Utah Tech Scholars Urge Embodied Learning to Protect Human Agency

Panel 6 - 3: Educating for Agency and Embodied Understanding in the Age of AI — Ai4

Ai4

Universities must replace superficial AI adoption slogans with hands-on, experiential problem-solving to prevent algorithmic optimization from displacing autonomous human reasoning.

AI-assisted summary of Panel 6 - 3: Educating for Agency and Embodied Understanding in the Age of AI — original source. Check the original for full context.

Source published:

Detailed notes

## Background Published on YouTube by the Ai4 channel on August 18, 2026, this talk features Emma Prendergast, Assistant Professor of Philosophy and Director of the Humanities Center at Utah Tech University. According to the page description, the presentation explores how educators can preserve human agency and embodied understanding in the age of artificial intelligence, challenging conventional slogans about responsible AI and advocating for experiential learning.

## TL;DR Drawing on educational theorists John Dewey and Paulo Freire, Emma Prendergast examines how the rapid spread of AI risks reducing education to mechanistic optimization and passive information consumption. She deconstructs popular institutional aphorisms surrounding AI usage, pointing out that treating LLMs as mere shortcuts for brainstorming or editing bypasses the essential cognitive struggle required to develop independent thought. To preserve self-governed human agency, Prendergast argues for shifting away from passive instruction toward embodied, polytechnic learning and collaborative Vertically Integrated Projects (VIPs) that ground education in real-world, hands-on problem solving.

## Atmosphere & style The presentation balances rigorous philosophical critique with witty, self-aware storytelling, using humor—such as tales of intrusive smart fridges and bizarre AI-generated Christmas lyrics—to put the audience at ease. Prendergast transitions smoothly from lighthearted classroom anecdotes to sharp dissections of technocratic culture, biohacking, and military-driven technology. The talk carries an engaging, grounded, and collaborative energy that invites educators and industry leaders to move beyond panic or passive compliance toward authentic human connection.

## Key takeaways - [00:04](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=4s) The rapid encroachment of AI into everyday tools and classrooms challenges educators to protect critical thinking, autonomy, and human agency rather than merely reacting to tech rollouts. - [02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) Educational autonomy—the capacity for self-governed authorship of one's life—is deeply rooted in ancient Athenian philosophy and requires recognizing human embodiment rather than treating minds in an abstract vacuum. - [02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) John Dewey's experiential education and Paulo Freire's critique of the passive 'banking model' demonstrate that true understanding requires active, embodied engagement with real-world problems. - [06:18](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=378s) Educators' critical skepticism is reinforced by recognizing that AI systems are fundamentally funded and developed as instruments of military power and corporate profit rather than purely benevolent tools. - [09:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=571s) Machine learning relies on optimizing a single objective function, making its mathematical framework fundamentally incapable of navigating conflicting human values and nuanced ethical deliberation. - [09:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=571s) The hyper-focus on machine optimization and biohacking risks training humans to operate like efficient machines rather than cultivating deeper social, moral, and emotional maturity. - [09:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=571s) The proliferation of 'AI slop' in academic work demonstrates how prioritizing frictionless output robs students of the essential cognitive struggle required for independent thought. - [12:51](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=771s) Slogans such as 'use AI as a tool, not a crutch' fail pedagogically because they offer no clear guidance on boundaries, while offloading drafting and editing eliminates critical reasoning practice. - [12:51](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=771s) Pressuring educators to mandate AI tools is counterproductive, as the prerequisite for effective technology use is strong foundational critical literacy rather than rote software instruction. - [17:32](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1052s) Deterring AI adoption by exposing errors and hallucinations is an obsolete strategy because models improve rapidly; the core concern is the displacement of human reasoning. - [17:32](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1052s) AI literacy must reframe the educational relationship away from autonomous AI agents directing passive users toward self-governed human agents critically employing assistive tools. - [19:09](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1149s) Developing effective educational approaches requires inclusive, cross-disciplinary dialogue through models like Vertically Integrated Projects (VIP) linking faculty, industry mentors, and students. - [19:09](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1149s) Embodied polytechnic education and experiential problem-solving foster active human empathy and critical agency that generative artificial intelligence cannot replicate or perform.

## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=4s) Introduction and the Ubiquity of AI in Education

Emma Prendergast introduces her presentation on research co-authored with her colleague Emily Fitzgerald, waving to her in the audience and joking that any especially difficult questions should be directed to Emily. Both are philosophy and humanities professors at Utah Tech University in St. George, located a couple of hours east of Las Vegas along the highway. Their research investigates how education can adapt to the sudden ubiquity of artificial intelligence while safeguarding long-standing educational goals: critical thinking, self-determination, and embodied human agency. Educators across all tiers currently face shifting goalposts, vague institutional leadership, and a rapidly transformed learning environment.

Prendergast humorously illustrates this saturated environment by walking through the relentless encroachment of AI features into everyday tools. From ChatGPT suggesting classroom discussion prompts to new workflows in Claude, automated summarizers for student discussion posts, Grok, and even AI built into smart refrigerators, the technology has inserted itself everywhere. For millennials and older generations, this pervasive intrusion feels reminiscent of Microsoft's Clippy—an assistant that can be helpful, but often pops up in places where it was never needed. This is the world inhabited by today’s students and the future workforce. In response, Prendergast and Fitzgerald examine how educators can preserve deep humanistic values rather than simply reacting to product rollouts.

```mermaid flowchart TD accTitle: Educational Paradigm Comparison accDescription: Contrast between the banking model of education and the embodied problem-posing approach in the presence of AI.

subgraph Banking ["Banking Model of Education"] B1["Instructor / AI Depositor"] -->|"Transfers static content"| B2["Passive Student Container"] B2 -->|"Rote memorization & efficiency"| B3["Standardized Optimization"] end

subgraph Embodied ["Embodied Problem-Posing Model"] E1["Human Agent"] <-->|"Relational dialogue"| E2["Physical & Social World"] E2 <-->|"Inquiry & self-governance"| E3["Autonomy & Embodied Growth"] E4["AI Technologies"] -.->|"Instrumental support"| E1 end ```

---

[02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) Theoretical Foundations: Autonomy, Embodiment, Dewey, and Freire

The core theoretical architecture of the research rests on the interplay between autonomy and embodiment:

- **Autonomy:** Defined as the capacity to be the self-governed author of one's own life, make independent decisions, and engage in critical introspection. This objective dates back to ancient Athens, where Plato framed education as "the art of turning the soul" to foster genuine freedom of thought. - **Embodiment:** Defined as the relationality between physical sensory experiences, the mind, the surrounding material world, and everything that constitutes human nature across individual and communal scales. True autonomy cannot exist in an abstract vacuum; it requires acknowledging that humans are physical beings embedded in a material world.

To anchor these concepts pedagogically, Prendergast draws on educational theorist John Dewey, who famously asserted that education is not mere preparation for life, but is life itself. Dewey pioneered progressive experiential schooling, arguing that real learning occurs through tangible, hands-on interactions rather than rote memorization. For Dewey, schooling must nurture continuous growth, self-control, habits of inquiry, reflective judgment, and cooperative problem-solving so individuals become capable of directing their own lives.

Complementing Dewey, the presentation invokes Brazilian educator Paulo Freire and his seminal work, *Pedagogy of the Oppressed*. Freire contrasted the oppressive "banking" concept of education—where passive students are treated like empty containers in which teachers deposit information—with "problem-posing" education. Working with illiterate peasants in post-World War II Northeast Brazil, where literacy was a prerequisite for voting, Freire built curricula around politically charged, lived vocabulary like *slum*, *wage*, and *land*. Problem-posing education treats learners as active subjects in the world who participate directly in their own liberation by grappling with real-world problems. Together, Dewey and Freire provide an educational framework asserting that genuine understanding is intrinsically embodied and aimed at cultivating self-determination.

---

[06:18](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=378s) Anchoring Biases: Unhinged Christmas Lyrics and Military Tech

Before diving deeper into current debates, Prendergast acknowledges that she and Fitzgerald approached AI with inherent skepticism, a mindset common among educators. Both incorporate AI into their teaching, but their baseline attitudes were shaped by memorable initial encounters:

- **Prendergast’s First AI Encounter (December 2021):** Living in Los Angeles for the first time during the holidays, she wrote a deliberately corny Christmas song about hanging lights on palm trees and wishing for snow (later discovering Kermit the Frog had already covered similar ground). Her husband encouraged her to test an early GPT model to generate additional lyrics. After she resisted—joking that as a philosopher she had "regular intelligence" and did not need artificial intelligence—she tested the system. To her shock, the model generated an unhinged, hostile response concluding that it would be glad when she was dead. She could never replicate the output, but the bizarre result cemented a lasting wariness about what is hidden beneath the surface of large language models. - **Fitzgerald’s Anchoring Bias:** Fitzgerald’s outlook was shaped by venture capitalist Peter Thiel’s frank characterization of artificial intelligence primarily as a military technology. While mainstream marketing pitches AI as a benevolent engine of optimization and personal convenience, Fitzgerald and Prendergast remain acutely aware that these systems are fundamentally developed and financed as instruments of power, warfare, and corporate profit.

---

[09:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=571s) The Cult of Optimization: Biohacking, Pope Leo XIV, and AI Slop

When artificial intelligence is held up as the apex standard of intelligence, education risks conditioning students to optimize themselves like machines. Prendergast quotes scholar Maximilian Casey to highlight the fundamental misalignment between machine learning and human development: machine learning operates on a single objective function that an algorithm must maximize, meaning the only failure mode is failing to optimize that objective. Consequently, the mathematical framework lacks the language to navigate or negotiate the messy, conflicting objectives of different human beings—precisely the capacity liberal education seeks to nurture.

``` "In machine learning, there's a single objective that the algorithm must maximize... In such a framework, we do not have the language to talk about the conflicting objectives of different people." — Maximilian Casey ```

This obsession with mechanistic efficiency is mirrored in contemporary cultural phenomena such as Bryan Johnson’s "Don't Die" biohacking movement. While biohacking leverages biometric algorithms to extend lifespan and treat death as an engineering flaw, Prendergast warns that such efforts often train people to become better machines rather than better human beings. She contrasts this technocratic drive with ethical warnings from figures such as Pope Leo XIV, whose encyclical cautions against transhumanist movements that expand technical means without a corresponding maturation in social, moral, and human depth.

In higher education, this hyper-focus on output produces what has become known as "AI slop." Prendergast shares an instance where a student submitted a paper on the philosophical mediation between technology and human experience that was completely written by AI. While the prose was grammatically flawless and sounded superficially sophisticated, it was practically devoid of substance and engagement. The core tragedy was not merely academic dishonesty, but that the student’s urge to optimize time and grades completely crowded out the opportunity to pause and think through the ideas independently.

---

[12:51](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=771s) Deconstructing the AI Aphorisms: Tools vs. Crutches and AI Literacy

Prendergast critiques the superficial slogans and "AI aphorisms" frequently repeated in university professional development seminars, administrative policies, and faculty hallway conversations.

```mermaid flowchart LR accTitle: Three Flawed AI Aphorisms in Education accDescription: Breakdown of the three common slogans heard by educators and why each falls flat.

subgraph Aphorisms ["Common AI Slogans"] A1["'Use it as a tool, not a crutch'"] A2["'Teach them to use it well; it's irresponsible not to'"] A3["'Ban it and show students where it fails'"] end

subgraph Critiques ["Underlying Flaws"] C1["A crutch IS a tool; ignores boundary-setting"] C2["Shames teachers; conflates tech tricks with critical thinking"] C3["Models rapidly improve; misses the loss of agency"] end

A1 --> C1 A2 --> C2 A3 --> C3 ```

- **Aphorism 1: "Teach students how to use AI as a tool and not a crutch."** - *Critique:* A crutch is literally a tool whose appropriateness depends entirely on whether a person needs it. The slogan offers zero guidance on where to draw the boundary between helpful assistance and cognitive dependency. - *Classroom Implications:* In writing pedagogy, advocates suggest AI can solve writer’s block by generating ideas, but philosophy demands the hard cognitive work of generating one's own ideas. Using AI to outline papers robs students of the structural discipline needed for independent reasoning. Even basic sentence-level editing via AI risks bypassing the essential intellectual habit of carefully rephrasing thoughts to achieve clarity. - **Aphorism 2: "We must teach students how to use it well, and it is irresponsible not to."** - *Critique:* This framing often serves to shame faculty who resist adopting AI. It falsely assumes educators are formally trained in ethical AI literacy (when most receive little pedagogical training of any kind). - *Critique:* It discounts the reality that many professionals learn AI tools independently because they possess strong foundational skills. The prerequisite for effective tool use is not rote software training in school, but a robust grounding in critical thinking and literacy.

---

[17:32](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1052s) Flawed Resistance and Reframing Human Agency

The third common aphorism—that educators should ban AI entirely by demonstrating its flaws, such as hallucinations and incorrect facts—is increasingly futile:

- **Aphorism 3: "Discourage AI use by demonstrating its failures."** - *Critique:* This tactic becomes obsolete by the day. Early models could not browse the live web or provide accurate citations, but newer models fix prior limitations at an astonishing pace. - *The Real Danger:* The fundamental pedagogical threat is not whether an LLM outputs an incorrect fact, but that copy-pasting answers displaces autonomous human reasoning.

Prendergast concludes this section by outlining a path forward. AI literacy must expand beyond prompt engineering, cheat-detection, or spotting factual errors to include critical reflection on how artificial systems affect our embodied, material lives. Educational institutions must abandon the false dichotomy of pitting autonomous "AI agents" against passive "human users," and instead cultivate a framework centered on self-governed **human agents** utilizing **AI tools** collaboratively. Crucially, addressing these profound pedagogical shifts is too complex for any individual educator to resolve in isolation.

[19:09](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1149s) Inclusive Dialogue, Vertically Integrated Projects, and Embodied Polytechnic Learning

Rather than handing down prescriptive slogans that attempt to tell educators "here's what to do"—which invariably fails to resonate or hold up in practice—meaningful progress requires genuine, inclusive dialogue. This conversation must actively bridge the gap between teachers, learners, and industry partners to figure out what practical strategies actually work. To put this into practice, a proposal is underway for a Vertically Integrated Project (VIP) model. This framework establishes a structured, collaborative space uniting university faculty, industry mentors, and students across various stages of their academic careers, enabling older students to mentor younger peers while collectively examining how pedagogy and tech integration must continually evolve over time.

```mermaid flowchart TD accTitle: Vertically Integrated Project Mentorship Structure subgraph VIP["Vertically Integrated Project Framework"] direction TB A["Industry Partners & Faculty Mentors"] --> B["Advanced / Older Students"] B --> C["Entry-Level / Younger Students"] C -. "Inclusive Dialogue & Feedback".-> A end subgraph Outcomes["Pedagogical Goals"] D["Hands-on, Real-World Problem Solving"] E["Embodied Experiential Learning"] F["Critical Thinking AI Cannot Replicate"] end VIP --> Outcomes ```

Rooted in the polytechnic educational model practiced at Utah Tech University, this pedagogical shift centers on active, experiential learning. Rather than relying on static instructional methods that generative tools can easily mimic or automate, higher education must pivot toward hands-on problem-solving that directly engages with the physical world and real-world challenges. Such embodied, active coursework fosters deep critical thinking and human empathy—cultivating the vital dimensions of agency and understanding that artificial intelligence simply "cannot generate or act out for us."

- **Vertically Integrated Projects (VIP):** Multi-tiered mentorship networks linking university faculty, industry stakeholders, and cross-level student cohorts to test and refine educational approaches collaboratively. - **Polytechnic & Active Learning:** Pedagogical focus at Utah Tech University emphasizing assignments grounded in hands-on, embodied experiences rather than passive consumption. - **Human-Centric Critical Thinking:** Prioritizing authentic problem-solving and embodied inquiry that AI systems cannot simulate, replace, or perform on behalf of human learners.

## Notable quotes > [09:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=571s) In machine learning, there's a single objective that the algorithm must maximize... In such a framework, we do not have the language to talk about the conflicting objectives of different people. — Maximilian Casey

> [02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) the art of turning the soul — Plato

> [12:51](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=771s) Teach students how to use AI as a tool and not a crutch. — Emma Prendergast

> [12:51](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=771s) We must teach students how to use it well, and it is irresponsible not to. — Emma Prendergast

> [17:32](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1052s) Discourage AI use by demonstrating its failures. — Emma Prendergast

> [19:09](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1149s) cannot generate or act out for us. — Emma Prendergast

> [02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) Pedagogy of the Oppressed — Paulo Freire

> [09:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=571s) Don't Die — Bryan Johnson

## Mentioned Ai4, Emma Prendergast, Emily Fitzgerald, Utah Tech University, ChatGPT, Claude, Grok, Microsoft Clippy, Plato, John Dewey, Paulo Freire, Pedagogy of the Oppressed, Peter Thiel, Maximilian Casey, Bryan Johnson, Pope Leo XIV, Kermit the Frog

## Video description Emma Prendergast, Assistant Professor of Philosophy and Director of the Humanities Center, Utah Tech University

Autonomy, or the ability to be the self-governed author of one’s own life, has been held up as an educational aim since Ancient Athens. This talk examines how we can preserve human agency as fundamental aim and why embodiment must become an equally central educational concept as AI reshapes teaching and learning. The talk will challenge familiar slogans about responsible AI use and explain how technologies such as biometrics and biohacking can reduce human lives to data, optimization, and efficiency. Attendees will learn why ethical agency depends on recognizing our embodied needs, relationships, limitations, and capacity for empathy; and why educators, learners, and technology leaders must address these questions together. The talk will highlight experiential learning as a method for helping students learn and critically engage with AI without surrendering their humanity.

[Source](https://www.youtube.com/watch?v=iVjtWO7CxZw)

中文简报

犹他科技学者呼吁开展具身学习以保护人类能动性 大学必须用注重实操的体验式问题解决取代流于表面的AI采纳标语，以防止算法优化取代人类自主推理。 借鉴教育理论家John Dewey和Paulo Freire的学说，Emma Prendergast探讨了AI的迅速普及如何可能导致教育沦为机械优化与被动的信息消费。她解构了有关AI应用的流行机构格言，指出将大语言模型（LLM）仅作为头脑风暴或编辑捷径的做法，绕过了培养独立思考所必需的认知磨砺。为维护自主的人类能动性，Prendergast主张告别被动式教学，转向具身化、应用型的学习模式以及协作式垂直整合项目（VIP），使教育切实扎根于解决现实问题的动手实践之中。 面对 AI 在日常工具与课堂中的快速渗透，教育工作者亟需保护批判性思维、自主性与具身人类能动性，而非仅仅被动应对技术产品的推出。 (0:04) 教育自主性植根于古雅典哲学，并由杜威的经验教育与弗莱雷对被动“储蓄式模型”的批判进一步发展，表明真正的理解需要对现实世界问题进行积极且具身的参与。 (2:17) 机器学习依赖单一目标函数的优化，这种数学框架从根本上无法处理冲突的人类价值，过度追求机器优化与生物黑客技术存在将人类训练成高效率机器而非培育道德深度的风险。 (9:31) “将 AI 作为工具而非拐杖”等空洞口号未能明确界定认知界限，外包起草与修改会剥夺学生进行独立思考和批判性推理所必需的认知磨炼。 (12:51) 犹他科技大学倡导结合垂直整合项目（VIP）与理工科主动体验式学习，通过跨层级协作与现实世界问题解决，培养 AI 无法模拟或替代的具身理解与人类能动性。 (19:09)

Reading summary

Philosophy professors Emma Prendergast and Emily Fitzgerald argue that higher education must resist the technocratic urge to treat learning as an algorithmic optimization problem, warning that pervasive artificial intelligence threatens student autonomy unless pedagogy is actively anchored in embodied, experiential inquiry. Philosophical Foundations of Autonomy and Embodied Learning (2:17) Prendergast and Fitzgerald ground their pedagogical approach in the balance between personal autonomy and physical embodiment. Autonomy, echoing Plato's classical ideal of education as turning the soul, denotes the capacity for self-governed judgment and critical reflection. However, Prendergast stresses that this intellectual independence cannot develop in an abstract vacuum; it requires embodiment—the continuous relational dialogue between human sensory faculties, the mind, and the material world. To translate these concepts to the classroom, the authors draw on John Dewey and Paulo Freire. Dewey maintained that education is life itself rather than mere preparation, requiring experiential, tangible interactions that cultivate habits of reflective judgment and cooperative inquiry. Freire directly critiqued the 'banking' model of education, wherein teachers passively deposit static facts into students. In contrast, Freire championed 'problem-posing' education that engages active human subjects in real-world struggles. In the presence of automated systems, Prendergast argues that genuine understanding still depends on this embodied, active participation rather than passive digital consumption. The Optimization Trap and Algorithmic Misalignment (9:31) The authors caution that treating artificial intelligence as the apex standard of intellect encourages students to optimize their own behavior like machines. Citing scholar Maximilian Casey, Prendergast observes that machine learning algorithms operate strictly on a single mathematical objective function. Because algorithmic models recognize failure solely as a failure to optimize that single metric, they entirely lack the conceptual language needed to negotiate messy, conflicting human values—the exact capacity that liberal education seeks to nurture. This drive toward mechanistic efficiency parallels cultural trends like Bryan Johnson’s biohacking movement, which Prendergast warns trains individuals to be more efficient machines rather than better human beings. In higher education, this logic produces 'AI slop'—grammatically polished student essays that simulate academic rigor while remaining devoid of personal intellectual effort. When students optimize exclusively for efficiency and grades, the core loss is not simply academic integrity, but the cognitive opportunity to struggle through complex ideas independently. Deconstructing Flawed Institutional AI Aphorisms (12:51) Prendergast critiques three common slogans circulated by administrators and professional development seminars. The aphorism urging educators to treat AI 'as a tool, not a crutch' fails because crutches are functional tools; the slogan offers no pedagogical guidance on where helpful scaffolding ends and cognitive dependency begins. In philosophy, using AI to outline arguments or refine sentences short-circuits the vital structural discipline and intellectual revision required to clarify thought. A second common mandate—that educators have an urgent duty to teach AI tools to prevent professional obsolescence—unfairly shames faculty while misidentifying the true prerequisites for technological literacy. Effective tool adoption in the workforce stems from strong foundational critical thinking, not institutional software walkthroughs. Finally, attempts to dissuade student use by demonstrating model errors or hallucinations are quickly rendered obsolete as models advance. Prendergast emphasizes that the true danger of generative AI is not factual inaccuracy, but the quiet displacement of autonomous human reasoning. Embodied Polytechnic Inquiry and Cross-Tier Mentorship (19:09) Rather than relying on prescriptive administrative rules, Prendergast points toward active polytechnic pedagogy, as practiced at Utah Tech University. By shifting educational assessment toward hands-on problem-solving that directly engages with the physical and social environment, institutions can emphasize human capacities that generative models cannot simulate or execute on a student's behalf. To institutionalize this approach, the authors propose a Vertically Integrated Project (VIP) framework. This structure establishes collaborative mentorship pipelines connecting faculty, industry partners, and multi-tier student cohorts. By having advanced students guide newer peers through evolving technological applications, universities create structured spaces where educators and learners can continually re-evaluate how pedagogical tools interact with genuine human agency. Prendergast and Fitzgerald conclude that resolving the educational challenges posed by artificial intelligence requires abandoning simplistic binaries between autonomous AI agents and passive human users. Instead, institutions must sustain open dialogue between teachers, learners, and industry, prioritizing embodied, collaborative inquiry that preserves human self-determination in an increasingly automated landscape.

精炼阅读摘要

Utah Tech University 的哲学与人文教授 Emma Prendergast 和 Emily Fitzgerald 指出，在生成式人工智能日益泛滥的背景下，高等教育必须摆脱单一的技术效率优化导向，依托具身认知与批判教育学传统重塑教学，确保学生保持自主性与深层思辨能力。 理论根基：具身认知与问题提出式教育 (2:17) Prendergast 与 Fitzgerald 的研究框架根植于“自主性”（Autonomy）与“具身性”（Embodiment）的辩证关系。自主性承袭自柏拉图将教育视为“灵魂转向之术”的哲学传统，强调学习者自我治理、独立抉择与批判反思的能力；具身性则指明人类智力并非抽象存在，而是深植于身体感官、物理环境与人际交往的物质现实之中。为将这一哲学脉络注入当代教育，研究结合了约翰·杜威（John Dewey）的进步主义经验教育论，主张教育即生活本身，真正的理解只能通过持续的动手交互、反思性判断与合作探究来习得。 此外，研究还引入了巴西教育家保罗·弗雷勒（Paulo Freire）在《被压迫者教育学》中阐发的批判理论。弗雷勒将单向灌输事实的传统教学斥为“储蓄式”（Banking）教育，认为这种将学生视作被动容器的模式具有压迫性；相对应地，他提倡以现实困境为起点的“问题提出式”（Problem-posing）教育，让受教育者作为主动主体介入生活世界。Prendergast 指出，单纯依靠算法推送答案或自动生成总结，正是储蓄式教育在技术时代的变体，教育者必须转向以自主性和具身实践为核心的能动教学。 优化的迷思：单一目标函数与算法平庸 (9:31) 当教育体系将人工智能推崇为智力顶峰标准时，极易诱使学生像算法一样追求自我优化。Prendergast 援引学者 Maximilian Casey 的论断强调，机器学习系统的本质是在单一目标函数下追求最大化输出，其数学底层完全缺乏调和人类多元、冲突目标的机制与语汇，而学会协调此类价值冲突恰恰是人文学科教育的核心使命。这种追求极致效率的倾向在当代文化中已显露无遗，例如 Bryan Johnson 的“生物黑客”（Biohacking）运动试图通过算法调控身体规避死亡，却可能把人规训为更高效的机器而非更健全的人；教宗良十四世在通谕中亦对技术手段盲目扩张而人文心智未同步成熟发出伦理警示。 在大学课堂中，这种对速度与分数的过度追求直接滋生了“AI 垃圾”（AI slop）。Prendergast 分享了一起学生完全利用 AI 生成哲学论文的案例，该论文文字看似严谨流畅，实则毫无见地与深度。她强调，此现象的实质危害并非单纯的作弊违规，而是对产出效率的执迷彻底挤占了学生停下脚步、独立消化复杂概念的思维空间；同时，Fitzgerald 也提醒注意 Peter Thiel 将 AI 视作军事技术的定性，警示教育界认清这些系统背后追求权力与资本扩张的技术本质，而非盲从商业宣传的便利神话。 解构流行格言：高校技术应对的三大误区 (12:51) Prendergast 严厉剖析了当前大学教师培训与行政规范中反复出现的三条流行格言。其一，“将 AI 视作工具而非拐杖”。她反驳称拐杖本身就是工具，其合理性完全取决于使用者是否切实需要；在哲学写作中，若依赖算法来突破写作障碍、生成论证大纲甚至润色语句，实质上会彻底剥夺学生构建认知结构与反复推敲字句的核心思维训练，导致格言无法划清辅助工具与认知依赖的界限。 其二，“必须教学生熟练使用 AI，否则就是不负责任”。这种论调既给未受过专业技术伦理培训的一线教师施加了不当的道德压力，也混淆了操作技能与批判性素养的区别；专业人士之所以能快速驾驭新工具，依靠的是扎实的底层思辨基础，而非学校对具体软件步骤的死板培训。其三，“通过展示幻觉与事实错误劝阻学生使用”。随着大语言模型以前所未有的速度修复缺陷并连接实时网络，这一策略正迅速失效；Prendergast 指出，真正的教学危机从来不是模型输出了错误事实，而是学生机械复制答案直接取代了自身的独立推理。 实践重构：多工科主动学习与垂直整合项目 (19:09) 针对自上而下灌输禁令或操作指令的局限，Prendergast 与 Fitzgerald 倡导通过包容性对话连接教师、学生与行业伙伴，寻求切实的落地路径。依托 Utah Tech University 的“多工科”（Polytechnic）教育特色，她们提出推行“垂直整合项目”（Vertically Integrated Projects, VIP）框架。该模式构建了一个跨层次的长期指导网络，将高校教师、行业导师、高年级本科生与低年级新生紧密联合，让学生在传帮带与校企协作中探索技术的演进边界。 该方案主张将教学重心转向需要亲身介入物理世界的主动体验式学习，通过解决真实的工程与社会问题，培养学生无法被生成式 AI 轻易复制的高阶批判思维与人际共情。Prendergast 总结道，高等教育必须彻底摒弃“自主 AI 智能体对阵被动人类用户”的二元对立设想，确立以“自主人类行动者运用协作性 AI 工具”为核心的全新教学范式。 Prendergast 与 Fitzgerald 明确提出，应对 AI 冲击的关键在于从被动适应转向深植于物质世界的主动探究。然而，面对商业技术势力的快速推进与学术体制对量化效率的偏好，如何在资源有限的常规人文教学中广泛落地多层级项目，并帮助缺乏工科背景的教师化解技术焦虑，仍是有待长期实践检验的系统性难题。

Detailed notes — zh

## Background 该视频由 YouTube 频道 Ai4 于 2026 年 8 月 18 日发布，主讲人为犹他理工大学哲学助理教授兼人文中心主任 Emma Prendergast。演讲基于与 Emily Fitzgerald 的合作研究，探讨在 AI 全面重塑教育生态的背景下，如何守护人类自古希腊以来所追求的自主性与身心具身性，并反思生物黑客与数据优化对人性的挤压，呼吁通过体验式学习重塑人机协作价值。 ## TL;DR 演讲从柏拉图、约翰·杜威的“做中学”以及弗莱雷的提问式教育理念出发，深入剖析了单目标算法优化、极端生物黑客思潮与“AI 废料”泛滥对学生自主思考能力的系统性剥离。Prendergast 逐一解构了教育界关于 AI 教学的常见空洞格言——包括“当作工具而非拐杖”、“不教即不负责”以及“通过指出幻觉缺陷劝退禁用”，揭示出单纯技术主义防守与口号化应对的局限性。最后，她主张教育应超越提示词操作，转向建立跨越学术与产业的垂直整合项目（VIP），并依托现实物理世界的具身体验式教学来牢固捍卫人类的主体能动性。 ## Atmosphere & style 演讲节奏明快、生动风趣却又充满严肃的哲学术语与深层思辨。主讲人通过分享早期 GPT 吐出荒诞诅咒等轶事引得全场哄堂大笑，随后迅速切入地缘军事技术现实与教皇通谕等宏大批判视角，形成鲜明的情感张力。整场分享由浅入深，既有对当下教育一线浮躁现象的犀利剖析，又充满对人性和具身体验的真诚关怀，带给听众一种清醒、务实且兼具启发性的学术思辨体验。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=4s) AI 工具的泛滥正在重构现代教育生态，教育者面临变动的标准，核心挑战在于如何在技术包围中维系学生的具身自主性（Embodied Agency）。 - [02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) 自主性自古希腊起就是教育的核心目标，而具身性强调人类思考与肉身经验、物理环境及群体关系的不可分割性。 - [02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) 杜威的‘做中学’经验教育与弗莱雷的‘提问式教育’为当下抵抗算法机械化、培养解放性主体提供了重要理论基石。 - [06:18](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=378s) 在效率叙事背后，AI 的底层驱动与早期研发本质上带有军事技术与资本驱动的烙印，技术本身并非纯粹中立。 - [09:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=571s) 机器学习的单目标优化逻辑无法容纳人类社会的多元冲突目标，盲目追求效率将人降维为机器，抹杀人性的丰富性与伦理反思。 - [11:37](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=697s) 学生使用 AI 生成无实质内容的‘AI 废料’，反映出功利化与效率至上的价值观正在挤压深度批判性思考的空间。 - [12:51](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=771s) ‘把 AI 当工具而非拐杖’的流行口号存在概念循环与定义模糊的问题，未能真正解决写作与思维训练中的教学边界划分难题。 - [15:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=931s) ‘不教 AI 就是不负责任’的观点混淆了软件操作技巧与根本的批判性思维，而深厚的人文思辨才是驾驭新技术的真正底色。 - [15:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=931s) 企图通过揭示 AI 的事实性错误或幻觉来劝退学生是短视且正在过时的策略，无法解决学生外包推理过程的核心危机。 - [18:25](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1105s) 重塑 AI 素养需超越提示词工程等操作层面，将人从被动的‘人类用户’重新确立为与 AI 工具协同共进的‘人类主体代理人’。 - [18:25](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1105s) 通过跨学科、跨年级并联动产业的垂直整合项目（VIP），构建开放包容的生态结构以长期应对技术伦理与教育演进。 - [18:25](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1105s) 推行主动体验式学习，引导学生投身真实物理世界的动手实践，在 AI 无法替代的具身经验中筑牢批判性反思与共情壁垒。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=4s) AI 无处不在：重塑教育与被遗忘的核心价值 演讲者 Emma Prendergast 是犹他理工大学（Utah Tech University，位于圣乔治，距拉斯维加斯仅两小时车程）的哲学助理教授兼人文中心主任。她开场时幽默地向台下共同执笔该研究的同事 Emily Fitzgerald 打招呼，并笑称如果大家有特别刁钻难答的问题，请务必直接提问 Emily。 Prendergast 指出，AI 目前已经无所不在，彻底重构了现代教育的生态。教育工作者每天都在面对不断变动的考核标准、模糊不清的管理领导以及彻底改变的学习环境。教育者和学生无时无刻不在被各种 AI 工具包围：从利用 ChatGPT 生成课堂讨论提示词、体验 Anthropic 的 Claude 新功能、一键生成全班学生讨论帖的 AI 摘要，甚至连日常冰箱都被强行植入了 AI 功能，更不用提社交平台上的 Grok。对于千禧一代及更年长的群体而言，这种现象神似当年微软办公套件里无处不在的“回形针助手”（Clippy）——它偶尔确实有用，但频繁出现在人们根本不需要它的荒唐场景中。 这种“AI 泛滥”正是当今年轻学子与未来劳动力所面临的现实生活与求学环境。在这一背景下，核心研究问题随之浮现：**当 AI 无处不在时，我们如何在教育中维系身心具身性（Embodied Agency）？** 教育工作者究竟该如何守住那些自古传承的核心教育目标——批判性思维、自主性（Autonomy）的养成以及具身的人类能动性？ ```mermaid flowchart TD accTitle: AI时代教育面临的环境与核心研究问题 subgraph S1["现实环境：无所不在的 AI 工具"] A1["LLM 工具<br>(ChatGPT / Claude / Grok)"] A2["日常与智能硬件<br>(AI 冰箱 / 讨论摘要插件)"] A3["教育者困境<br>(变动的标准 / 模糊的指引)"] end subgraph S2["危机：对传统教育价值的冲击"] B1["过度依赖工具导致思考被剥离"] B2["技术侵入不必要的学习场景<br>(现代版 Clippy)"] end subgraph S3["核心研究目标"] C1["培养学生的自主性<br>(Autonomy)"] C2["践行具身的人类能动性<br>(Embodied Agency)"] end S1 --> S2 S2 --> S3 ``` --- ### [02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) 自主性与具身性：古希腊哲学与现代教育学理论 Prendergast 梳理了演讲脉络，随后深入拆解了本次研究的理论基石：“自主性”（Autonomy）与“具身性”（Embodiment）。 - **自主性（Autonomy）**：指的是一个人能够成为“自己生命自我治理的作者”（the self-governed author of one's own life），拥有独立决断与批判性内省的能力。这一直是自古希腊雅典以来的终极教育宗旨。正如柏拉图（Plato）曾将教育描绘为“灵魂转向的艺术”（the art of turning the soul），教育的本意是赋予人独立思考的权利，使生命更为自由。 - **具身性（Embodiment）**：追求自主性绝不能脱离人作为物理存在的肉身现实。具身性是指人类的身体物理经验、精神世界、周围环境以及个体与群体层面上所有构成“人性”之要素间的关联性（Relationality）。 为了阐述具身自主性在教学法中的演进，Prendergast 引入了两位关键的现当代教育理论家： 1. **约翰·杜威（John Dewey）与进步教育**： 杜威的名言深入人心：“教育不是生活的准备，教育就是生活本身”（Education is not preparation for life, but is life itself）。杜威开创了“进步学校”（Progressive School）理念，提倡以身临其境的“做中学”（Hands-on Experiential Learning）取代枯燥死板的死记硬背。杜威深刻洞察到，真实的学习恰恰发生在具体可感的物理与具身经验之中。在杜威看来，教育旨在提升人的自我管控能力，以持续的经验滋养生命生长，最终使人有能力掌舵自己的人生航向，养成探究、合作与反思性判断的习惯，而非培养盲从与死记硬背。 2. **保罗·弗莱雷（Paulo Freire）与“提问式教育”**： 弗莱雷在《被压迫者教育学》（Pedagogy of the Oppressed）中界定了两种对立的教学范式： - **银行储蓄式教学（Banking Education）**：将学生视为空白容器，教师单向地把预设好的知识碎片像“存钱进银行”一样塞进学生的脑海。 - **提问式教育（Problem-Posing Education）**：源自二战后弗莱雷在巴西东北部的扫盲运动。当时不识字便无法享有投票权，弗莱雷教导贫困农民时，选用与他们生活息息相关的鲜活语汇（如“贫民窟”、“工资”、“土地”）。提问式教育具备解放性与自主性，视学习者为置身物理世界的鲜活主体，教师向学生抛出问题，引导他们以主体身份协同探索解法，在具身参与中实现真正的思想解放。 --- ### [06:18](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=378s) 锚定偏见：从恶搞歌词的 AI 威胁到军事技术现实 在进入具体的“问题提出”（Problem Posing）阶段前，Prendergast 坦承自己与合作者 Emily 在面对 AI 时抱有某种“锚定偏见”（Anchoring Biases）。和很多教育工作者一样，即便如今她们在课堂教学中多有运用，但两人对 AI 的接纳都始于强烈的怀疑论。 Prendergast 讲述了自己在 2021 年 12 月第一次使用早期 GPT 模型的啼笑皆非的经历。当时她的丈夫热衷于摆弄早期生成式 AI，而她作为哲学家对此嗤之以鼻，甚至反驳：“我不在乎这个，我是个哲学家，我不需要人工智能，我有正常的人类智能（I have regular intelligence）。”但丈夫力劝她一试，声称其效果绝对令人吃惊。当时他们刚搬到洛杉矶，那是她第一次在温暖的加州过圣诞节，她随手写了一首俗气的圣诞打油歌词，大意是“把彩灯挂在棕榈树上，我想要的圣诞礼物唯有一场大雪”。丈夫提议让 AI 顺着这首俗气的歌续写几句歌词。 不料，AI 输出的内容却让 Prendergast 惊掉下巴。在展示幻灯片前，她给观众打了个“预警”（Trigger warning），因为 AI 的回复异常荒诞且脱轨（unhinged）。屏幕亮出后引得全场大笑：那台早期模型不仅没有写出温情的歌词，反而吐出一句恶毒的诅咒，称“当你死掉的时候，它会非常高兴”（it would be glad when I'm dead）。尽管后来反复尝试均无法复现这一离谱回答，但这彻底奠定了她的初始心理阴影：“这黑盒里到底装了什么东西？它居然说等我死了它会很高兴！” 而她的合著者 Emily 的锚定偏见则更加冷峻严肃：科技投资人彼得·蒂尔（Peter Thiel）曾公开断言，**“AI 是一项军事技术”（AI is a military technology）**。社会各界不断灌输给大众和学生的叙事是“AI 的诞生是为了让我们生活更美好、更高效”，但在其乐融融的效率叙事背后，不可回避的事实是这项技术的核心驱动力和大量早期研发本质上是作为武器系统、为了地缘政治与资本利润而制造出来的。 --- ### [09:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=571s) 优化的陷阱：单目标算法、肉身黑客与超人类主义的反思 Prendergast 警告，当我们盲目将 AI 奉为“智力的终极标准”时，最大的潜在危机在于我们正在潜移默化地规训年轻一代，让他们误以为人类也必须像 AI 一样去自我“优化”（Optimize）。这种“跟上 AI 效率”的庞大焦虑正全方位压垮当下的大学生。 学者麦克西米利安·凯西（Maximilian Casey）的论述精准剖析了这一机制的死穴： > "在机器学习中，算法必须最大化一个单一目标；而在单目标框架下，可能发生的最坏结果无非是‘未能优化该目标’。在这种框架中，我们根本没有语言去讨论和容纳不同人群之间相互冲突的多重目标。" 然而，理解并协调人类社会中复杂、交织且彼此冲突的目标，恰恰是教育最渴望培养的核心素养。如果一切都被降维至“单一目标的最优解”，人性的广度将被彻底抹杀。 Prendergast 进一步将此现象与硅谷科技圈风靡的极端实践联系起来： - **布莱恩·约翰逊（Bryan Johnson）的“Don't Die”（别死）运动**：科技创业者试图通过算法、精密监控和生物黑客技术（Biohacking）追求长生不老，甚至宣称要让死亡过时。但 Prendergast 犀利反问：这种狂热的生物黑客技术充其量只是在“把人类改造成更精良的机器，而不是更完善的人”（become better machines, not necessarily better people）。人类到底在优化什么？难道生命的价值仅仅是一组代谢效率的指标吗？ - **精神与宗教维度的批判**：她引用了教皇对当今技术主义的警告——若技术狂飙突进却没有伦理与社会的协同演进，最终结果只会是工具手段的空前扩张，伴随着人性的急剧萎缩。教皇通谕对这种后人类主义（Post-humanism）和超人类主义（Transhumanism）思潮提出了深刻反思，呼吁社会重构人文价值与科技的对话机制。 在剖析完宏观层面的技术崇拜后，Prendergast 将目光拉回教学一线最刺眼的现实：AI 垃圾内容（AI Slop）泛滥。大量学生将 AI 生成的内容不假思索地拼贴成作业、抄袭作弊，并冠冕堂皇地包装成个人成果提交，直接破坏了学术诚信与深层认知的发生环境。 ### [11:37](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=697s) “AI 废料”与效率至上：学生自主思考的丧失 Emma Prendergast 分享了一个颇具讽刺意味的真实案例：一位学生在写关于“技术如何介入与调解人类经验”的课程论文时，向她坦白自己提交的内容实际上完全是由 AI 生成的“废料”（AI slop）。从表面上看，整篇文本在语法上无懈可击，听起来极具学术深度且显得十分聪明，但只要真正开始阅读，就会发现它几乎毫无实质内容，读者极难抓住核心论点，更无法在阅读过程中保持专注。然而，Prendergast 指出，这起事件中最严峻的问题并非单纯的“作弊”，而在于这名学生完全陷入了追求效率、优化流程和争取高分的功利逻辑。当追求效率和绩点的价值观占据主导地位时，便无情挤压了他停下来独立审视“技术与人性的关系”这一深刻哲理命题的思考空间。 这种外包思考的现象早已由长期的“谷歌效应”（Google effect）提供了前车之鉴，而如今越来越多的科学家更是在敲响警钟，担忧过度依赖 AI 可能会实质性削弱人类的大脑认知功能。随着 AI 无处不在，人们对自己作为具身化存在（embodied beings）的感知开始变得模糊混乱，当代学习者在培养批判性思维方面面临着巨大的结构性阻碍。面对这种复杂局面，教育界目前充斥着大量追求即时成效的短期思维与看似万灵的空洞口号。Prendergast 将这些在教师研修工作坊、政策提案以及教职员工茶余饭后的走廊闲聊中不断重复的建议总结为三大“格言”（aphorisms），并指出这些泛泛而谈的陈词滥调在实际教学指导中早已彻底失效。 --- ### [12:51](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=771s) 格言一：“把 AI 当作工具而非拐杖”的模糊与局限 教育界流传甚广的第一条格言是：“我们应当教会学生将 AI 作为‘工具’使用，而不是当作‘拐杖’。”Prendergast 犀利地指出了这句话在逻辑与语言上的根本缺陷。首先，从概念框架来看，拐杖在物理意义上本身就是一种工具；一个人是否需要拐杖，完全取决于其真实的身体需求——在腿脚健全时使用拐杖是不必要的，但在腿部受伤需要支撑时使用拐杖则是完全合情合理的。其次，这一表述存在极大的模糊性，它翻译过来无非是一句毫无信息量的废话：“在有帮助时使用 AI，在没有帮助时不要使用 AI”，然而教育工作者真正面临的核心难题恰恰是：这条合理与不合理的界线究竟应当划在哪里？ 在某些明确的领域，例如医疗诊断与药物剂量的精确计算中，AI 的应用显著提升了人类的生存体验，其带来的益处显而易见。但在课堂写作教学中，AI 的边界却极其微妙且充满争议： * **头脑风暴与构思**：许多写作教师将 AI 誉为打破“写作者瓶颈”（writer's block）的绝佳利器，认为它可以为学生提供初始灵感。但作为哲学教授，Prendergast 对此深表疑虑。哲学的核心本质正是通过艰苦深入的思辨来提炼自己的独创观点，并对前人的思想作出独立反应；如果连最初的想法都让渡给 AI，便背离了哲学训练的本意。 * **逻辑框架与大纲组织**：有人主张利用 AI 来自动生成论文提纲。但优秀的写作本质上是清晰思维的外在体现，唯有亲历组织思路、梳理逻辑的挣扎过程，学生才能真正成长为独立自主的思考者。 * **文本编辑与润色**：尽管 Prendergast 相对能够接受将 AI 用于快速的基础语法检查，但许多人文学科同行依旧极度看重反复推敲、重塑句式结构的手工打磨过程，因为每一个精准句式的构筑，都是锤炼认知清晰度不可或缺的具身化心智体操。 --- ### [15:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=931s) 格言二与三：“不教就是不负责任”与“通过指出 AI 缺陷来劝退” 教育界盛行的第二条格言带有明显的道德绑架色彩：“我们必须教会学生用好 AI，不这么做就是对学生的不负责任；否则他们未来投简历、找工作都会被 AI 筛选机制淘汰。”这种说辞常常被用来羞辱那些对在课堂引入 AI 持谨慎保留态度的教师。对此，Prendergast 提出了层层反思： 1. **概念界定的缺失**：究竟何为“用好”（use it well）？它指的是针对特定行业提高产出的操作技巧，还是涉及深刻价值考量的伦理审视？目前的口号从未给出确切界定。 2. **师资培训的脱节**：该论调默认广大 K-12 和高等教育阶段的教师已经天然掌握了伦理化 AI 素养的教学技能。但事实恰恰相反，绝大多数大学教授甚至从未接受过系统性的通用教学法培训，更遑论在 AI 时代探索人机共教的新模式。 3. **批判性思维才是根本**：现实中有大量从未在学校接受过任何 AI 培训的职场人士，如今都能在日常工作中驾轻就熟地驾驭 AI。这充分证明，决定能否用好新技术的关键并不在于机械的操作教学，而在于深厚的批判性思维积淀与跨学科专业底蕴；因此，过去的老师没有未卜先知地教导今天的 AI 工具，绝不意味着他们存在任何失职。 第三条格言则来自部分传统防守型教师：“我们应当通过向学生展示 AI 的技术缺陷来全面禁止其在课堂中的使用。”这类教师试图抓住 AI 的“幻觉”（hallucinations）和事实性错误来劝退学生。Prendergast 指出，随着大语言模型以前所未有的速度迭代演进，早期无法联网检索或提供真实参考引用的硬伤早已被新一代模型攻克，这种基于工具漏洞的防守策略正在以肉眼可见的速度迅速过时。更重要的是，问题的核心从不在于 AI 是否会出错，而在于如果学生只习惯于对模型输出进行简单的“复制与粘贴”，他们就永远无法自主培育独立的理性推理能力。 ```mermaid flowchart TD accTitle: 批判教育界关于 AI 的三大虚妄格言 subgraph Slogans["教育界的表面化格言与核心漏洞"] A["格言一：当工具而非拐杖<br/>（Tool not a Crutch）"] --> A1["缺陷：概念循环且定义模糊<br/>未解决教学边界划分难题"] B["格言二：不教即是不负责<br/>（Irresponsible Not to Teach）"] --> B1["缺陷：混淆操作与思维素养<br/>忽视教师培训缺失的客观现实"] C["格言三：揭示幻觉以劝退禁用<br/>（Demonstrate AI Failures）"] --> C1["缺陷：技术缺陷正被迅速攻克<br/>掩盖了学生自主推理能力的丧失"] end ``` --- ### [18:25](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1105s) 走向具身化与行动：未来 AI 素养教育的三大出路 在逐一解构了流行口号的局限后，Prendergast 为 AI 时代的教育重构指明了三大切实可行的方向： 首先是**深化与重塑“AI 素养”（AI Literacy）的内涵**。真正的 AI 素养绝不能仅停留在编写高效提示词（prompt engineering）、防范学术抄袭或识别虚假信息的实用技术层面，而必须引导学生反思技术如何深刻重塑我们身处的具身化、物理化与具象化的现实世界。在主体关系的定位上，教育必须坚决摒弃“一边是 AI 智能体，另一边是人类用户”的被动二元对立范式，转向将人牢固确立为“与 AI 工具协同共进的人类主体代理人”（human agents alongside AI tools）。 ```mermaid flowchart LR accTitle: AI 时代人机关系范式的转变 subgraph OldParadigm["陈旧的被动范式"] O1["AI 代理 (AI Agent)"] <--> O2["人类用户 (Human User)"] end subgraph NewParadigm["提倡的主体范式"] N1["人类主体代理人 (Human Agents)"] --- N2["AI 辅助工具 (AI Tools)"] end ``` 其次是**构建跨界包容的对话机制（Vertically Integrated Projects, VIP）**。面对技术引发的伦理与教学难题，没有任何个体能够独自给出标准答案，简单的标语式宣导根本无济于事。Prendergast 透露，她正与合作者推进一项跨部门垂直整合项目方案。该模式将大学学术资源、行业产业伙伴、跨专业导师指导以及不同年级的学生紧密联结在一个开放包容的生态结构中，由高年级学生对低年级学生展开朋辈传帮带，共同探索适应技术演进的长期教育方案。 最后是**全面践行具身化的“做中学”（Active & Experiential Learning）**。立足于犹他科技大学（Utah Tech University）作为应用技术型大学（Polytechnic Institution）的实践土壤，Prendergast 与同事 Emily 极力倡导主动学习与体验式教学。教育的终极壁垒在于那些 AI 既无法虚拟生成、更无法代为亲历的领域——引导学生走出屏幕，卷起袖子投身于真实的物理世界，在直面复杂现实问题的动手实践中激发真正的批判性反思与共情力量，捍卫人类作为具身存在的核心主体性。 * **新型 AI 素养内涵**：超越提示工程与反剽窃，聚焦于对具身物理世界技术影响的批判性反思。 * **垂直整合项目（VIP）结构**：高校学术界 + 行业一线 + 跨层级师生导师制，构建动态演进的对话研讨场域。 * **理工应用型教育模式（Polytechnic Model）**：强化现场感与触觉感，通过解决现实复杂困境，守护 AI 无法替代的具身体验与人类尊严。 ## Notable quotes > [02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) the self-governed author of one's own life — Emma Prendergast > [02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) the art of turning the soul — 柏拉图 (Plato) > [02:17](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=137s) Education is not preparation for life, but is life itself — 约翰·杜威 (John Dewey) > [06:18](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=378s) I have regular intelligence — Emma Prendergast > [06:18](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=378s) it would be glad when I'm dead — AI 模型 > [06:18](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=378s) AI is a military technology — 彼得·蒂尔 (Peter Thiel) > [09:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=571s) 在机器学习中，算法必须最大化一个单一目标；而在单目标框架下，可能发生的最坏结果无非是‘未能优化该目标’。在这种框架中，我们根本没有语言去讨论和容纳不同人群之间相互冲突的多重目标。 — 麦克西米利安·凯西 (Maximilian Casey) > [09:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=571s) become better machines, not necessarily better people — Emma Prendergast > [12:51](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=771s) 我们应当教会学生将 AI 作为‘工具’使用，而不是当作‘拐杖’。 — Emma Prendergast > [15:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=931s) 我们必须教会学生用好 AI，不这么做就是对学生的不负责任；否则他们未来投简历、找工作都会被 AI 筛选机制淘汰。 — Emma Prendergast > [15:31](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=931s) 我们应当通过向学生展示 AI 的技术缺陷来全面禁止其在课堂中的使用。 — Emma Prendergast > [18:25](https://www.youtube.com/watch?v=iVjtWO7CxZw&t=1105s) human agents alongside AI tools — Emma Prendergast ## Mentioned Emma Prendergast, Emily Fitzgerald, 犹他理工大学 (Utah Tech University), ChatGPT, Claude, Anthropic, Grok, Clippy (微软回形针助手), 柏拉图 (Plato), 约翰·杜威 (John Dewey), 保罗·弗莱雷 (Paulo Freire), 《被压迫者教育学》(Pedagogy of the Oppressed), 彼得·蒂尔 (Peter Thiel), 麦克西米利安·凯西 (Maximilian Casey), 布莱恩·约翰逊 (Bryan Johnson), Don't Die, Vertically Integrated Projects (VIP) ## Video description Emma Prendergast, Assistant Professor of Philosophy and Director of the Humanities Center, Utah Tech University Autonomy, or the ability to be the self-governed author of one’s own life, has been held up as an educational aim since Ancient Athens. This talk examines how we can preserve human agency as fundamental aim and why embodiment must become an equally central educational concept as AI reshapes teaching and learning. The talk will challenge familiar slogans about responsible AI use and explain how technologies such as biometrics and biohacking can reduce human lives to data, optimization, and efficiency. Attendees will learn why ethical agency depends on recognizing our embodied needs, relationships, limitations, and capacity for empathy; and why educators, learners, and technology leaders must address these questions together. The talk will highlight experiential learning as a method for helping students learn and critically engage with AI without surrendering their humanity. [Source](https://www.youtube.com/watch?v=iVjtWO7CxZw)

Source quotes

[2:17] Emma Prendergast: the self-governed author of one's own life [2:17] 柏拉图 (Plato): the art of turning the soul [2:17] 约翰·杜威 (John Dewey): Education is not preparation for life, but is life itself [6:18] Emma Prendergast: I have regular intelligence [6:18] AI 模型: it would be glad when I'm dead [6:18] 彼得·蒂尔 (Peter Thiel): AI is a military technology [9:31] 麦克西米利安·凯西 (Maximilian Casey): 在机器学习中，算法必须最大化一个单一目标；而在单目标框架下，可能发生的最坏结果无非是‘未能优化该目标’。在这种框架中，我们根本没有语言去讨论和容纳不同人群之间相互冲突的多重目标。 [9:31] Emma Prendergast: become better machines, not necessarily better people [12:51] Emma Prendergast: 我们应当教会学生将 AI 作为‘工具’使用，而不是当作‘拐杖’。 [15:31] Emma Prendergast: 我们必须教会学生用好 AI，不这么做就是对学生的不负责任；否则他们未来投简历、找工作都会被 AI 筛选机制淘汰。 [15:31] Emma Prendergast: 我们应当通过向学生展示 AI 的技术缺陷来全面禁止其在课堂中的使用。 [18:25] Emma Prendergast: human agents alongside AI tools

Source quotes

[9:31] Maximilian Casey: In machine learning, there's a single objective that the algorithm must maximize... In such a framework, we do not have the language to talk about the conflicting objectives of different people. [2:17] Plato: the art of turning the soul [12:51] Emma Prendergast: Teach students how to use AI as a tool and not a crutch. [12:51] Emma Prendergast: We must teach students how to use it well, and it is irresponsible not to. [17:32] Emma Prendergast: Discourage AI use by demonstrating its failures. [19:09] Emma Prendergast: cannot generate or act out for us. [2:17] Paulo Freire: Pedagogy of the Oppressed [9:31] Bryan Johnson: Don't Die

Original transcript

[0:04] Hi, I'm Emma Prendergast.  I'll be presenting a work [0:08] that I've co-authored with my colleague Emily  Fitzgerald. Emily, will you give a wave? [0:12] If you have very hard questions to ask about  our research, please direct them to her. [0:17] So, we are philosophy and humanities professors at  Utah Tech University, which is in St. George, just [0:26] east on the highway a couple of  hours from Las Vegas. You'll find us. [0:32] So, AI it's everywhere. It's fundamentally  changed education, and we're interested in [0:38] thinking about what education looks like now now  that can preserve long-standing educational aims, [0:45] critical thinking, the cultivation of  autonomy and embodied human agency. [0:51] Professionals in education are  facing like shifting goalposts, [0:55] unclear leadership, and a totally changed  learning environment. So, our research question [1:04] Wait. This one? Oh, okay. Our research question,  how do we educate for embodied agency when AI is [1:11] everywhere? Oh. That's embarrassing. ChatGPT Oh.  I can use AI to suggest a discussion prompt now [1:22] for my for my students. Oh, and there's new ways  to work with Claude. Oh, I can open a summary [1:28] now of the discussion posts that my students  write. Oh, it's in the refrigerator, also. [1:35] Now we have Grok. If you're like a millennial  and up, this might feel like a little familiar. [1:40] It's kind of like Clippy showing up again.  He can be helpful, but sometimes appears in [1:45] places that it's not clear if we needed him to be  there, and this is the life environment that we [1:50] face these days in the learning environment  that students today, the future workforce, [1:55] uh faces. Um AI is everywhere. It's helpful,  sometimes it's not. I don't know if anyone [2:00] has the AI refrigerator. If it helps you, please  let me know. I'm curious. Um but that this is the [2:05] learning environment that we face. And so, we're  interested in thinking about how we can preserve [2:10] uh long-standing values in education um when this  this is the situation. Okay, so a quick outline. [2:17] Um I'll I'll discuss autonomy and embodiment as  core educational concepts. Um I'll do a little [2:24] bit of what's called problem posing, and I'll  explain what that means. Um then we'll discuss [2:30] uh what we're referring to as the AI aphorisms.  And these are um kind of like sloganized bits [2:36] of advice that um educators, professionals in K  through 12, professionals in higher education, [2:43] um that we've been kind of hearing over and  over again, um but that my co-author and I find [2:47] a fall a bit flat. Um and then finally, we'll  talk about sort of where do we go from here. [2:56] Okay, so uh in our work, we describe autonomy and  embodiment as important educational values. So, [3:02] autonomy, um which is like the ability to be like  the self-governed author of your own life, to make [3:08] your own decisions, to critically introspect. Uh  this has been a fundamental aim in education uh [3:14] since ancient Athens when Plato wrote of education  as the art of turning the soul. Education serves [3:21] the purpose of uh helping us think for ourselves  and making us freer. We argue that in order to aim [3:28] at autonomy, we need to keep in mind that we are  embodied creatures in a physical world. Embodiment [3:33] we define as the relationality between our  physical experiences, our mind, the world around [3:38] us, and whatever else makes us human on individual  and communal scales. So, I'll say a little [3:42] bit more in the next uh few slides about how  educating for autonomy and embodiment are related. [3:52] Uh Uh educational theorist John Dewey,  he's known for the idea that education [3:57] is not preparation for life, but is life itself.  He piloted the idea of the progressive school, [4:04] which focused on experiential hands-on learning,  rather than just rote memorization, which is what [4:09] he had seen sort of happening in the schools. Um  Dewey had the insight that our tangible sort of [4:13] embodied experiences are where the real learning  happens. Uh and for Dewey, education aims at our [4:19] capacity to govern ourselves, to exert self-  control. Education is something that sustains [4:25] our growth, so that we can become capable of  directing our lives. And we need experiences [4:30] uh for that continued growth. So, schools then  ideally cultivate habits of inquiry, cooperation, [4:36] reflective judgment, and not just uh you know,  memorization, rote obedience, uh and so on. [4:45] Similarly, educational theorist uh  Paulo Freire describes a distinction [4:49] between teaching as banking or as problem posing. [4:53] He theorized that banking is a teaching method  where students are seen as like containers and [4:58] the teacher just deposits the knowledge uh into  their minds, like putting money into the bank. [5:05] Uh Freire uh instead, he saw education for its  liberatory potential, and this is from a book, [5:12] Pedagogy of the Oppressed. Um having worked in  post-World War II Northeast Brazil at a time when [5:18] uh education, when literacy was a requirement  to be able to vote, uh he worked with illit- [5:24] illiterate peasants, focusing on vocabulary that  mattered to them, like slum, wage, land. Freire [5:30] describes a method of teaching that's liberatory  and autonomy-cultivating, uh known as problem [5:35] posing. It sees learners as uh beings in the world  who can participate in their own liberation. So, [5:41] as a teacher, you pose problems to the student,  and they actually engage in the solving of that [5:45] problem, rather than just like, here's what I  know and trying to stick it into their minds. Um [5:51] So with these this sort of theoretical background,  we have a framework that understands the most [5:56] important type of learning as embodied  and connected to the world with the aim of [6:00] cultivating autonomy. So now I'm going to go get  into a little bit of problem posing and this is [6:05] where AI comes into the picture some of the sort  of problems or barriers or complications that AI [6:11] poses for all of us um when trying to realize  these values in teaching and learning. [6:18] Okay, first this is part of the problem  posing, but we did want to sort of acknowledge [6:24] my co-author and I that we have some anchoring  biases when it comes to AI. So similar to a lot of [6:29] professionals in education, we've kind of started  from a place of skepticism about AI and how it [6:36] could feature into education. Both my co-author  and I use AI in the classroom in various ways, [6:42] but we both kind of started from a point of  skepticism. So I'm going to tell you about my [6:48] anchoring bias. This goes back to the first time  that I ever prompted an AI. It was in December [6:52] of 2021. So this was like a really early GPT  model. Um my husband had been sort of playing [6:59] around with them and I was kind of like I don't  care about that. I don't want to know about it. [7:03] I'm a philosopher. I don't need artificial  intelligence. I have regular intelligence. [7:07] But he was sort of like you know priming me  to like just try it out. You're going to be [7:11] kind of surprised at how good the results were and  even then the results were pretty surprising for [7:16] for you know uh pre-2021 what we thought AI  could be capable of. So I decided to try it [7:22] out. At that time I was living in Los Angeles  during Christmas time for the first time and I [7:26] had written like a corny Christmas song with like  lyrics about like hanging the lights on the palm [7:32] trees and all I want for Christmas is snow. And  it turns out that Kermit the Frog already did [7:37] that. So it was not especially original. It  was very corny. So so husband was like this [7:42] is the perfect thing. I bet it'll give you some  corny Christmas lyrics that totally fit the song. [7:47] Um, okay. So, here's what it said. This is  the first thing an AI ever ever gave me. [7:53] I have to give you a trigger warning  cuz it's like a little unhinged. [7:59] You see it? [8:03] Yeah. It's okay to laugh. [laughter] So,  the first thing AI ever said to me is that [8:09] it would be glad when I'm dead. Um, so you  can't really blame me for sort of starting [8:14] from a place of skepticism when it comes to  AI. And this result, by the way, I like tried [8:19] prompting it again. I couldn't replicate  this. I have no idea where this came from. Um, [8:24] but basically I've started with an  understanding of AI is such that I'm like, [8:28] I don't know what's in this thing. Um, it it might  be glad when I'm dead. Okay, so next my co-author [8:34] has shared the following as her anchoring  bias, uh, early exposure to the idea of AI. [8:40] Uh, and we have uh, Peter Thiel declaring  that AI is a military technology. [8:49] Um, we're told, our students are told that AI is  something that's here to make our lives better and [8:55] in various ways it has and has so much potential  to do so. Um, that might be through efficiency, [9:00] optimization, although as I'll explain in a later  slide, it's not clear that that's always what [9:05] we should want either. Um, but meanwhile the  technology is, uh, developed as a weapon and [9:11] that's not necessarily the thing that people have  in the front of their minds when they're thinking [9:15] about AI, but it is something that my co-author  and I think about, that it is a weapon, uh, [9:20] developed as a weapon. It's not always used as a  weapon, but it's developed as a weapon for profit. [9:28] Okay. [9:31] Uh, even if we implicitly sort of hail AI as sort  of a standard for intelligence, uh, my co-author [9:37] and I worry that we risk teaching young people  that they need to optimize in the way that AI [9:43] optimizes. Um and if we teach young people that,  we we do run the risk of losing a lot of what's [9:50] fundamental to uh humanity. So, uh the pressure to  keep up or be as efficient as an AI is enormous. [9:57] It's a pressure that we see um our college  students facing. As Maximilian Casey writes, [10:02] in machine learning, there's a single objective  that the algorithm must maximize in a framework [10:07] with a single objective. The worst that can happen  is that we fail to optimize the objective. In such [10:12] a framework, we do not have the language to talk  about the conflicting objectives of different [10:17] people. The latter being exactly the type of  thing we hope for young learners to be capable of. [10:23] Uh another example we think about, so we  have movements today such as don't die from [10:28] entrepreneur Bryan Johnson focusing on how to use  AI to extend human life or make death obsolete. [10:37] Biohacking we worry enables us to become better  machines, not necessarily better people. Um [10:43] and we have to ask exactly what it is that we're  seeking to optimize here. Um so, this stands in [10:47] contrast with wisdom we can find from other public  figures um like Pope Leo the 14th who has recently [10:53] emphasized that if technology advances without  corresponding ethical and social progress, [10:58] we might see an increase in means um without  uh a corresponding growth in humanity. And in [11:06] his encyclical letter recent encyclical letter,  he's critical of this post- or transhumanism uh [11:13] movement. So, um some contrasting perspectives a  source of a need for dialogue about these issues. [11:20] Um okay, so maybe like the most like well-known  sort of problem in education these days is sort of [11:26] AI slop where students are using  AI to plagiarize or cheat um, [11:33] passing off AI slop as their own writing. So, the  result this is an example of something I received [11:37] from a student, um, who did confess to me after  I asked that it was a indeed AI slop. And it was [11:43] a paper about, uh, the mediation between, uh,  technology and our our human experiences. So, [11:47] it's just kind of interesting. Um, so the  result is a text that looks grammatically [11:52] perfect and it sounds smart until you try to read  it and then you find that it says almost nothing. [11:57] It's very hard to find the content. It's very  hard to sort of stay engaged with the writing. [12:01] Um, this is just a quick sample. We're not  sure that cheating is really the worst, uh, [12:07] problem that we we see in this kind of  example, um, so much as this student [12:12] was trying to be efficient and was trying to  optimize and was trying to get a good grade. [12:17] Um, and those values crowded out his ability to  take a moment and think for himself about, uh, [12:24] what does sort of the mediation between  technology and humanity, uh, mean. [12:30] Um, the Google effect, that's been  well documented for a long time. [12:33] Now scientists are sounding the alarm that  AI might be impairing our brain function. [12:39] Okay, so AI is everywhere. Our sense of  ourselves as embodied becomes muddled, uh, [12:45] in the process and learners today face serious  barriers to the development of critical thought. [12:51] What we're skeptical about  is short-term thinking, um, [12:56] slogans that sound like easy solutions. And so  next I'm going to present what we're calling the [13:00] aphorisms and these are bits of advice, uh, that  we've heard offered to teachers over the last few [13:06] years and this happens in professional  development seminars, policy proposals, [13:12] hallway chats between faculty. And a lot of this  advice we find really falls flat and fails to give [13:17] us much actual guidance. So, aphorism number one  and I've heard it put in exactly these terms that, [13:25] uh, we should teach students how to  use AI as a tool and not a crutch. [13:33] Okay, so the issue here, first the framing, a  crutch is literally a tool. Whether we should [13:39] use a crutch depends on our needs. I shouldn't  use a crutch if I don't need one, but if I do [13:44] then I should. Um, and second, there's an issue of  vagueness. All this is really saying is we should [13:50] use AI when it's helpful and we should not use it  when it's not helpful, but the whole question in [13:54] the first place was where do we draw that line?  Um, there are obvious cases where AI has improved [14:00] the human experience when it comes to like medical  technology and being accurate in diagnosis and [14:05] dosing. Um, AI has been operationalized in that  context in ways that just seem obviously good [14:11] and obviously better for people. Um, but in  the classroom, we wonder, so you might think, [14:16] okay, a good place to use AI is to help students  improve their writing. So, should we use it for [14:22] brainstorming? So, some writing professors have  told me that AI is like the great solution to [14:28] writer's block. It can give you an idea to run  with. As a philosophy professor, I take a little [14:34] bit of issue with the idea that we should let the  AI give us the ideas because our whole thing is [14:40] thinking hard to come up with our own idea and our  own sort of reaction to the ideas of others. Um, [14:46] maybe a better use for AI is organizing and  writing. Um, AI can give a really great paper [14:52] outline, but good writing demonstrates  clear thinking. Organizing your thoughts [14:58] for yourself is the kind of thing that makes you  a stronger independent thinker. Um, so we might [15:03] worry about using it in that context as well. And  then editing, I'm probably most comfortable with [15:08] the use of AI for like a quick like grammar  check, um, to to help in the editing process. [15:14] Um, but many professors who I've talked to really  value the skill of reworking a sentence as a part [15:19] of like a careful edit. And again, um, part  of learning how to think clearly, how to come [15:24] up with that sentence that has that structure  that demonstrates your clarity of thought. Okay, [15:31] a different approach, another aphorism, bit of  advice we've heard. Um, we should teach students [15:36] how to use it well, and it's irresponsible  not to. They're going to get screened out, [15:40] um, of sort of job applications if they don't have  these skills, or if they don't have the AI help [15:44] them craft their resume. Um, and there definitely  seems to be truth to this. Um, this is often said, [15:49] I think, as a way of shaming teachers that  resist AI in the classroom. Um, but again, [15:57] like, obviously we need AI literacy. Um,  so a a couple of issues that we have with [16:02] this just offered to us as a slogan. So,  one, what does it mean to use it well? So, [16:07] there are like industry-specific ways to use it  well to get the results that you want. Um, there's [16:13] the question of what it looks like to use it well  ethically, which is a kind of open conversation [16:17] dialogue that we all need to be a part of. So,  this isn't really give you that answer. Um, or [16:21] in terms of AI literacy, and when we talk about AI  literacy, like, what exactly should that encompass [16:26] and does that encompass? Um, the other issue  that we have with this aphorism is that it is it [16:31] either assumes that we're trained teachers, K-12  teachers, higher education teachers, that we're [16:36] trained in ethical AI literacy. We're usually  not. Most higher ed teachers, at least, don't [16:43] get very much training at all in how to be a good  teacher, let alone how to teach alongside AI. Um, [16:50] or it assumes that we've been able to figure  it out for ourselves. And many have. Um, [16:55] I've figured out how to use AI. I  know a lot of people my age and older [16:59] who were not trained in AI use in school, but  now are using it every day in their jobs. Um, [17:07] we think it's telling that a lot of people have  been able to figure this out without having been [17:11] directly taught how to use it. So, we suspect that  having a good foundation of critical thinking, [17:17] expertise, literacy, that's what's really  needed to sort of be good at using it. So, [17:22] I don't think my teachers in school were  irresponsible for not seeing into the future and [17:26] teaching me about uh the AI that I would need  to know when I enter the workforce. Okay. [17:32] Aphorism three, a lot of teachers have said,  we should just discourage its use entirely, [17:39] try to eliminate it from the classroom, and we  should do so by demonstrating its failures. Uh so, [17:46] AI hallucinates, it can uh produce factually  incorrect information. Uh this aphorism becomes [17:53] more outdated uh by the day. Earlier models could  not search or provide references, um while the new [18:01] ones do. Uh new models continue to overcome the  failures of the previous ones at rapid speed. [18:09] But anyway, the problem isn't really whether  AI fails or whether it can always give us [18:13] the right answer. Um the problem is that if  all we do is sort of copy-paste from an LLM, [18:20] we're not reasoning autonomously and cultivating  those skills for ourselves. Okay. So, finally, [18:25] um where can we go from here? Um first, we need  to cultivate AI literacy in students, and we think [18:32] this is not just in the form of effective prompt  engineering, anti-plagiarism, or the ability to [18:38] spot information, but also the ability to think  critically and reflect on how AI impacts our [18:44] embodied, physical, tangible world. We need to  move away from a framing of our relationships with [18:51] AI as focused on like, okay, there's the AI agents  on the one hand and the human users on the other [18:56] hand, um toward a a way of thinking where we've  got human agents alongside AI tools. Second, [19:05] we believe that no individual can answer hard  questions about AI alone. That's what That's why [19:09] I'm kind of trying to give a slogan to teachers to  tell them here's what to do. That doesn't really [19:13] seem to hit. Uh Um what we need is inclusive  dialogue that brings teachers and learners uh [19:20] and industry partners into conversation with  each other. So, this idea is the basis for a [19:25] proposal that my uh co-author and I are working  on for a uh vertically integrated pro- project um [19:32] which basically a vertically integrated it's like  a partnership involves like a partnership between [19:37] like university industry faculty mentorship  and then students at various levels where [19:41] older students uh mentor younger students and we  think uh some kind of structured space like that [19:47] that's inclusive is the kind of space that we need  to have more of these conversations about okay [19:52] that advice doesn't work like what kinds of stuff  will work and how do we need to evolve over time. [19:57] And third uh this is often referred to as active  learning or a polytechnic model in education. So, [20:02] Emily and I work at a polytechnic  institution Utah Tech University. Um so, [20:08] please feel I think I'm running out of time.  So, please feel free to ask us about the kinds [20:12] of classroom assignments that we've been using  to facilitate active learning sort of both in [20:16] relation to AI and not. We need to move toward  a model of education that focuses on hands-on [20:22] embodied experiential learning that deals with  real problems stuff that inspires critical [20:27] thinking the stuff that AI cannot generate or  act out for us. Okay, that's my talk. Thank you.