 实际影响 (Impact)"] end N --> Success["经济发展机遇与职业安全保障"] A --> Success I --> Success Success --> ActiveFaculty["教师安心投入 AI 跨学科探索"] ``` Gregory 指出，只有具备这三个要素，才能创造真正的经济发展机遇，从而让教师拥有足够的职业安全感去大胆尝试。此外，他强烈呼吁高校必须将产业界拉得更近（"bring industry closer"），绝不能闭门造车在真空中设计解决方案。在谈到高校颠覆性变革给院长带来的巨大领导力压力时，现场更是幽默地调侃起“院长们需要喝威士忌来减压”。 ### [25:19](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1519s) 以人为中心的教育转型与教师恐惧 接力这一话题，Jason Baker 进一步提炼了全场讨论的核心主线——“以人为本”（centering things on humans）。他重申，这场变革关乎的从来不是冰冷的技术本身，而是活生生的人。高校必须正视并理解广大教职员工心中存在的真实恐惧，唯有站在理解人的立场上，才能引导教育体系完成平稳转型。 ### [25:39](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1539s) 重塑教师角色：从信息传递者到学习引导者与激励机制 嘉宾们探讨了高校教师面对 AI 浪潮时的职业焦虑与角色转变。如果教师将自己的职责仅仅定义为“信息的传递者或搬运工”，那么 AI 的确会暴露出这种模式的局限性，因为这并非教师时间的最优价值所在。大学管理层需要引导教师理解，在 AI 时代，教学的内涵远不止于信息灌输。然而，要求原本在自身领域处于顶尖水平、在课堂上“掌控全场且自认最聪明”的学者去经历再培训（reskilling），往往会让他们体验到一种能力不足的挫败感。 为了推动实质性变革，大学领导层必须提供配套的激励机制与成长机会。比起一遍遍重复既有的教学套路，尝试创新需要投入更多的时间与精力，因此学校必须对付出努力的教师给予实质性奖励。在推广方式上，同行展示远比自上而下的行政命令更有效：当一位教师自发向同事展示“我刚学会了这个工具”、“我刚刚构建了一个新应用”或“你看我在课堂上这样互动”时，所激发的灵感与火花，是开办一整天自上而下的 AI 培训工作坊都无法比拟的。 ```mermaid flowchart TD accTitle: Faculty Transformation in the AI Era accDescription: Flowchart illustrating the transition from traditional teaching to peer-driven AI adoption and institutional incentives. subgraph OldModel["传统教学模式"] A["信息传递者 / 搬运工"] --> B["重复性讲义与既定知识灌输"] end subgraph Transition["面临挑战与转型"] C["知识更新周期骤缩"] --> D["顶尖学者面临技能重塑的挫败感"] end subgraph SupportSystem["赋能与落地路径"] E["学校激励机制与时间投入奖励"] F["同行成果展示 (Peer Showcase)"] G["师生双向协作与动态共创"] end OldModel -.-> Transition Transition --> SupportSystem ``` --- ### [27:46](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1666s) 知识半衰期缩短：课堂教学向“双向对话与共创”演变 Lin Zhou 从物理学和教学动态的角度指出了传统课堂与现代知识演进的脱节。在过去，大学教学极度依赖记录下来的既有知识：物理课讲解 300 年前发现的牛顿第三定律，或是 100 年前的相对论，知识更新的速度极为缓慢（堪称“冰川级速度”）。但在今天的 AI 时代，知识创造的周期已经从过去的数百年骤降到以“月”为单位，很多最新进展甚至在开课前刚刚发布。这要求高校在三个维度进行革新： - **建立常态化的教师专业发展体系**：构建能够持续、可重复地帮助教师再培训和技能升级的流程。 - **重构课堂支持机制**：打破“讲课-测验-固定答案”的死板框架，包容课堂中即时更新的知识动态，引导师生在课堂上“共创知识”（co-create）。 - **建立实时反馈与去权威化沟通**：如今学生在课堂上会使用 AI 工具对教授进行实时事实核查（Fact-check），“且往往有 70% 到 80% 的概率学生是对的”。教学不再是单向权威灌输，而是转变为平等的双向对话。 --- ### [31:13](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1873s) 跨学科实践社群建设：雷鸟商学院的 AI 推广范式 Anjelina Belakovskaia 分享了亚利桑那州立大学（ASU）雷鸟全球管理学院的落地经验。教师面对 AI 时不仅有失去工作的担忧，更有“工具太多不知道该选哪个、最佳实践是什么”的困惑。为此，雷鸟学院 AI 与新兴技术委员会在凤凰城市中心创办了面向全校的 AI 实践社群（Community of Practice），并每月组织跨领域的“AI 科技沙龙”（AI Tech Talks）。 沙龙打破了学院壁垒，广泛吸纳教授、教职员工、学生以及校友参与。讨论议题非常务实且全面，涵盖工具实操、版权与知识产权归属（如果用 AI 开发了产品，成果归谁所有）、AI 在教学与科研中的应用，以及如何提升个人生产力。这种开放、共享的交流平台成功促使许多最初持怀疑态度的教师转变态度，亲口承认“我原本完全不相信，但现在彻底被折服转化了”。 --- ### [32:51](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1971s) 全球化视角下的 AI 架构：嵌入模型依赖、成本控制与数据隐私 随着讨论转向全球竞争与跨国团队协作，Lin Zhou 深入剖析了教育与科研机构部署 AI 底层架构时的关键风险。针对现场观众普遍使用 Chatbot 却对底层技术细节不甚了解的情况，他重点分析了三大核心技术考量： | 考量维度 | 核心风险与行业现状 | 应对与建设建议 | |:--- |:--- |:--- | | **嵌入模型锁定 (Embedding Lock-in)** | 向量知识库极度依赖特定 Embedding 模型。若提供该模型的商业公司倒闭或下架服务，原先向量化的全部知识资产将彻底作废，一切必须从零重建。 | 建立自主受控的私有化 Embedding 模型，避免单一第三方商业依赖。 | | **模型算力与 Token 成本** | 前沿商业大模型性能虽强（参数量达 2 到 3 万亿，远超人脑约 900 亿神经元），但调用成本高昂（如每百万 Token 需 5 至 10 美元），大学难以大规模无限制承受。 | 结合实际预算，选择性能“足够好”、性价比高且易于获取的开源/商用模型组合。 | | **数据主权与全生命周期安全** | 师生在提问和交互过程中，核心数据与研究成果直接向外传输，存在知识资产外泄及长期留存不可控的隐患。 | 参照过去 20 年硬盘厂商从数十家整合至两三家的历史规律，从全生命周期、隐私保护和数据可控性出发构建自主 AI 平台。 | --- ### [37:02](https://www.youtube.com/watch?v=-eykunyQ55Y&t=2222s) 时空大数据赋能：突破资源局限解决全球未解难题 Gregory Triplett 介绍了圣路易斯大学与美国国家地理空间情报局（NGA）的紧密合作。NGA 负责绘制全球及其所有活动的图谱，在时空坐标系下，任何带有时间和空间属性的信息都是极具价值的数据点。圣路易斯大学将 AI 与地理空间技术（Geospatial Technologies）深度融合，已广泛应用于全球供应链优化、极端天气模式预测以及精算评估等多个领域。 Gregory 驳斥了“AI 会让世界问题过度简化”的悲观论调，他坚信世界上面临着无限多的复杂问题。过去几代人往往只聚焦于有资源群体的需求，而忽视了大量缺乏资源和技术话语权的弱势群体。AI 与地理空间技术的结合，恰恰能够打破这种资源壁垒，将前沿算力与洞察力延伸到原本未被赋能的领域，解决真正关乎全人类福祉的长远挑战。 ### [38:34](https://www.youtube.com/watch?v=-eykunyQ55Y&t=2314s) 跨国协作与打破孤岛：AI 时代防范封闭化 Gregory Triplett 探讨了全球视角下的技术发展机遇，强调通过挖掘时空数据（time and space data）所带来的洞见，我们能够真正帮助世界各地背景各异的群体。他指出，要实现这一点，前提是必须建立国际合作伙伴关系并开展团队协作。他表达了自己最大的担忧：AI 可能会导致所有人变得封闭自守（insular），每个人都只坐在自己的电脑前使用 AI，进而放弃了团队合作。 因此，无论未来如何发展，唯有团队的共同努力才能真正激发人类最优秀的一面。将问题置于全球视野下，意味着大家必须打破信息孤岛（silos），主动与同行交流互动。结合地理空间信息与人工智能（GeoAI）便是一个绝佳范例，因为它跨越了地域与传统边界的限制。 ### [39:27](https://www.youtube.com/watch?v=-eykunyQ55Y&t=2367s) 绿地开发效应：后发国家的跃迁式创新与教育启示 Jason Baker 对团队协作与人性价值的观点表示高度赞同（"Amen to all the human and working together aspect"），并提出了“绿地开发”（green field development）的视角。他解释道，当一个组织或国家无需拆除沉重的遗留系统（legacy systems），而是能够从零开始、白纸作画时，往往会涌现出全新的愿景和创新解决方案。 这一现象在很多历史进程中都曾得到验证。例如在移动支付领域，部分国家由于没有发达的传统银行遗留体系，直接跨越到了点对点的移动端交易；同理，缺乏固定电话网络的地区，其移动电话的普及速度反而更快。因此，国际上有许多值得借鉴的经验——那些没有历史包袱的实体和国家，虽然可能起初并非行业领导者，却拥有极强的创新敏捷度。Baker 补充道，在线教育是他的研究领域，在亚利桑那州立大学（Arizona State University）等名校崛起之前，凤凰城大学（University of Phoenix）等机构率先开创了全新模式，正是因为它们没有需要改造的传统生源与体制束缚。 ```mermaid flowchart TD accTitle: 绿地开发与创新跃迁机制 subgraph LegacyConstraints["传统系统约束（存量包袱）"] A["成熟遗留系统 / 既有利益链"] --> B["高昂的改造与试错成本"] B --> C["渐进式缓慢革新"] end subgraph GreenfieldAdvantage["绿地开发优势（从零开始）"] D["无传统基础设施束缚"] --> E["直接采用最新技术架构"] E --> F["跃迁式创新 (Leapfrogging)"] end subgraph Examples["典型跃迁范例"] F --> G["无传统银行系统 ➔ 直奔移动支付 / P2P"] F --> H["无固话网络 ➔ 快速普及移动通信"] F --> I["无传统校园包袱 ➔ 开创新型在线教育模式"] end ``` ### [40:58](https://www.youtube.com/watch?v=-eykunyQ55Y&t=2458s) 终局预测：碳基与硅基共生及 AI 的未来福祉 在圆桌会议仅剩两分钟的总结环节，主持人 Anjelina Belakovskaia 引入了一个新的思考维度。她提到，自己除了是副教授外，还是一名职业国际象棋棋手（professional chess player），目前正参与顾问工作，为各地的资优儿童（gifted and talented kids）开发国际象棋项目。作为棋手，习惯于“多看几步棋”，因此她邀请嘉宾用一两句话展望尚未被大众清晰看见的未来。 Lin Zhou 预测，未来我们将见证数字与人类之间日益深化的共生关系（symbiosis）。人类作为碳基生命，身体约 17% 由碳构成，而如今硅基技术展现出越来越强大的力量。随着碳基与硅基融合的不断增强，他大胆预测，到 2030 年当大家再次参加这一会议时，现场或许会出现技术原型甚至是“行走的混合物种”（walking hybrid species）。随后嘉宾补充强调，只要引导得当，“充裕的智能（abundant intelligence）将成为全世界的福祉”，并极大改善人类的健康水平。Belakovskaia 随后向全场致谢，并邀请有更多疑问的听众在台下会场外继续交流，圆满结束了本次圆桌讨论。 ## Notable quotes > [00:04](https://www.youtube.com/watch?v=-eykunyQ55Y&t=4s) 不仅要为采购技术软件付费，还要投入巨资为教职员工进行再培训、技能重塑和业务流程重塑。 — Jason Baker > [04:40](https://www.youtube.com/watch?v=-eykunyQ55Y&t=280s) 持久能力绝不能脱离具体情境而在抽象中教授。 — Jason Baker > [04:40](https://www.youtube.com/watch?v=-eykunyQ55Y&t=280s) 治理绝非单纯的合规或死板的规章制度，而是提出正确问题的能力。 — Gregory Triplett, Jr. > [04:40](https://www.youtube.com/watch?v=-eykunyQ55Y&t=280s) 教育应当始终聚焦于何以让人成为人的特质上。 — Lin Zhou > [14:48](https://www.youtube.com/watch?v=-eykunyQ55Y&t=888s) 如果大学仅仅在‘技能证明/成绩单’的维度上竞争，必败无疑。 — Jason Baker > [18:09](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1089s) 大学阶段很可能是年轻人在步入职场之前，所能经历的‘最后一次自由形式的学习’。 — Lin Zhou > [20:34](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1234s) 高校必须将产业界拉得更近，绝不能闭门造车在真空中设计解决方案。 — Gregory Triplett, Jr. > [27:46](https://www.youtube.com/watch?v=-eykunyQ55Y&t=1666s) 如今学生在课堂上会使用 AI 工具对教授进行实时事实核查，且往往有 70% 到 80% 的概率学生是对的。 — Lin Zhou > [39:27](https://www.youtube.com/watch?v=-eykunyQ55Y&t=2367s) 当一个组织或国家无需拆除沉重的遗留系统，而是能够从零开始、白纸作画时，往往会涌现出全新的愿景和创新解决方案。 — Jason Baker > [40:58](https://www.youtube.com/watch?v=-eykunyQ55Y&t=2458s) 充裕的智能将成为全世界的福祉，并极大改善人类的健康水平。 — Lin Zhou ## Mentioned Anjelina Belakovskaia, Lin Zhou, Jason Baker, Gregory Triplett, Jr., 亚利桑那州立大学 (Arizona State University), 雷鸟全球管理学院 (Thunderbird School of Global Management), 德克萨斯理工大学 (Texas Tech University), 摄政大学 (Regent University), 圣路易斯大学 (Saint Louis University), 美国国家地理空间情报局 (NGA), 凤凰城大学 (University of Phoenix), 苏格拉底 (Socrates), LinkedIn, GeoAI ## Video description Anjelina Belakovskaia — Associate Teaching Professor, Chair of the AI & Emerging Technologies Committee, Arizona State University, Thunderbird School of Global Management Lin Zhou — Vice President, Chief Information Officer, and Executive Director of AI and Quantum Computing, Texas Tech University Jason Baker — Associate Provost & Senior Technology Strategist, Regent University Gregory Triplett, Jr. — Dean of the School of Science and Engineering, Saint Louis University From chatbots that coach students to AI that adapts in real-time to learning styles, education is being reinvented. In this panel, educators and edtech leaders explore how AI is improving learning outcomes while also raising critical questions. What does effective, ethical implementation look like? Can AI address equity gaps, or will it reinforce them? Learn what’s working today and how leaders are preparing for tomorrow’s AI-driven classroom. [Source](https://www.youtube.com/watch?v=-eykunyQ55Y)

[0:04] Jason Baker: 不仅要为采购技术软件付费，还要投入巨资为教职员工进行再培训、技能重塑和业务流程重塑。 [4:40] Jason Baker: 持久能力绝不能脱离具体情境而在抽象中教授。 [4:40] Gregory Triplett, Jr.: 治理绝非单纯的合规或死板的规章制度，而是提出正确问题的能力。 [4:40] Lin Zhou: 教育应当始终聚焦于何以让人成为人的特质上。 [14:48] Jason Baker: 如果大学仅仅在‘技能证明/成绩单’的维度上竞争，必败无疑。 [18:09] Lin Zhou: 大学阶段很可能是年轻人在步入职场之前，所能经历的‘最后一次自由形式的学习’。 [20:34] Gregory Triplett, Jr.: 高校必须将产业界拉得更近，绝不能闭门造车在真空中设计解决方案。 [27:46] Lin Zhou: 如今学生在课堂上会使用 AI 工具对教授进行实时事实核查，且往往有 70% 到 80% 的概率学生是对的。 [39:27] Jason Baker: 当一个组织或国家无需拆除沉重的遗留系统，而是能够从零开始、白纸作画时，往往会涌现出全新的愿景和创新解决方案。 [40:58] Lin Zhou: 充裕的智能将成为全世界的福祉，并极大改善人类的健康水平。

[44:41] Jason Baker: A degree... the institution is saying something more than you have achieved a collection of skills. It's actually giving its imprimatur to say that they have formed you into somebody in a particular discipline... if universities only want to compete on the transcript side, we'll lose. [40:51] Gregory Triplett, Jr.: abundant intelligence will be a blessing to the world

[0:04] Welcome to our session. Uh it's going to be very  excited. I'm uh glad to be here and have amazing [0:11] panelists with me. So uh the discussion is going  to be redesigned in the classroom. But we actually [0:18] going to uh go behind uh or above the idea of just  how do we use AI tools and more into what AI means [0:28] for education. how everything is changing and so  with that I would like uh each of you to introduce [0:36] uh what you are doing what is your role at the  uh at the university and also briefly say you [0:43] know like what is the most critical challenge you  are dealing with right now so Lynn uh let me get [0:52] started so uh my name is Lynn I'm um um a vice  president at Texas Tech university. Uh my team [1:00] responsible for uh all the information technology  in our campus and uh we leverage technology to [1:08] raise education standard accelerating research  broadening student career pathways and also [1:15] raise our educational brand. One of the most um  I would say presenting challenge for us today is [1:25] to quickly identify and adopt technologies which  will accelerate all on those four fronts I just [1:33] talked about. Okay, thank you. Good morning.  My name is Jason Baker. I'm associate provost [1:39] at Regent University. We're a private faith-based  teaching university in Virginia Beach. currently [1:45] expanding our scholarship and research areas as  well. Thank you for coming out this morning. We [1:50] really appreciate it. Um, in terms of our biggest  challenge, I'd actually say it's the economics [1:56] because you go down to the vendor display and  every single vendor will tell you that you if you [2:00] adopt their AI, it will save time and money. In  the long run, that's true. In the short run, it's [2:08] actually costing us twice. First, we have to pay  for the technology. Then we have to pay for the [2:13] people for the reskilling for the reusing for the  reimagining. So that is a challenge that we are [2:20] facing individually but I think also as a sector  how does higher education particularly teaching [2:26] universities handle the economics of college and  an AI world. Thank you. Good morning. My name is [2:34] Gregory Triplet. I'm the dean of the school of  science and engineering at St. Louis University. [2:40] It is a Catholic Jesuit institution. Um, our  biggest challenge is staying true to our mission. [2:49] Uh, we have a saying at St. Louis University  is for the greater good and incorporating AI [2:55] technology is challenging many of our assumptions.  We think about educating the whole person. And how [3:03] can you do that with a disruptive technology  like AI and all of the potential that it has [3:09] to streamline for creation uh for design? And so  one of the things that we grapple with is how do [3:16] we help our students who are really coming to have  a skill set and also become a whole person because [3:23] they will have families. How do you do all of  that while adapting to changes on the technology [3:30] forefront? That's our biggest challenge.  Great. Thank you. So, let me also briefly [3:36] uh mention about what I'm doing. So, my um as uh  as I said, my name is Angelina Belikovska. I'm [3:42] associate teaching professor at Thunderberg School  of Global Management, Arizona State University. [3:46] And our biggest challenge is to figure out how to  bring everybody on the same page because a lot of [3:54] uh faculty you know like at the different level of  adoption of AI and we also have to to rethink what [4:01] exactly we are teaching. So um what are the skills  that our students will need going forward and [4:08] uh if you think about there is a long time of the  university you know the undergraduate degree is [4:13] four years. So we really have to think not about  what we doing you know like day-today basis but [4:18] where it's going to lead our students and this  is going to be my first question. Um what are the [4:24] skills that students will need to possess uh that  will be absolutely necessary in the uh a era of [4:33] AI and also which skills we are uh overestimating  right now that actually not going to be necessary. [4:40] Um who's going to start? I can start. [clears  throat] So skills are interesting because there [4:47] are the durable skills and then there are  the immediate skills. The durable skills, [4:52] communication, conflict resolution, problem  solving, creative thinking, critical thinking, [4:58] adaptability. These things last. In fact, if you  ask most alumni say 10 years after they graduate, [5:05] what did you learn in college? Most of them will  say something along the lines of, I learned to [5:10] think like an engineer. or a nurse or I had a new  view of the world, they won't necessarily remember [5:17] the individual uh equation or or um test or  assignment that they did. At the same time though, [5:27] you need the immediate skills in order to be  able to get the job. Because if you ask those [5:31] same people when they are applying for college,  what are you looking for? Many of them are saying, [5:37] I'm looking to get a degree so I can get a job. So  how do we do this? How do we take these long these [5:44] durable skills and then also these immediate  skills which may in fact change constantly? [5:51] I have three kids in their 20s. For each one of  them, the skills that were hot when they were a [5:57] freshman were different than the skills when they  graduated. How do you handle that? And the way you [6:03] handle that is to actually work together. Here's  what we know from the research. durable skills, [6:10] those human skills are actually not taught in  the abstract. If you really want them to land, [6:16] you need to teach them while doing the tangible  things. So the key there is to actually engage [6:23] with the students so that as they learn the  immediate skills, you are getting behind it. [6:29] You are looking at the process. You are getting  them to do metacognition. You're getting them to [6:33] observe why they're doing it. Not just what's  the right answer, but how did you get it? What [6:39] does it mean? Where does it go in the future?  Okay, awesome. And Gregory, do you want to [6:46] elaborate and give your perspective? Uh, at St.  Louis uh University, again, we tend to focus on [6:55] um uh technologies that that improve society and  quality of life. And so in all of our programs, [7:02] we really do think about discernment. That's  a really big thing in the Jesuit community. Um [7:09] the I think there's a problem with discernment is  that you can just you know you can think about it [7:13] all day but you don't actually take an action.  So that could be a little bit of a problem. Uh [7:17] and so when we think about you know what is the  next generation of classes going to look like and [7:22] in fact that's now is how do we allow students  to use these toolkits that allows for them to [7:30] do mundane tasks a lot quicker and give them the  opportunity to to impart their core values. And [7:36] so we try to to always emphasize in everything  we do human- centered AI human- centered AI. [7:43] We just keep driving that point home over and  over again. And it's one of the I think salient [7:48] features that we have at Slooh is because we  never want the the the student to lose their [7:52] identity and that's like critically important. So  in terms of the workforce development, we feel we [7:58] have an advantage because when our students are  really thinking about the ethical implications, [8:04] companies want that like and they don't really  define it, but they do want to understand a little [8:09] bit more about governance, right? And governance  in our in our term, the way to define it is not [8:15] compliance. It's not saying what you should  do or what you shouldn't do. It's about asking [8:19] the right question. And so we really do for all  four years or however many years you're at Slooh, [8:26] we always incorporate let's the thinking part of  this like that's the skill set that cannot get [8:31] lost and it cannot be embedded in a in a platform.  So those are the skills that our employers want. [8:38] Those are the skills that our employers desire  even if they don't articulate it very well. Right. [8:44] Thank you Lynn. Um I felt uh the same question  is repeating uh in history. Um 2,000 years ago [8:55] uh when writing was first invented uh  Socrates was facing the same question right [9:03] um what skills are important and how the education  will evolve and his answer actually at that point [9:09] was very different. Um so you can you can you  can do some history research. Um and uh he [9:16] was actually against writing at that point. Mhm.  And uh just as AI is reinventing itself um today [9:25] it's not even it's not new but it's reinventing  today. We are facing the same question right and [9:32] um uh my I don't think we have the right answer  and I think we do have some conjectures. My view [9:40] is that um the education would continue focus on  what makes humans human, right? We have how can [9:53] we make good connections in our brand for those  synapsis, right? 100 trillions or more of those. [9:59] And how can we focus on the basic humans needs,  right? the reading, writing, passion, dedication, [10:07] commitment, all those good qualities human  do really well. So, uh not a complete answer, [10:14] but I felt this is a moment where we can have  a huge opportunity to determine or change our [10:22] destiny. How to make humans human. I like that.  But um uh if you think about we're uh way beyond [10:32] the point where uh you know like the question is  if AI is here to stay. So we we all know that it's [10:38] it's already happening. Uh it's implemented those  are new tools. If you're thinking you know like [10:43] how agriculture has changed um so just now the  question is what are the skills what they have [10:49] to be and for example you know like I'm fortunate  because Arizona State University has been 11 years [10:54] in a row number one in uh innovation so uh we  have uh AI you know like all in front of us [11:02] uh but we really work on um where's uh where human  fit in and the skills and in our case you like [11:12] the skills should be the management of the AI,  the management of the tools, uh the ability to [11:17] understand how to ask the proper questions,  how to evaluate the results, how to apply [11:24] uh you know what uh what has been um what has  been produced. Uh but now if you think about the [11:32] education, so you get all those diplomas, you get  certificates. Uh so you get all those credentials [11:38] but now you know like uh it seems like the turn  is toward general knowledge instead of specialized [11:46] and so what exactly from the educational viewpoint  uh you're going to see in uh in in a sense of are [11:54] we going to stay with diplomas undergraduate  degree graduate degree is it going to be like [11:58] micro uh credential certificates or anything  else uh Gregory you want to start? So as as [12:06] a a dean of school of science and engineering,  right? So we have tremendous opportunity because [12:10] I feel like the nerve center kind of exists  in my school. Um we've actually expanded our [12:17] um sort of involvement in other programs. So  computer science is in my school and we've [12:21] created new programs in the school of business.  We basically have uh partnerships there where we [12:26] teach classes with their business students. Um we  have a partnership with our school of medicine. [12:31] We actually have an AI and medicine program  and we're actually teaching 476 residents. Um, [12:37] we have AI and aviation where we're thinking about  transportation safety. In some ways, it's just an [12:43] AI dot dot dot. I mean, that's almost almost kind  of how it feels or an AI plus. It's like whatever [12:49] you want, you need to just kind of incorporate  it. I do think that from my perspective, again, [12:54] just kind of being housed in a center, we try to  help people understand how to use these tools. [13:00] uh so that they don't have to peak behind the  curtain because when you peek behind the curtain [13:05] those are the hardcore you know programmers they  really understand the tools and it's a little [13:09] bit different when you have folks who are in  philosophy or who are in arts or in history they [13:14] don't understand it they're just trying to figure  out how do I apply it so it's kind of up to us [13:19] to bring them up to speed as quickly as possible  and give them practical applications say this is [13:24] exactly what you can use this for so one example  that we give and is a pretty common example [13:30] is that when kids are abducted um in many cases,  I mean it's obviously horrible for the parent, [13:36] it's horrible for the community, but in sometimes  those pictures are posted online and it's really [13:42] just it's a horrible scenario. So, we actually  have researchers that will take those photos [13:48] and they use open source to identify the location  using AI. Uh and then we send the authorities and [13:55] then the kids are rescued. And so when you provide  that practical example, the light bulb begins to [14:01] turn on. They're thinking, "Oh, that's a practical  application." Um, we use AI and aviation. We think [14:06] about um the landing space, like is the are the  conditions appropriate for landing a plane and [14:12] people don't think about those kinds of things,  but it's all about safety. Um, you can kind of [14:18] get the idea why I'm really excited about what  happens, right? what's going to emerge is because [14:23] safety and looking out for our kids is very much  in line with our core values at a faith-based [14:29] institution. And so I think that there is an  infinite number of problems that we can solve in [14:34] that space. And our goal as as um a technologist  is to help the other people understand how to use [14:41] AI to get there. All right. Uh so let's see. Thank  you uh Greg Jason. So let's let's talk about this [14:51] uh from from your point of view. So what what  do you do? What kind of certificate programs or [14:57] anything you uh uh you see at your university?  So I want to echo what Gregory said in that we [15:04] also have developed courses that are used across  the university in different programs because we [15:10] think that AI shouldn't just be the purview of  one particular school or department. It's too [15:16] pervasive and crossf functional of a technology  and we want people bringing their different [15:22] perspectives in there. We want to just not look  at humanities people and say, "Well, those degree [15:28] programs just need tech skills." No, the tech  people need the humanities thinking every bit as [15:35] much as the humanities people need to understand  the technology aspect of it. But to go to the [15:40] issue of credentials specifically, I think we need  to distinguish between transcripts and diplomas [15:48] because transcripts are the thing that documents  individual classes and their performance or maybe [15:54] in the more advanced with micro credentials sort  of buckets of skills that you have demonstrated [16:00] mastery of. But a degree when you're awarded a  bachelor's degree, a master's degree or something [16:06] else, the institution is saying something more  than you have achieved a a collection of skills. [16:15] It's actually giving its impromoner to say that  they have formed you into somebody in a particular [16:25] discipline that they're basically vouching that  you have they have seen you think they have seen [16:33] you master and they are sending you out into  the world. I think if universities only want to [16:42] compete on the transcript side, we'll lose. One,  as Gregory said earlier, the role of a university [16:52] is bigger than just skill development. It actually  is about formation and shaping humans and building [17:01] societies. But the other thing is there are other  companies and organizations that can do pure play [17:09] skill development and assessment and transcription  better, faster and cheaper than universities can. [17:18] So yes, we need to leverage all the capabilities  with micro credentials and we've done badging and [17:24] certificates and stackable credentials  and the whole thing and those are good, [17:29] but they're actually not sufficient. We need to  make sure that we can watch the students think and [17:36] that we can actually say we are sending somebody  out with a degree not just a transcript of skills [17:44] and it's actually what companies are still looking  for. look at most company lists they will look at [17:49] two things you will see skills but even in recent  years there's been an increase in the number of [17:56] companies for instance on LinkedIn that actually  want the degree as well because they're asking [18:02] for something more than just the skills I thank  you Lyn do you have something to add so we have [18:10] uh so in Texas Tech we have uh about 13 colleges  uh over 42,000 uh students uh which goes all the [18:18] from undergraduate to PhDs and we also have  a online program. We do the certificates [18:24] uh micro credentials all those good stuff. Um one  of the things um uh resonate with me a lot uh I [18:33] also teach by the way I also teach in uh computer  science I teach quantum computing. So um what I [18:39] have witnessed was that uh many of the parents  uh are hesitating sending their kids to school. [18:46] A large actually this is not a small number it's a  very large percent of parents are rethinking and I [18:51] think what they have been reacting is the value of  the skill which were conveyed through the degree [18:59] or the master degree program whatever the program  is. Um but the big part is missing as many of you [19:05] sitting in the room you know you gone through at  least 20 years of education if not more right if [19:10] you go through the PhD that's even more extensive  the way you gone through that journey is not just [19:16] for the skill mastery that's only a small piece of  it the big piece of we gone through the 20 20 25 [19:25] years of education is we learned how to earn. Mhm.  We also learned how to commit. We also learned [19:36] how to follow through, right? Just to be able to  complete that journey is not a simple one course, [19:43] two course. It's a whole portfolio you had  to go through. I think that I go back to my [19:48] earlier comment. What makes the humans human  is our education system or to deliver. We have [19:57] been delivering that. Now we have to rethink go  through the in the lenses of degree programs, [20:04] certificates, micro credentials. How can we  continue support the students to do that? [20:10] After all, this is probably the last free  form learning they could experience before [20:16] they go to the they get a professional job. So  it's our job it's from education point of view [20:21] it's our job to make sure they are very well  prepared for the complex world skill-wise but [20:28] also from the humanity side as well. Right.  Thank you Lyn. Well you know like it's it's [20:35] very interesting. So what we hear you know from  from panelist uh a a lot of is happening. So [20:42] uh we have you know like diff rethinking of  education uh different um opportunities for [20:50] the students different pathways different  new classes collaboration uh and um cross-d [20:55] disciplinary uh work uh now I'm also faculty and  from our perspective faculty has to deal with it [21:09] so faculty are the one who have to redesign and  and sometimes you know like it comes from the dean [21:13] hey you're teaching the new class but but uh but  actually a lot of times it's faculty themselves [21:20] trying to figure out even in the classes that  I teach how do I change how do I apply those [21:25] um changes to to to have students learn the skills  that they have to have and meanwhile you know like [21:34] who's helping faculty so how do you upskill  faculty to uh to meet the challenge Uh that's [21:41] the next question. So we now I just be before  before we proceed. How many of you have school [21:47] age kids? Anybody? All right. Okay. Awesome. Well,  [laughter] so so we we all facing that problem. [21:55] We we have how do we take the kids to get where  they supposed to be and how do we change what we [22:02] teach everything that we do and so Gregory back  to you. uh how do you help faculty like me and [22:09] you know like many many others? Thank you for  asking the hardest question. Yeah. [laughter] [22:19] If here's here's how I view it, right? So  as a leader I try to recognize the the most [22:28] uh important aspect of being human. Usually  you want to provide for your family. That's [22:34] what you want. You want to have the opportunity  uh to create um you know safe home environment. [22:43] Faculty are feeling very challenged right now and  they're thinking man if I do move to AI that means [22:50] the university won't need me. I mean it is very  real and and this is why it hasn't been widely [22:57] adopted as quickly as I would like. Um, and so  the question that I keep posing to myself is how [23:04] do I get faculty to engage in something where  they feel most vulnerable. That is a skill that [23:11] requires a lot of thinking. Um, and so my approach  is to really focus on innovation to think about [23:20] what is new that you can do that has not been  done before. Um, is there something that you is [23:26] there something new that can be widely adopted?  That's the second part. And can it have impact? [23:34] If you don't check all those three boxes, don't do  it. And it requires that students, not students, [23:41] faculty get outside of their comfort zone and to  think about interdisiplinarity in a in a brand new [23:48] way. And the reason those three things novelty,  adoption, and impact are the three pillars that [23:55] I drive because it ensures that we have economic  development opportunity. And if we're creating [24:01] economic development opportunity, then people  feel very safe to pursue interdisiplinary work [24:08] and AI. And they also have an opportunity to have  gainful employment. And as a dean right now in [24:15] this midst of higher education being completely  turned upside down, it is requiring a new set [24:21] of skills as a leader that I was not trained to  have. And so it is it is always challenging but [24:27] to listen and have conversations and look for  impact in every way. My last point is that I [24:34] want to mention this. How do we get faculty to  do it? We really need industry to come closer. [24:41] We need industry to come very very close because  industry understands all the problems that exist [24:46] and we don't want to create solutions in a  vacuum. Right? Just bringing industry closer [24:51] is to me one of the most important uh actions  that we can take as a community. Bring them [24:56] closer to high red. Okay. That's an awesome  point and you know based on what you said [25:02] next time I will ask you know like what it  takes to upskill deans [laughter] whiskey. [25:11] [laughter] No, you just says that it's  challenging and I'm sure that you're [25:15] not the only ones that there are a lot of  deans who are dealing with it. But Jason, [25:19] now it's your turn. To pull forward a theme that  you've heard a lot from all the panelists is this [25:27] idea of centering things on humans. That this  isn't really about the technology, it's about [25:31] the people. There is legit fear, as Gregory said,  among faculty who think that the whole purpose of [25:39] bringing this in is essentially to suck all of  their talent out so that we can replace them. [25:46] And unfortunately, if a faculty member's  view of their teaching responsibility is [25:53] simply to be a mere conveyor or transmitter  of information, well then okay, I'm not saying [26:00] that we want to replace them, but that's not  really the richest and best use of their time. [26:08] We need to help faculty understand what the  responsibilities that they have in an AI era [26:16] which is more than just information transmission.  We need to do that not just by providing training [26:22] but providing incentives to them. If they have  to reskill, if they have to learn new things, [26:29] that's a difficult task for anybody. It's  especially difficult for people who are top [26:33] of their game, who can hold court in a classroom  where they are literally the smartest one in the [26:38] room. And now you're basically asking them  to feel incompetent to learn a new skill [26:43] or to even rethink what their role is as  a faculty member. As university leaders, [26:49] we have to provide incentives and opportunities  for them to learn and grow. We have to reward them [26:56] when they invest the time because to do something  different takes more time than than to do the same [27:02] thing over and over again. And the final piece  I'd say is that faculty showing one another, [27:10] showcasing what they've done, hey, I just learned  this. Hey, I just built this new thing. Did you [27:14] see what I did in the classroom here? Is far  more powerful, at least in our experience, [27:20] than anything that we do top down. We can  run workshops all day long and we do. But [27:27] when you have one faculty member in a particular  discipline show a colleague what they've done or [27:33] how they've rethought of something, you get the  light going off and the sparks flying in a way [27:38] that you don't have with a top-down professional  development workshop on AI reskilling. All right, [27:46] Len, I'm going to change the word later.  That's okay. That's okay. Thank you. Um, [27:57] so I I I speak from uh I I have observed of  my uh many of my colleagues uh as we uh go [28:04] through the teaching process. uh if you really  think about for the audience in the room if [28:09] you really think about what's happening in the  classroom when we talk about teaching there was [28:15] uh pseudo definition of what teaching is  all right look at all the class we have [28:20] usually those classroom always have a textbook  if not have a lecture slides someone created and [28:26] uh in our world I'm a physicist right we  are teaching like Newton third law which was [28:32] invented 300 years ago right and uh If you talk  about theory of relativity which was invented [28:38] about 100 years ago. So largely teaching has been  conducted based on recorded knowledge. The cycle [28:49] of knowledge refreshment has been glacial speed if  not that but very slow right but that's not what [28:56] happening in the classroom today. Dynamics I have  a kid in the classroom as well. um the knowledge [29:02] creation that cycle has been reduced from h  measured in hundreds of years in a matter of [29:09] months right we have seen that acceleration and uh  so this brings the challenge which my colleagues [29:15] mentioned over here is as a professor as a teacher  in the classroom what is your role and how can we [29:22] enable you right if the knowledge just literally  created a few months ago how could you quickly get [29:27] that knowledge and relay that back to students  probably it's not relay actually probably could [29:32] be co-create with the students together right so  I would say that probably there are three things [29:38] we need to do uh in the colleges uh and we are  doing that number one is we uh uh conducting [29:46] professional development for our faculties  so that we have a process repeatably right [29:54] reskill our faculty so that's number one number  two is to support our faculty in the classroom [30:02] because the knowledge creation cycle is so short  many times the knowledge get refreshed literally [30:09] right before the class happened. So understanding  that the what has been transpired in the classroom [30:15] between the faculty and student could change and  that usually not the traditional teaching is when [30:20] you teach it you give a quiz it is it is what it  is it's not going to change but that's not how [30:25] dynamic happen. So that's number two and number  three is to constantly get feedback from from [30:33] the outside world and and also from the students  because guess what many times students actually [30:39] fact check the professors in the classroom real  time right see people nodding head they bring [30:44] up hat they fact check real time when you say  something that no professor is not right this [30:49] is what whatever whatever is I would say probably  80% 70% they are right okay so I think we need [30:56] creating that supporting system right uh not  recognize the authoritative source but recognizing [31:03] that the teaching the dynamic will change  more more more of more become a conversation [31:08] is a two-way street so let me pause over here I  think there are a lot more we can talk about okay [31:13] great so yes it's you know like uh if you think  about faculty absolutely need the upskilling and [31:21] uh they also have to feel comfortable yes I  understand that one is uh the fear to lose a [31:26] job but another is how do I use those tools like  there are so many different tools how do I know [31:32] which one to use what is the best practice what  is the best way to use so for example um what we [31:37] do at Thunderbird and I'm uh the chair of AI and  emerging technologies committee at uh Thunderbird [31:43] um we organize monthly AI tech talks so uh we  invite faculty staff students and uh alumni so [31:51] we are like really open and we invite people from  all um uh ASU uh colleges. So technically we are [32:00] uh we created the downtown Phoenix um community of  practice AI community of practice and we share the [32:06] ideas we share the the best practices we share you  know the uh the use of the tools we we talk about [32:14] uh copyright uh who is uh uh if if the product  is developed. So who is uh who it is belongs to [32:21] uh how to use it in teaching how to use it in  research how to use it in uh um productivity [32:28] you know etc and I think you know like it makes  people to convert this is exactly it's not even [32:33] my word it's their words that like hey you know  like I did not believe and now you know like you [32:37] converted you converted but it's really you know  like it's happening because uh it's it's a real [32:43] it's a real tools and real knowledge now so with  at and since I started to talk about Thunderbird [32:51] we are a global school and if you think about we  have a lot of conversation about everything what's [32:57] happening uh in the United States but if we look  at industries so we see a lot of multicultural [33:07] multi-reional uh teams working together uh we also  see that different countries move at the different [33:14] speed of AI adoption and So I would like to uh  ask Lynn, you know, like now it's we're going [33:21] to start with you. So uh what do you see globally  that we can either use, you know, like uh here in [33:29] the states or um we can compete with um so there's  a lot uh I I repeat what Greg said. It's very hard [33:40] question. We have nine minutes left. I know. Let  me let me just be very succinct. So um how many [33:46] of you actually use chatbot but maybe who who has  not who has not really used chatbot I don't see a [33:52] single hand right nobody wants to admit how many  of you actually embedded your knowledge raise your [33:58] hand okay so if you embedded a knowledge if you  embedded your knowledge you know how critical [34:03] the embedded model is right what happened when  embedded model goes away or the company disappears [34:15] You start from scratch, right?  Everything you done is gone, right? So, [34:20] do you have a private controlled embedding  models? Okay, so I'll start from here. So, [34:28] the technology obviously has becomes very  sophisticated. Um, models become very capable. [34:36] Which one is the best? Which one is the worst?  I think they are all good enough these days, [34:41] right? They become very they become very good.  There is one front talking about the benchmarking, [34:48] but the other part is talking about the open and  the private models, right? Commercialized models [34:54] versus open models. Um, now you can you continue  seeing the open models becomes very capable, [34:59] right? The number of parameters is over probably  two trillion parameters number now and uh is [35:07] certainly more than the number of neurons we have  in our head right we have 90 billion neurons there [35:11] this is 2 probably three trillion you know three  trillion parameters very sophisticated uh models [35:19] the question is how can we get those models  um those capabilities to our faculty to our [35:27] researchers and there's a cost component of it,  right? If you try to get the best to the faculty, [35:34] whether you somebody mentioned economic, right?  You maybe you mentioned, right? Can you afford it, [35:39] right? Some of those costs were expensive,  right? $5, $10 for 1 million tokens. So, [35:45] think about that computer, thinking about that  technology landscape. Think about picking the [35:50] models which is good enough, you can afford, but  also is accessible. But very importantly thinking [35:57] about the privacy the data control your data goes  out when you ask a question your data goes out are [36:06] you sure the data is going to be controlled  you still have control in that data are you [36:12] sure whatever you have done is going to be with  you for many years I started as a technologist [36:17] in design drives you know 20 years ago there  were 20 30 disr companies today how many there [36:26] two three what happened I go back to my first  question what happened you use the embedding [36:33] model today there are many there are 20 30  of those what happened over 20 years there [36:38] are only two left where your knowledge goes  right so this is something we are very mindful [36:45] uh we are building as we creating that  platform to enable our faculty enable [36:50] our researchers right we're thinking about  the whole pipeline the entire life cycle [36:54] privacy capabilities and those things uh since you  I would advise you if you're not doing it at least [37:02] to think about it. Thank you Lyn. Uh Gregory so  briefly to you. So in in St. Louis University we [37:11] um we're very close to a a federal agency called  the National Geospatial Agency. Uh and they map [37:18] the entire world and all of its activity. And  anytime you have a um uh uh a time and space, [37:27] that's a data point. There is tremendous insight  from looking at data that has time and space. And [37:36] so we use that in all kinds of applications. And  this is where using AI and geospatial technologies [37:44] can have a real impact uh across the globe because  you can think about problems that are supply chain [37:50] related. You can think about weather patterns.  You can think about uh actuarial impact. I mean [37:56] it is I think that to me I I certainly feel that  there is this concern that everything is going [38:05] to get simplified and I believe total totally  differently. There are infinite problems that [38:12] exist and they've existed for generations and what  we've done is we've only focused on problems that [38:20] are important to those who have resources and  there are a lot of people out there who don't [38:25] have empowerment, who don't have access. So to  me, I think that this is going to emerge into an [38:34] uh an opportunity set that really allows us to  help our brothers and sisters who don't look like [38:40] us across the world. And that is by thinking about  the insights that come from time and space data. [38:46] You can only do that when you have international  partners. You can only do that when you talk as [38:51] a team. My biggest fear is that AI is going to  cause everybody to be insular. They're going to [38:58] use AI from their desktop and they're going to  stop working as a team. And so whatever you do, [39:05] the team effort is going to really unearth  the best that we are as humans. And I think [39:11] to think about it in a global context is kind of  your question means that we have to get outside [39:16] of our silos and start having conversations with  our peers. And using geo and AI is a good example [39:22] because it doesn't have any limitations. All  right. Thank you. And I think just to add on [39:27] to that, amen to all the the human and working  together aspect of this. Um I think one of the [39:34] things we can also take away is when you see  green field development um where you don't have [39:41] to undo all of the legacy systems that are  within an organization but can start fresh [39:46] or start clean. New visions emerge, new solutions  emerge. We saw this for instance in mobile payment [39:54] systems in countries where they didn't have the  legacy banking system that we had and they went [40:00] straight to peer-to-peer mobile transactions and  even just mobile phones uh had faster penetration [40:07] in countries that didn't already have the landline  system that was already there. So there's actually [40:13] a lot for us to learn internationally as we  see countries and entities and organizations [40:20] that don't have to undo an existing legacy  system. Maybe they're not the leader in a space, [40:27] but that gives them the ability to innovate new  things. Online learning was my research area. And [40:34] before the Arizona States and the others rose up,  it was the University of Phoenix's and the others. [40:39] These were these were schools that didn't have  an existing constituency that they were trying [40:44] to repurpose. They saw new opportunities and they  built new models and then others proceeded from [40:50] that. Awesome. All right. And so as we have two  minutes left so we we need to to hurry up. So [40:58] uh I will take us into different direction. uh  what I did not mention so in addition to being the [41:03] uh associate professor I'm also uh professional  chess player and uh right now you know like I just [41:10] got the consulting position where I'm helping  to build um chess program for uh gifted and [41:17] talented kids anywhere uh and as a chess players  we definitely think a few moves ahead and so my [41:26] question for you maybe you know like one sentence  or like very very short So how do you see the [41:31] future? Futures that we don't see yet. So Lynn  if I could take just one one or two sentence I [41:37] think the uh we have been seeing the increased  level of symbiosis between digital and uh human [41:45] right we are carbon based uh species 17% of body  is carbon now we see the silicon side takes more [41:51] and more power. So I think going forward um that  integration between silicon and the carbon species [41:59] is going to continue enhance and maybe 2030 when  we come back I know to this conference we probably [42:06] will see some uh prototypes or even some walking  hybrid species in the room. That's my prediction. [42:15] If we do this right abundant intelligence will  be a blessing to the world. If we do this right, [42:22] we can improve human health. All right. Well,  that sounds great and I really appreciate all [42:29] your wisdom sharing with us. I'm sure that if  you have uh questions and probably you do, so we [42:35] don't really have time uh to take it on stage,  but we will be available outside. So, please, [42:41] you know, like come and ask questions. We would  be happy to talk. Thank you very much. Thank you.