Cal State Teams With OpenAI, Google to Drive AI Into Local Businesses
Ai4
Closing the AI adoption gap depends on regional state universities rather than elite research hubs because local graduates remain within their communities to deploy practical tools for small and mid-sized employers.
## Background
Published on YouTube by the Ai4 channel on August 16, 2026, this presentation features Edmund Clark, Chief Information Officer at The California State University. According to the description, the session explores how universities, employers, and knowledge workers can form market-responsive partnerships to bridge the AI adoption gap and build scalable workforce readiness pathways.
## TL;DR
Following warnings from major tech firms regarding university preparedness in the AI era, California created the AI Workforce Acceleration Board to leverage the massive footprint of the California State University system. Recognizing that 80% of graduates stay within 50 miles of campus, the initiative targets the adoption gap in local small and mid-sized businesses rather than elite tech giants. The model combines shared, NSF-subsidized compute infrastructure (CENIC AIR) with 10-to-40-hour micro-internships facilitated by intermediary organizations like Breakthrough Tech. This five-part framework provides a replicable blueprint for delivering low-risk student talent and accessible AI solutions to regional economies.
## Atmosphere & style
The presentation carries a pragmatic, policy-oriented, and collaborative tone, blending statewide strategic vision with practical grassroots execution. Edmund Clark delivers a clear, authoritative, and structured briefing that feels like an institutional playbook for public sector innovation. The delivery strikes an optimistic yet urgent rhythm, emphasizing actionable public-private cooperation over theoretical hype.
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=4s) Major tech leaders including Google, OpenAI, Nvidia, and Anthropic warned Governor Newsom in 2024 that universities were not preparing students quickly enough for AI workforce disruption.
- [00:04](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=4s) California established the AI Workforce Acceleration Board via an MOU between the Governor's office and the CSU system as a replicable model for other states.
- [02:49](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=169s) State university systems like CSU are critical AI talent anchors because 80% of their graduates remain within 50 miles of their campus, fueling local regional economies.
- [04:31](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=271s) While individual consumers and big tech firms adopt AI rapidly, small and mid-sized businesses (SMBs) face a major adoption gap that local university talent must help bridge.
- [04:31](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=271s) SMBs need incoming graduates to act as operational evangelists who help identify quick AI wins and transition businesses from chatbot prompting to autonomous AI agents.
- [09:10](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=550s) To eliminate prohibitive compute costs for SMBs and students, California created CENIC AIR, deploying 987 GPUs across the statewide educational research network.
- [09:10](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=550s) The National Science Foundation (NSF) permits up to 10% of National Research Platform capacity to be used for workforce development and student-business collaborations.
- [12:01](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=721s) Traditional 10-12 week summer internships are high-risk and selective, making short, 10-40 hour project-based micro-internships a more scalable and cost-effective alternative.
- [12:01](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=721s) Micro-internships allow students to build diverse multi-project AI portfolios across several employers while earning $20 to $30 per hour.
- [14:56](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=896s) Intermediary organizations like Breakthrough Tech are essential to handle the logistical overhead of project scoping, student cohorts, and stipend distribution.
- [14:56](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=896s) Micro-internship projects focus on practical student evaluation and sandbox testing rather than immediate full commercial product deployment.
- [18:04](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=1084s) The sprint internship framework offers flexible participation pathways for international students, adapting between paid and unpaid structures based on visa and financial aid rules.
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=4s) The Clarion Call: Industry Concerns and California’s AI Workforce Acceleration Board
In 2024, major technology leaders in California—including Google, OpenAI, Nvidia, and Anthropic—approached Governor Gavin Newsom’s office with a pressing concern. They warned state leadership that traditional universities were not adapting quickly enough to prepare students for the looming disruptions in the workforce. In response, the Governor’s office issued an urgent call to California’s higher education systems: the California Community Colleges, the University of California (UC) system, and the California State University (CSU) system.
To confront this challenge, the state and the CSU system established the **AI Workforce Acceleration Board**. While developed in California, the strategic blueprint is designed to be fully replicable across other states and regions looking to develop a sustainable, community-rooted talent pipeline.
- Industry leaders from OpenAI, Google, Anthropic, and Nvidia raised the alarm to Governor Newsom about university workforce preparedness.
- The Governor's office partnered across California Community Colleges, the UC system, and the CSU system.
- An official Memorandum of Understanding (MOU) was signed between Governor Newsom's office and the CSU system to create the AI Workforce Acceleration Board.
- Legislative outreach extended to both the California Democratic Caucus in Washington, D.C. (including Nancy Pelosi and Ro Khanna) and Republican representatives to build bipartisan support.
---
[02:49](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=169s) The CSU Scale and the First Ingredient: State University Systems
State university systems occupy a unique and vital role in economic development compared to elite research institutions. While research universities focus on expanding the frontiers of theoretical knowledge, state universities are specifically charged with educating the local professional workforce. The scale of the California State University system illustrates this immense local footprint: it is the largest four-year public university system in the United States, granting 1 in every 20 college degrees nationwide, with an alumni network of 4.3 million living graduates—surpassing the populations of 23 individual U.S. states.
The primary reason state university systems are indispensable to AI workforce strategy is regional retention: **80% of CSU graduates remain within 50 miles of their campus after graduation**. By training students locally, universities ensure that newly minted AI capability directly enters and sustains local regional economies.
```mermaid
flowchart TD
accTitle: Workforce Development Ecosystem
subgraph S1["1. State University Systems"]
A["CSU & Regional Campuses"] -->|"80% remain within 50 miles"| B["Local Talent Pool"]
end
subgraph S2["2. Economic Engine"]
B --> C["Small & Mid-Size Businesses"]
C --> D["Thriving Local Communities"]
end
subgraph S3["3. Support Infrastructure"]
E["CENIC AIR / NSF AI Infrastructure"] -.->|"Affordable Compute & GPUs"| B
F["Experienced Intermediary (Breakthrough Tech)"] -.->|"Project & Cohort Management"| C
end
```
---
[04:31](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=271s) The Missing Middle: Partnering with Small and Mid-Sized Businesses
Although frontier AI firms sit on advisory boards, the overwhelming majority of university graduates will not work for giant tech enterprises; they will work for local small and mid-sized businesses (SMBs). However, there is a pronounced "adoption gap" across the business landscape. As illustrated in discussions with OpenAI COO Brad Lightcap, individual consumer adoption of AI has surged rapidly, and large tech firms are building deep native capabilities, but small and mid-sized companies remain stranded in the middle without clear paths to implementation.
During a CIO mastermind event in Los Angeles, regional business leaders articulated two core needs: they need incoming graduates to act as "evangelists for new ways of working," and they need accessible expertise to help them identify and deliver initial AI quick wins. Implementing AI within these businesses requires a fundamental mindset shift—moving from simple chatbot prompting to trusting autonomous AI agents with operational decisions.
---
[09:10](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=550s) Ingredient Three: Affordable, Shared AI Infrastructure (CENIC AIR)
A major hurdle preventing small businesses and universities from experimenting with AI is the steep cost of infrastructure, compute, and token credits. Employers worry about paying for interns to waste expensive compute on unproven pilot projects. To solve this, California leveraged its existing statewide research network, **CENIC** (Corporation for Education Networked Initiatives in California).
By securing National Science Foundation (NSF) support, every node across the CENIC research network was equipped with dedicated GPUs, establishing the **CENIC AI Resource (CENIC AIR)**. This infrastructure hosts open-weight foundation models (including DeepSeek, Moonshot, and Kimi), enabling researchers to advance domain-specific knowledge and students to solve practical problems for local industries.
- **CENIC AIR Capacity**: Features 987 GPUs along with substantial CPU compute across the network.
- **Federal Resource Allocation**: The NSF allows up to 10% of National Research Platform capacity to be dedicated to workforce and student-business collaboration projects.
- **Real-World Agricultural Case Study**: At Sonoma State University, students utilized drone imagery and computer vision models running on this shared infrastructure to analyze crop variations at Iron Horse Vineyards, helping winemakers preserve signature flavor profiles across different micro-zones in the vineyard.
---
[12:01](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=721s) Ingredient Four: The Micro-Internship Model
Traditional semester-long summer internships carry immense financial and recruiting risks for employers while failing to scale. Top companies spend heavily on corporate housing and large salaries, only to have interns reject return offers. For students, traditional pipelines filter down thousands of applicants to a lucky few, leaving the rest without practical portfolio experience.
The proposed solution is the **micro-internship**: focused, project-based engagements of 10 to 40 hours, compensated at $20 to $30 per hour. Students work in small cohorts to deliver concrete milestones (such as building a generative campaign in Adobe Firefly for Adobe or solving an automation task for a local business). Over a college career, a student accumulates multiple distinct project experiences on their résumé, while employers tap into a broader, more diverse talent pool with lower financial overhead.
| Dimension | Traditional Internship | Micro-Internship Model |
|:--- |:--- |:--- |
| **Duration & Commitment** | 10–12 weeks full-time (summer) | 10–40 hours total (project-based) |
| **Employer Cost & Risk** | High (housing, full salary, high risk of non-conversion) | Low ($20–$30/hour per deliverable) |
| **Student Reach** | Highly selective (top 1% of applicants) | Broad, inclusive cohorts across multiple majors |
| **Outcome** | Single employer experience | Multi-project AI portfolio across several organizations |
---
[14:56](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=896s) Ingredient Five: The Experienced Intermediary (Breakthrough Tech)
Scaling micro-internships requires managing project scopes, payment distributions, and student cohorts—a logistical load that universities and lean SMBs cannot handle alone. Effective programs rely on an experienced intermediary to bridge the gap.
The CSU system partnered with **Breakthrough Tech**, a national initiative originally out of Cornell University funded by philanthropy to expand AI workforce readiness. Breakthrough Tech structures student cohorts of 3 to 4, vets project deliverables with partner companies, manages student compensation, and runs structured AI skills curricula.
- Over 200 CSU students and 16 faculty members actively teach and participate in the Breakthrough Tech AI curriculum.
- Projects are framed around realistic testing rather than full commercial deployments: "Most of the times you're not going to release whatever they're doing to the broader world. You're just trying to see, does this student get it? Is this a person I want to bring on as a permanent employee?"
---
[16:29](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=989s) Recipe Summary for a Regional AI Talent Engine
The session closed with a recap of the essential five-part "cookbook" needed to replicate this workforce engine in any state:
1. **State University Anchor**: Prioritize regional university systems where the vast majority of alumni remain local.
2. **Small & Mid-Sized Business Focus**: Target the adoption gap in local commercial markets where students serve as AI evangelists.
3. **Subsidized Shared Compute**: Tap into regional research networks (such as CENIC) and NSF allocations (up to 10% platform access) for free GPU compute.
4. **Low-Risk Micro-Internships**: Replace high-stakes traditional internships with short, 10–40 hour cohort sprints to build student portfolios.
5. **Operational Intermediary**: Partner with specialized third-party organizations like Breakthrough Tech to manage projects, stipends, and business relations.
[18:04](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=1084s) International Students and the Sprint Internship Model
In the Q&A portion of the session, an audience member raises a question regarding how the university handles international students in the context of preparing students for work and finding them employment. Edmund Clark begins by acknowledging the inquiry and turning directly to the attendee, checking whether they work at Columbia University to gauge whether they are experiencing similar trends across institutions. Clark notes that across higher education, and specifically observed at the California State University, there has been a notable decline in the number of incoming international students coming over.
Despite these shifting enrollment numbers, Clark points out that the sprint internship framework provides a viable pathway for international students to get involved and gain the practical, career-building experiences universities aim to provide. He observes that while participation is feasible, the structure can vary—some sprint internships may be unpaid depending on visa considerations, work authorization rules, and the potential impact on financial aid—before asking the audience member to clarify whether their concern centers specifically on the financial implications or another operational dimension of supporting international cohorts.
- **International Student Trends:** Cal State, mirroring broader patterns across many higher education institutions, has observed a significant drop in international student enrollment.
- **Sprint Internship Flexibility:** The sprint internship model allows international students to actively participate and acquire hands-on experience, though arrangements may vary between paid and unpaid opportunities depending on financial aid regulations and student eligibility.
## Notable quotes
> [02:49](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=169s) 80% of CSU graduates remain within 50 miles of their campus after graduation — Edmund Clark
> [04:31](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=271s) evangelists for new ways of working — Edmund Clark
> [14:56](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=896s) Most of the times you're not going to release whatever they're doing to the broader world. You're just trying to see, does this student get it? Is this a person I want to bring on as a permanent employee? — Edmund Clark
## Mentioned
Edmund Clark, California State University, Ai4, Google, OpenAI, Nvidia, Anthropic, Gavin Newsom, California Community Colleges, University of California, AI Workforce Acceleration Board, Nancy Pelosi, Ro Khanna, Brad Lightcap, CENIC, National Science Foundation, CENIC AIR, DeepSeek, Moonshot, Kimi, National Research Platform, Sonoma State University, Iron Horse Vineyards, Adobe, Adobe Firefly, Breakthrough Tech, Cornell University, Columbia University
## Video description
Edmund Clark - Chief Information Officer, The California State University
In the AI era, the advantage will go to universities, employers, and workers that can develop capabilities faster than their peers.
As AI reshapes everyday work, the challenge is no longer just adoption. It is aligning education, employer demand, and human capability quickly enough to prepare people for success in a rapidly shifting labor market.
Attendees will learn how universities, employers, and knowledge workers can partner to accelerate workforce readiness, translate emerging AI needs into practical capability, and move from isolated experimentation to scalable talent strategy. They will also gain a clearer view of the human skills, work-based learning models, and upskilling pathways that will define success in an AI-enabled economy.
Attend this session to see how collaborative, market-responsive partnerships can help your organization and your workforce upskill fast enough to compete and win.
Following warnings from major technology companies about lagging workforce readiness, California partnered with the California State University system to establish a regional AI training model designed to equip local businesses with student talent and shared compute.
Regional State Universities as Economic Anchors (0:04)
In 2024, executives from Google, OpenAI, Nvidia, and Anthropic warned California Governor Gavin Newsom that traditional universities were failing to prepare students rapidly enough for workforce disruptions. In response, the state established the AI Workforce Acceleration Board through a Memorandum of Understanding with the California State University (CSU) system, gaining bipartisan support from federal lawmakers including Nancy Pelosi and Ro Khanna.
CSU was selected as the operational anchor because of its sheer scale and local demographic footprint. Granting 1 in 20 college degrees nationwide with an alumni base of 4.3 million, CSU retains approximately 80% of its graduates within 50 miles of their campuses. This high retention rate ensures that state investments in emerging technical competencies directly reinforce regional economies rather than migrating to elite technology hubs.
Targeting the Small Business Adoption Gap (4:31)
While frontier tech companies possess advanced technical capabilities and consumers have rapidly adopted AI tools, small and mid-sized businesses (SMBs) face a pronounced adoption gap. Discussions with OpenAI COO Brad Lightcap and regional business executives at a Los Angeles CIO mastermind highlighted that SMBs frequently struggle to identify practical use cases or afford custom implementation.
Because the vast majority of state university graduates enter local SMBs rather than big tech firms, the initiative trains students to act as workplace evangelists. The goal is to assist local employers in transitioning from basic conversational prompting to autonomous operational agents that can handle concrete business tasks and deliver near-term productivity gains.
Subsidized Public Infrastructure via CENIC AIR (9:10)
Prohibitive infrastructure expenses, token costs, and GPU scarcity often deter small enterprises and universities from testing experimental AI workflows. To mitigate this barrier, California upgraded its existing Corporation for Education Networked Initiatives in California (CENIC) research network to launch CENIC AIR, supplying 987 GPUs alongside CPU capacity across network nodes. The system leverages National Science Foundation provisions allowing up to 10% of National Research Platform capacity to support student-business workforce collaboration.
CENIC AIR hosts open-weight foundation models—including DeepSeek, Moonshot, and Kimi—enabling applied research without enterprise software fees. At Sonoma State University, for example, students used drone imagery and computer vision models running on this network to map micro-zone crop variations at Iron Horse Vineyards, assisting winemakers in protecting specific flavor profiles.
Lowering Risk Through Managed Micro-Internships (12:01)
Traditional 10-to-12-week summer internships carry high recruitment costs and retention risks for employers while excluding the majority of applicants. The California model replaces this with micro-internships: focused 10-to-40-hour project sprints paying students $20 to $30 per hour to deliver discrete milestones, such as building marketing campaigns or automating internal workflows.
To prevent administrative burdens on lean companies and academic departments, the CSU system partnered with Breakthrough Tech, an intermediary originating from Cornell University. Breakthrough Tech organizes cohorts of three to four students, standardizes project scopes, and oversees compensation. Over 200 CSU students and 16 faculty members participate, using low-stakes sprints to showcase student competence without requiring full-scale commercial deployments.
While the framework offers a scalable template for regional workforce development, questions remain regarding international student inclusion. CSU representative Edmund Clark noted a broader institutional drop in international enrollments and observed that visa regulations and financial aid restrictions may complicate international student participation, sometimes limiting these engagements to unpaid arrangements.
## Background
本视频由科技会议频道 Ai4 于 2026 年 8 月 16 日发布于 YouTube，由加州州立大学（CSU）首席信息官 Edmund Clark 担任主讲。内容聚焦于在 AI 时代大学、雇主与知识工作者应如何协同打破技术采纳鸿沟，通过市场敏捷对接与新型实践培养模式，加速打造具备实战能力的规模化 AI 劳动力生态。
## TL;DR
面对科技巨头对大学人才培养脱节的警示，加州州立大学联合州政府与产业界成立了 AI 劳动力加速委员会，探索出以公立高校为支点赋能地方经济的可复制范式。针对绝大多数中小型企业（SMBs）面临的技术采纳断层与算力成本壁垒，加州利用 NSF 资助的 CENIC AIR 骨干网络部署近千张 GPU，为学生提供免除 Token 计费压力的开源大模型实践沙盒。同时，该项目力推 10 至 40 小时的敏捷“微实习”模式，并在专业中介机构 Breakthrough Tech 的协调下，让学生深入索诺马酒庄等真实业务场景化身“AI 布道者”，以低试错成本实现规模化的高校实战与产业赋能。
## Atmosphere & style
整场演讲风格沉稳务实却又不失风趣幽默，呈现出高度接地气的高校实干家气质。Edmund Clark 将庞大的全州教育战略拆解为一本亲切易懂的“实操食谱”，穿插了诸如与纽森州长合影抓拍、女儿在亚马逊实习的沉没成本等鲜活故事，极大拉近了与观众的心理距离。现场没有浮夸的宏大叙事，而是以直面痛点的坦诚、条理清晰的落地路径与充沛的公共关怀，带来了一场既有大局观又充满可操作性的教育变革分享。
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=4s) 2024 年，Google、OpenAI、Nvidia 和 Anthropic 等科技巨头向加州州长预警大学人才培养与 AI 变革脱节，促使加州政府与三大公立高校系统联合成立加州 AI 劳动力加速委员会（AI Workforce Acceleration Board）。
- [01:15](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=75s) 加州州立大学（CSU）作为全美规模最大的公立大学系统，拥有 430 万健在校友，全美每 20 位本科学历获得者中就有 1 人毕业于该校，是赋能本地经济与推动社会流动性的关键支柱。
- [03:38](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=218s) CSU 高达 80% 的毕业生选择在母校方圆 50 英里内工作，因此打造 AI 就绪劳动力的战略重点并非服务硅谷巨头，而是深入赋能广大的本地中小型企业（SMBs）。
- [05:53](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=353s) 当前市场在个人端和科技巨头端 AI 采纳迅速，但在广大中小企业端存在严重采纳断层，中小企业面临资金匮乏、缺乏技术落地经验以及需要重塑管理心智等挑战。
- [09:10](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=550s) 为消除学生与中小企业试错时产生的高昂 Token 和计算积分焦虑，加州利用科研网络与 NSF 支持打造了搭载近千张 GPU 的 CENIC AIR 公共算力平台，免费托管开源权重模型。
- [10:56](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=656s) CSU 强调将 AI 技术注入实际产业场景，如索诺马铁马酒庄利用无人机与 AI 多模态模型分析葡萄地块差异，以保障经典葡萄酒风味的稳定性。
- [12:01](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=721s) 传统暑期实习存在竞争过度、成本极高且试错风险大的问题；CSU 大力推行 10 至 40 小时的“微实习”（Micro-Internship）模式，帮助学生低风险积累实战作品集。
- [14:10](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=850s) 引入 Breakthrough Tech 等专业“体验中介机构”协助企业拆解项目、组建学生小组并统一管理流程，有效解决了高校与中小企业在敏捷项目协同上的管理门槛。
- [16:29](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=989s) 构建全州级 AI 劳动力生态需依托公立大学系统、聚焦中小企业协同、提供普惠算力基建、推行微实习模式以及引入第三方中介服务机构等五大支柱。
- [16:29](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=989s) 灵活的敏捷短周期微实习/冲刺实习（Sprint Internship）为国际学生融入本地 AI 实训网络、积累合规工作经验提供了有效路径。
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=4s) 行业巨头的警钟与加州 AI 劳动力加速委员会的诞生
2024 年，加利福尼亚州的科技产业迎来了一个关键转折点。包括 Google、OpenAI、Nvidia 和 Anthropic 在内的顶级科技巨头集体前往加州州长纽森（Gavin Newsom）的办公室发出警示。这些企业直言不讳地指出：“我们并不确定大学是否正在采取必要措施，让学生为即将到来的职场变革做好充分准备。”这声警钟引发了州政府的高度重视，州长办公室迅速召集了加州的三大公立高等教育系统——加州社区学院（California Community Colleges）、加州州立大学（California State University, CSU）以及加州大学（University of California, UC），直截了当地发问：“面对这种局面，你们打算怎么做？”
为了给出系统性的应对方案，各方联合推动成立了“加州 AI 劳动力加速委员会”（AI Workforce Acceleration Board）。作为加州州立大学的首席信息官（CIO），演讲者 Edmund Clark 强调，本次分享的核心目标并不是单纯展示加州正在做的事情，而是提炼出一套可复制的范式。不论听众身处哪一个州或地区，都可以参考这套“配方”，在自己的社区中培育出具备 AI 能力的劳动力生态。
```mermaid
flowchart TD
accTitle: 加州 AI 劳动力倡议发起架构
subgraph IndustryCall["产业界发声与政府号召"]
A["头部科技企业<br>(Google, OpenAI, Nvidia, Anthropic)"] -->|"敲响警钟：大学人才供给滞后"| B["加州州长办公室<br>(Governor Newsom)"]
end
subgraph HigherEd["高等教育系统协同响应"]
B -->|"问策与行政推动"| C["加州公立高校系统<br>(CSU / UC / 社区学院)"]
C --> D["签署合作备忘录 (MOU)"]
D --> E["成立 AI 劳动力加速委员会<br>(AI Workforce Acceleration Board)"]
end
```
### [01:15](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=75s) 演讲者背景与加州州立大学的庞大体量
Edmund Clark 分享了自己的职业经历与推进此项倡议的历程。在来到加州之前，他在明尼苏达州工作了 17 年，先后任职于明尼苏达大学（University of Minnesota）、明尼苏达州立大学系统以及私立的圣托马斯大学（University of St. Thomas），并曾荣膺明尼苏达州非营利类别的“年度首席信息官（CIO of the Year）”。大约四年前，他正式加入加州州立大学（CSU）。在推进 AI 劳动力加速委员会的过程中，加州州立大学与州长纽森签署了合作备忘录（MOU）。Clark 展示了一张在现场与州长的自拍照，打趣道：“只要有人拍照，纽森州长总能第一时间捕捉到镜头。”此外，就在几周前，他还前往华盛顿特区向加州民主党党团（包括 Nancy Pelosi 和 Ro Khanna 等国会议员）以及共和党成员进行专题汇报，探讨跨党派支持构建全州 AI 劳动力的战略。
谈及 CSU 自身，其庞大体量至今仍令 Clark 感到震撼。作为全美规模最大的公立大学系统，CSU 承担着不可替代的社会流动性与人才供给任务：
- 全美所有拥有大学本科学历的人口中，每 20 人里就有 1 人毕业于加州州立大学系统（CSU）。
- CSU 目前健在的校友人数高达 430 万，这一数字甚至超过了全美 23 个州的单独总人口。
- 该战略旨在打造一个可持续、高影响力的可复制模式，让各个地区都能以州立大学为支点赋能本地经济。
### [03:38](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=218s) 打造 AI 劳动力的核心要素：公立大学与中小企业的协同
Clark 将构建 AI 劳动力的策略比喻为一本“食谱（cookbook）”，列出了成功所需的五大核心原料。其中第一项也是最重要的基石，就是必须与各州的公立大学系统展开深度合作。无论是加州的 CSU、明尼苏达州的明尼苏达州立系统，还是康涅狄格州的公立大学，州立大学系统的天然属性就是具备无可替代的本地深耕能力。
在 CSU，高达 80% 的毕业生在毕业后选择留在距离母校 50 英里（约 80 公里）的范围内工作与生活。这意味着公立大学培养的人才并非流向硅谷孤岛，而是直接输送给周边社区的各行各业，成为支撑本地实体经济发展的骨干力量。
在其余四项要素中，与中小企业（Small and Mid-size Businesses, SMBs）的合作被提到了首位：
- **合作对象脱敏：** 尽管 AI 劳动力加速委员会有 Google、OpenAI、Anthropic 和 AWS 等科技巨擘坐镇，但现实情况是绝大多数毕业生并不会进入超大型科技企业，而是服务于本地的中小型企业。
- **平民化与可负担的基础设施：** 必须建立易于获取、成本低廉的 AI 算力与开发环境，防止因试错成本过高而将学生与小微企业拒之门外。
- **微实习（Micro-internship）模式：** 打破传统的长时间实习壁垒，让学生以轻量、灵活的项目制形式参与企业实践。
- **高水准的中介协调机构（Experienced Intermediary）：** 无论是管理大学生还是管理敏捷的技术项目，对高校院系和中小企业而言都存在极高的实操门槛，必须借助专业的第三方平台和桥梁机构来保证项目交付与落地质量。
### [05:53](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=353s) 中小企业面临的 AI 采纳断层与思维转变
在分析行业采纳现状时，Clark 引用了他与 OpenAI 首席运营官（COO，Sam Altman 的二把手）Brad Lightcap 讨论时绘制的一张草图。Lightcap 指出，AI 展现出了一种极具颠覆性却又异常吊诡的发展态势：在个人端，技术普及极其迅猛，甚至 Clark 已经离世的 85 岁母亲都在日常使用这项技术；在巨头端，开发这项技术的科技大厂自然运用得如鱼得水。然而，在以广阔的中小企业为代表的中间地带，却出现了一个巨大且惊人的“采纳断层（adoption gap）”。
中小企业普遍面临缺乏技术Know-how与资金预算的窘境，以至于像 OpenAI 和 Anthropic 这样的顶尖 AI 公司近期都开始建立专门的咨询业务板块，亲自下场指导中小企业如何落地 AI 技术。
```mermaid
flowchart LR
accTitle: 当前市场 AI 采纳鸿沟示意图
subgraph HighAdoption1["个人端 (Individual)"]
A["个人消费者 / 大众用户<br>(技术自发扩散，全年龄层使用)"]
end
subgraph MiddleGap["中间断层：中小企业 (SMB Gap)"]
B["小微企业 / 独立创业<br>(有部分个人经验外溢)"]
C["中小型企业核心层<br>(严重缺乏技术落地指导与资金)"]
end
subgraph HighAdoption2["巨头端 (Big Tech)"]
D["大型科技公司 / 研发企业<br>(原生采纳，基础设施完备)"]
end
A -->|"有限外溢"| B
B -.->|"断层阻隔"| C
C -.->|"差距悬殊"| D
```
在洛杉矶举办的“CIO 集思会（CIO Mastermind Event）”上，Clark 接触了大量中小企业 CIO，他们一致提出了两重迫切诉求：
1. **期盼毕业生担当“变革布道者”：** 企业直言希望大学毕业生入职时能扮演“全新工作方式的布道者（evangelists for new ways of working）”，主动教导企业该如何重塑工作流。Clark 感叹这对职场新人来说无疑是一项艰巨甚至略显严苛的挑战。
2. **渴求落地指导与快速成效：** 中小企业迫切希望有机构进驻协助打造可衡量的 AI 初期成果（AI wins），而公立州立大学恰恰是充当这一外脑的最佳合作伙伴。
3. **在职员工的心智转型（Mindset Shift）：** AI 的落地绝非仅仅是“在日常业务中塞进一点 ChatGPT 就会改变一切”。企业面临的核心瓶颈是信任机制与管理心智的重构——“我们能否真正放手让 AI 智能体（Agent）去独立做决策？它能够做哪些维度的决策？”这种面对不确定性与失控感的心理跨越，需要系统性的再培训方案。
### [09:10](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=550s) 基础设施破解成本困境：国家科学基金会与 CENIC AIR 算力网络
面对中小企业 CIO 们反映的另一大痛点——AI 工具与模型调用的高昂成本，大学探索出了公共基础设施层面的破局之道。中小企业普遍对让年轻实习生在未经验证的项目中随意“烧掉代币和计算积分（wasting tokens and credits）”感到心惊胆战。为了消除阻碍试错的高昂门槛，必须提供低成本、高可用的公共计算平台。
Clark 指出，全美各州普遍建有服务于高校科研的区域性“研究与教育网络（Research and Education Networks）”。加州的对应骨干网络即为 CENIC（加利福尼亚州教育网络倡议联盟）。借助国家科学基金会（NSF）的资助，加州团队提出了一项关键构想：既然全州分布着密集的科研网络节点，为什么不直接在每个网络节点部署 GPU 算力集群，从而面向所有师生和研究人员免费托管 AI 开源模型？
这项倡议最终孵化为了 CENIC AI 资源库（CENIC AIR，CENIC AI Resource）：
- 全州节点整合并搭载了 987 张高性能 GPU 以及大量 CPU 计算资源。
- 该公共网络专门托管主流开源权重模型（Open-weight models），为学生提供无计费压力、免受 Token 消耗焦虑的安全沙盒。
- 它搭建了连接高校、学生实验与中小企业实际业务需求的纽带，使得地方企业无需承担昂贵的基础设施初期投入，即可借助大学的人才与算力平台展开 AI 赋能测试。
### [10:56](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=656s) AI 模型落地与地方产业赋能：以索诺马酒庄为例
加州州立大学（CSU）的研究人员与学生正在积极探索各类前沿开源与商业大模型（例如 DeepSeek、Moonshot 旗下 Kimi 等），并将这些工具拓展至制造业、计算机科学等领域的研究前沿。然而，相较于纯理论探索，加州州立大学更注重将 AI 能力注入具体的中小型企业场景，解决现实中的业务痛点。
在加州著名的索诺马（Sonoma）葡萄酒产区，学生们将 AI 工具与无人机航拍技术结合，应用于葡萄种植监测。例如在知名的铁马酒庄（Iron Horse Vineyards），学生利用无人机采集葡萄园的高清图像数据，并通过模型分析不同地块之间葡萄风味产生差异的原因，协助酒庄精准调整种植管理，确保整片葡萄园产出的果实都能保持酒庄标志性的独特风味。这种让学生直接参与中小型企业真实业务的模式，不仅为地方产业创造了数字化价值，也让学生掌握了扎实的 AI 实操经验。
```mermaid
flowchart LR
accTitle: AI 赋能地方产业实践流程
subgraph S1["数据采集"]
A["无人机航拍与地块勘测"] --> B["获取葡萄园视觉数据"]
end
subgraph S2["AI 分析与优化"]
B --> C["多模态模型分析差异"]
C --> D["锁定风味差异成因"]
end
subgraph S3["落地赋能"]
D --> E["赋能铁马酒庄保持风味一致性"]
end
```
---
### [12:01](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=721s) 传统实习的困境与“微实习”（Micro-Internship）模式
传统暑期实习模式在规模化和投入产出比上面临严峻瓶颈。传统实习竞争极度激烈，成百上千名学生竞争极少数岗位，未能入选者便失去实践机会；同时，企业也要承担极高的招募成本与沉没风险。Edmund Clark 分享了自己女儿的亲身经历：她曾获得亚马逊（Amazon）在西雅图的实习机会，公司为其支付了高昂薪资并全额承担公寓租金，然而实习期结束后，女儿却决定不在亚马逊继续发展——这种高昂的试错成本对雇主而言是一笔巨大的开支与风险。
为了解决这一痛点，加州州立大学力推以项目为导向的“微实习”（Micro-Internships）机制：
- **灵活精炼的工时**：每个项目周期仅为 10 至 40 小时，企业按每小时 20 至 30 美元的标准支付报酬。
- **高针对性的实战项目**：学生以团队形式协作攻坚具体课题。例如，学生团队为 Adobe 设计基于 Adobe Firefly 的全新创意营销方案；或为 OpenAI 及本地中小型企业交付针对性 AI 解决方案。
- **动态积累的作品集**：学生在完成项目后可将成果沉淀在个人 LinkedIn 页面与简历中，在大学四年间建立起多元、厚实的 AI 实战作品集，向未来雇主直观证明自身的业务落地能力。
---
### [14:10](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=850s) 雇主收益与“体验中介机构”的枢纽作用
微实习模式不仅让学生在不同专业背景下磨炼出 AI 应用与职业沟通技能，也为企业重构了人才招聘与技术引入的通道。企业无需再从数千份简历中盲选极少数人，而是能以极低的成本和风险接触到更多元化的优质人才池。同时，带着创新方法进驻企业的学生团队，还能在组织内部扮演“AI 布道者”角色，为现有团队的工作流变革带来启发。
微实习的高效运转离不开“体验中介机构”（Experience Intermediary）的专业化运营。由于企业并不需要实习生直接上线生产级系统，而是希望检验其思维模式与问题解决能力，中介机构能够协助企业提炼交付目标、组建 3 至 4 人的学生小组，并统一管理薪资发放与合作关系维护。
加州州立大学在此领域的核心合作伙伴是由慈善基金资助、源自康奈尔大学的 Breakthrough Tech：
- 推动 AI 就绪型人才培养体系在校企之间快速对接。
- 截至当前，已有超过 200 名学生及 16 名教职人员参与该技能培训课程。
- 通过结构化的小组协作，大幅减轻了企业管理零散实习生的行政负担。
---
### [16:29](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=989s) 打造 AI 就绪劳动力的关键要素与国际学生考量
Edmund Clark 总结了构建全州级 AI 就绪劳动力生态系统所必需的关键支柱，强调了不同高等教育机构的差异化分工：
| 核心支柱 | 职能与实施路径 |
|:--- |:--- |
| **州立大学系统** | 区别于以拓展人类知识边界为核心的研究型大学，州立大学的使命是为本地经济输送高素质的专业实干人才。 |
| **中小企业协同** | 绝大多数毕业生最终就职于本地广大中小企业，校企合作必须深植于地方商业生态。 |
| **普惠计算基础设施** | 借助美国国家科学基金会（NSF）国家研究平台（NRP）政策，最高可将 10% 的算力资源免费用于支持学生与企业的 AI 实践。 |
| **微实习扩展机制** | 替代高成本、高风险的传统实习，通过敏捷短周期项目实现大规模技能认证与人才交付。 |
| **中介服务机构** | 由 Breakthrough Tech 等第三方专业机构统一协调项目定义、资金结算与校企关系。 |
在演讲结束后的问答环节中，来自哥伦比亚大学（Columbia University）的参会者就“国际学生如何参与该培养体系”进行提问。Edmund Clark 回应指出，虽然全美高校的国际生生源数量近期呈现整体下降趋势，但微实习/冲刺实习（Sprint Internship）具备高度灵活性，国际学生同样可以通过项目制模式（包括结合助学金与合规要求的灵活参与形式）融入实训网络，积累宝贵的工作经验。
## Notable quotes
> [00:04](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=4s) 我们并不确定大学是否正在采取必要措施，让学生为即将到来的职场变革做好充分准备。 — 头部科技企业代表
> [00:04](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=4s) 面对这种局面，你们打算怎么做？ — 纽森州长办公室
> [01:15](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=75s) 只要有人拍照，纽森州长总能第一时间捕捉到镜头。 — Edmund Clark
> [01:15](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=75s) 全美所有拥有大学本科学历的人口中，每 20 人里就有 1 人毕业于加州州立大学系统。 — Edmund Clark
> [05:53](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=353s) 全新工作方式的布道者 — 中小企业 CIO
> [05:53](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=353s) 我们能否真正放手让 AI 智能体（Agent）去独立做决策？它能够做哪些维度的决策？ — Edmund Clark
> [09:10](https://www.youtube.com/watch?v=4S3U_mQMTOQ&t=550s) 既然全州分布着密集的科研网络节点，为什么不直接在每个网络节点部署 GPU 算力集群，从而面向所有师生和研究人员免费托管 AI 开源模型？ — Edmund Clark
> 在 AI 时代，优势将属于那些能够比同行更快培养能力的大学、雇主和劳动者。 — Edmund Clark
## Mentioned
加州州立大学 (CSU), 加州大学 (UC), 加州社区学院 (California Community Colleges), 加州 AI 劳动力加速委员会 (AI Workforce Acceleration Board), Edmund Clark, 纽森 (Gavin Newsom), Google, OpenAI, Nvidia, Anthropic, AWS, Brad Lightcap, Nancy Pelosi, Ro Khanna, 明尼苏达大学 (University of Minnesota), 圣托马斯大学 (University of St. Thomas), CENIC (加利福尼亚州教育网络倡议联盟), CENIC AIR, 美国国家科学基金会 (NSF), 国家研究平台 (NRP), DeepSeek, Kimi (Moonshot), 铁马酒庄 (Iron Horse Vineyards), Adobe, Adobe Firefly, Breakthrough Tech, 康奈尔大学 (Cornell University), 哥伦比亚大学 (Columbia University)
## Video description
Edmund Clark - Chief Information Officer, The California State University
In the AI era, the advantage will go to universities, employers, and workers that can develop capabilities faster than their peers.
As AI reshapes everyday work, the challenge is no longer just adoption. It is aligning education, employer demand, and human capability quickly enough to prepare people for success in a rapidly shifting labor market.
Attendees will learn how universities, employers, and knowledge workers can partner to accelerate workforce readiness, translate emerging AI needs into practical capability, and move from isolated experimentation to scalable talent strategy. They will also gain a clearer view of the human skills, work-based learning models, and upskilling pathways that will define success in an AI-enabled economy.
Attend this session to see how collaborative, market-responsive partnerships can help your organization and your workforce upskill fast enough to compete and win.
[Source](https://www.youtube.com/watch?v=4S3U_mQMTOQ)
Source quotes
[0:04] 头部科技企业代表: 我们并不确定大学是否正在采取必要措施，让学生为即将到来的职场变革做好充分准备。
[0:04] 纽森州长办公室: 面对这种局面，你们打算怎么做？
[1:15] Edmund Clark: 只要有人拍照，纽森州长总能第一时间捕捉到镜头。
[1:15] Edmund Clark: 全美所有拥有大学本科学历的人口中，每 20 人里就有 1 人毕业于加州州立大学系统。
[5:53] 中小企业 CIO: 全新工作方式的布道者
[5:53] Edmund Clark: 我们能否真正放手让 AI 智能体（Agent）去独立做决策？它能够做哪些维度的决策？
[9:10] Edmund Clark: 既然全州分布着密集的科研网络节点，为什么不直接在每个网络节点部署 GPU 算力集群，从而面向所有师生和研究人员免费托管 AI 开源模型？
Edmund Clark: 在 AI 时代，优势将属于那些能够比同行更快培养能力的大学、雇主和劳动者。
Source quotes
[2:49] Edmund Clark: 80% of CSU graduates remain within 50 miles of their campus after graduation
[4:31] Edmund Clark: evangelists for new ways of working
[14:56] Edmund Clark: Most of the times you're not going to release whatever they're doing to the broader world. You're just trying to see, does this student get it? Is this a person I want to bring on as a permanent employee?
Original transcript
[0:04] Good morning and I'm so excited to be here
today to talk about building an AI-ready
[0:09] workforce together. Um I'll give you a little
bit of background first. I'm I live and work in
[0:15] California. I'll talk about that a little bit,
but what was happening in 2024 is our industry
[0:21] partners, and when I'm when I mean by industry
partners, I mean uh big companies in the state of
[0:26] California. Um Google, Open AI, Nvidia, Anthropic,
you name it. They were going to the governor's
[0:34] office, Governor Newsom's office, and they were
saying, "We're not sure that universities are
[0:38] doing what they need to do to prepare students
for what's about to happen in the workforce."
[0:45] Right? So, that was kind of this clarion call.
Everybody was like, "What are we going to do about
[0:49] it?" And so, go the governor's office reached
out to all of the uh systems in the California
[0:55] uh state, so like community colleges, California
State Universities, which I'm part of,
[1:00] and the University of California system, and said,
"What are you all going to do about it?" And one
[1:04] of the things we did we decided to create was this
thing called the AI Workforce Acceleration Board,
[1:10] which I'm going to talk about in a little bit, but
that's part of the background of where we came to.
[1:15] Today, I'm going to go through
a little bit of background,
[1:17] talk about the goals of the model because yes,
I'm in California. You may not be in California,
[1:22] but the idea is that you could replicate
this model wherever you are as well.
[1:26] And then the ingredients you're going to need
to be successful, and then we'll recap it,
[1:31] and I'll say thank you, and I'll be done in 20
minutes, hopefully. So. Um Yes. Okay. Um so, uh
[1:41] a little bit about me. Um in the lower right, you
know, I was at the University of Minnesota for
[1:46] many years, 17 years. I was in the Minnesota State
University system. I was I was at the University
[1:51] of St. Thomas, which is a private college in
Minnesota. And as a result of all these efforts,
[1:57] I was eventually named a CIO of the year at uh for
the state of Minnesota in the nonprofit category.
[2:04] But this AI Workforce Acceleration Board, you
know, there's an MOU between the California State
[2:09] University system and Governor Newsom's office,
and that selfie you see in the upper right,
[2:13] that's me. And then you can see Governor Newsom
in the background there. He when someone's taking
[2:18] a photo, he always looks up. So, you can see he
he caught himself up there. He signed that thing,
[2:23] which was very a proud moment for me and the
Cal State system. And then just recently,
[2:28] uh a couple of weeks ago, I was in uh Washington,
D.C., and you can see I was presenting to
[2:32] uh the California Democratic Caucus, you know, so
you can see Nancy Pelosi and Ro Khanna there. But
[2:37] then we also did have conversations with the
Republican side as well about how to build
[2:42] the workforce, um the AI-ready workforce for the
state of California. Um so, a little bit about Cal
[2:49] State. Um you know, I I came to Cal State from
Minnesota about 4 years ago. So, some of these
[2:55] stats are still mind-blowing to me. So, I'll just
pick out a couple of them, which is, you know,
[3:00] it we are the nation's largest university system,
and one in 20 people with a college degree in the
[3:06] United States earned it at the CSU. That's a big
impact, not only in California, and but beyond.
[3:13] Um and our alumni, you know, the the 4.3 living
alumni is bigger than the population of 23 uh
[3:20] states. So, huge impact in California and beyond.
And um the goal here was to create a replicable
[3:29] approach for you to develop a sustainable and
impactful AI workforce for your communities.
[3:38] So, I'm just going to go through it like a
cookbook. You know, what ingredients do you
[3:41] need here? Um the first, in my opinion, you
should work with a state university system.
[3:47] Many of you come from states, and you might be
where I was at Minnesota. We have a Minnesota
[3:52] State University system. Or you might be in
Connecticut, they have a state university system.
[3:56] Why state university systems? Because they
have so much local impact. In the CSU,
[4:02] 80% of our graduates stay within 50 miles of
their campus after graduation. What that means
[4:08] is they're going to have a local impact. They're
going to be employees in your local communities.
[4:13] They're going to help your local businesses
thrive. And then when you have that happening,
[4:18] that local community grows. And that's what all
of our states want and need, right? So that's an
[4:23] important part. State university systems, people
stay local and they help build your state up.
[4:31] And so the other ingredients, two through five,
you also need to work with small and mid-size
[4:35] firms. Um although on our AI workforce board
because they called for it, we have Google,
[4:41] we have Anthropic, OpenAI, AWS, you name it,
but they're probably on our workforce board.
[4:47] Most of our students aren't going to work for
these large tech firms. They're going to work
[4:51] for your local community small and mid-size
businesses. So that's why again, to have this
[4:57] scale up and be impactful, you want to really have
partnerships with your small and mid-size firms.
[5:03] You also need an affordable and accessible AI
infrastructure. I'm going to talk about that a
[5:06] little bit. A micro-internship model, I'm going to
talk about that. And then um a highly experienced
[5:13] intermediary. By that I mean managing students is
hard. Managing projects is hard. Universities and
[5:22] even your small and mid-size firms aren't going
to have the experience that is probably necessary
[5:26] to deliver this. So have a good good partner and
I'm going to I'll give you a suggestion on that.
[5:31] Okay. We talked about small and mid-size
firms. Why? Because of that local impact.
[5:38] If your small firm starts to thrive and
grow and your businesses are growing,
[5:42] then that community gets built up. Your town will
grow. It'll have the kind of economic impact that
[5:47] you want for all of your citizens. Small and
mid-size firms are really key to this model.
[5:53] And uh this is something uh I was talking to
Brad Lightcap. He is the uh chief operating
[6:00] officer for OpenAI. So, he's second in command
under Sam Altman, right? And he drew this chart.
[6:06] I'll make I made prettier versions of it, but
we were kind of brainstorming around this.
[6:11] But what the chart says in general is
like AI is a very strange technology.
[6:15] Individuals have adopted it in mass. I mean,
every even my mom was using this technology,
[6:20] right? She passed away recently, but she was using
it and she was 85 years old. And then obviously
[6:28] the big firms that invented the technology, you
know, on the far right, they've been using it.
[6:32] But there's this gap. There's a little bit of
little bit of carryover for individuals going to
[6:37] small businesses because individuals start small
businesses, right? So, there's a little carryover,
[6:42] but there's a sort of like adoption gap in the
middle, small and mid-size businesses, that's
[6:48] whoa, weird. And if you can believe it, OpenAI
and Anthropic have noticed this and have started
[6:53] all these consulting functions around like, "Hey,
we're going to help small and mid-size businesses
[6:57] figure out how to leverage these technologies."
Cuz it's a huge market they're missing, right?
[7:03] Um and of course our AI strategies, we don't
want there to be a gap. You saw on the left,
[7:07] non-STEM a little bit lower than STEM, so
we want to There shouldn't be a gap. We
[7:11] want all of our students to graduate with
a certain level of skill in their context.
[7:18] But I already talked about this. 80% of
our graduates are going to work in small
[7:21] and mid-size firms in local communities near
their universities, and that's where we want to
[7:26] have the impact with you and your communities.
And this again applies to any state you're in.
[7:34] So, I was hosting a bunch of CIOs at a what
we call a CIO mastermind event in Los Angeles.
[7:42] And again, it was a lot of small and
mid-size businesses. And they were saying,
[7:45] "Look, what we really need is your students
to graduate, and they're they need to be
[7:48] evangelists for new ways of working." That's
a big challenge for new students, right? Like,
[7:53] "Hey, they're going to come in. We're going to
show you how you should do your work." I mean,
[7:56] that's a big challenge, right? But that was the
desire that was um kind of uh uh expressed there.
[8:04] And then uh many of them also said this, "Hey,
do you know a firm that can come in and teach
[8:08] us how to build you know, some AI wins in our
environment?" Again, this gets to that that
[8:14] knowledge and adoption gap in the middle there,
right? What are we going to do about this? And uh
[8:20] I would say that your local state university
is a good partner for this kind of solution.
[8:26] And then, retraining So, not only the students
coming in and being these evangelists, but
[8:31] we need to retrain our existing workforce
because, as you know, many of you know,
[8:35] the it implementing AI in your environments is
not only just like, "Hey, we're going to do a
[8:40] little bit of chat GPT in here, and it's going to
change everything." It's really a mindset shift.
[8:45] Can we let that agent go out and make decisions?
And what kinds of decisions? Those are things that
[8:51] it takes a while for you to say, "Okay, I'm going
to be okay with this." It's a little scary. It's a
[8:56] mindset shift, right? And that's part of what we
need to do. But this is part of our AI strategy
[9:01] because we've been asked, again by the state of
California, to build the workforce for the state.
[9:10] Um so, the third ingredient you'll
need is an accessible and affordable
[9:14] AI infrastructure. When I talked to
those CIOs at small and mid-size firms,
[9:18] a lot of them talked about how expensive
it is to get into the AI tooling, and
[9:24] how scary it is to say, "Hey, we're going to
have interns, students wasting all our, you know,
[9:28] like tokens and credits on like experimenting
in this context." Right? That's expensive.
[9:34] So, you know, um there is a model I'm about to
show you where the National Science Foundation is
[9:40] has supported this. And it's a great
place for students to experiment.
[9:44] And it allows universities and businesses
to connect to support the local community.
[9:50] So, this map is a map of of the the research and
education networks in the United States. These are
[9:58] all networks that support research at universities
everywhere. And I've I've circled in the dotted
[10:03] circle CENIC. That's California's network.
Why is this important? It Well, first of all,
[10:10] it shows you that wherever state you are, you
probably have a research network available.
[10:16] So, in CENIC, uh one of our uh one of our board
members had this brilliant idea that look, we
[10:22] have this research network and every one of those
nodes could be hosting some GPUs. Let's ask the
[10:28] NSF to load up every node with GPUs so we can host
AI models for students and researchers to use.
[10:36] And that's what led to CENIC AIR, the CENIC AI
resource. Right? And that's So, that thing that uh
[10:43] that serves our California state includes a bunch
of GPUs. And it looks like uh what, 987 GPUs and
[10:50] a lot of CPUs as well. And they have And we're
hosting a lot of open weight models where students
[10:56] and researchers can team together. And they're
Mostly these Chinese models, right? So, meaning
[11:01] the Deep Seek, Moonshot, which you've heard about,
Kimmy. Um but what happens is our researchers
[11:08] are using these tools to expand the boundaries of
knowledge, whether it's in you know, you name it,
[11:13] manufacturing or computer science. And our
students are using it in these small business
[11:19] context. I'll give you an example. At Sonoma, you
know, there's a big wine making area in Sonoma.
[11:25] And uh our students started using these tools
to figure out at certain vineyards, hey, what's
[11:32] happening in this crop? How can drones, first of
all, using drones to look at the crops and gather
[11:37] all that visual data, but you know, they noticed
that flavors were different from one part of the
[11:42] vineyard to the other. How can we capture this
one flavor that Iron Horse Vineyards is known for?
[11:48] Um and make sure that all the other
grapes taste the same way. And again,
[11:53] our students were using AI at that small and
mid-size firm to solve that problem for them.
[12:01] So, the fourth model is a micro-internship model.
The standard internship model is not scalable
[12:08] because what happens is you get a lot
of students applying for an internship,
[12:13] one or a few students gets that internship. The
rest of them are out of luck. That's not going to
[12:18] be scalable in the big big term, and it's a lot of
risk for the employer. I'll give you an example.
[12:24] I mean, my daughter get mad that I'm bringing
this up. She got an internship with Amazon,
[12:29] and they paid for her apartment in Seattle,
gave her a huge salary, right? So, she's
[12:35] moose in mini apples in Seattle, gets a huge
salary, and at the end of the internship,
[12:39] she decides, I don't want to work for Amazon. No
No offense to any Amazon folks out here. But, um
[12:46] huge risk. She went through all this stuff.
They went through all this stuff. They paid a
[12:49] lot of money, and then she decided she didn't
want to work for them anymore. Right? So,
[12:54] that's a lot of expense. Micro-internships are a
way to come down to like a very small snippet, a
[13:01] project-based thing, 10 to 40 hours. You might pay
20 to 30 dollars an hour for that 10 to 30 hours,
[13:08] uh 10 to 40 hours. And then what happens to for
the student is, oh, I went to Adobe. They wanted
[13:14] me to create a new creative campaign using Adobe
Firefly for something they're trying to promote.
[13:20] I worked with a team of other students to do
that. And if Adobe likes it, they can put on on
[13:25] LinkedIn page, I did this for Adobe. Then they
can do the next project for Open AI. You name it.
[13:32] Or their local businesses. What we hope is that
at the end of that, by the time they graduate,
[13:38] they've gained this experience and they'll have
a very impressive resume, a portfolio of things
[13:43] they've done with AI across their entire academic
career. We're hoping that allows you to say, "Man,
[13:52] this person could really make a difference in
our our our organization." The other thing is,
[13:58] as they build their experience and they contribute
to your businesses, again, they're going to
[14:03] actually impact your local communities using
that AI infrastructure I talked about before.
[14:10] I'm going to speed up a little bit. So,
the benefits are the students will get the
[14:13] professional skills. They get to leverage
AI in different career and major paths.
[14:19] They develop the portfolios and you get
a talent pipeline. You might find talent
[14:24] you wouldn't have found otherwise. You know,
when you get out when out there and say, "Hey,
[14:27] I got a thousand applicants. I'm going to get five
of them for my internship." Instead, you start
[14:32] to see a broader talent pipeline, more diverse
talent pipeline that might identify candidates
[14:37] you wouldn't have found otherwise, and it's
lower risk and lower cost to you as employers.
[14:42] And then you want students to evangelize new
ways of workers working. They're going to come
[14:46] in as these cohorts and they're going to come
in and say, "Hey, here was a way we solved that
[14:50] problem for you." And hopefully that is also
an inspiration to your current organization.
[14:56] And then finally, you're going to need an
experience intermediary because managing
[15:00] students and micro-internships is hard, right?
You got to define the right projects. You know,
[15:06] I mean, someone was talking to me about
this yesterday on the plane as we flew over.
[15:10] I was like, "No, no, actually, you don't
when you're doing an internship and your
[15:14] interns are working on deliverables, most
of the times you're not going to release
[15:18] whatever they're doing to the broader world.
You're just like, you're just trying to see,
[15:22] does this student get it? Is it someone that Is
this a person I want to bring on as a permanent
[15:26] employee? Right? So, you define the right
project, something they can deliver for you.
[15:32] They get managed by our firm, and they manage
the money and the relationships so that the
[15:36] businesses say, "Yeah, that was worthwhile. I
want to go in again next year and do this again."
[15:43] The partner we've been working with most
is a company or company organization called
[15:47] Breakthrough Tech. They came out of Cornell.
They were funded by philanthropy to do exactly
[15:51] what I'm talking about. How can we build
AI ready people in the workforce? And uh
[15:59] I can't say enough about Breakthrough Tech.
Um I see I'm running out of time, so you know,
[16:04] at this point, you know, we have over 200 students
and 16 faculty teaching that skills curriculum.
[16:10] Our students are going through these programs um
in cohorts, so like cohorts of three and four.
[16:17] The The firm manages projects on their behalf. Uh
um of of you, and then they manage students and
[16:22] payments, and that's kind of the the the big model
that I wanted to share. And so, the ingredients,
[16:29] you know, state university systems have local
impact. Don't forget about your state university
[16:33] systems. Um research universities are great. Their
mission is to expand the boundaries of knowledge.
[16:39] State university systems are there to build
a professional workforce for your state.
[16:45] You need the partnerships with small and
mid-size firms because most of your students
[16:49] most of your community will not be working for a
Google. They're going to be working for all the
[16:54] businesses around you. You're going to need that
accessible AI infrastructure that's affordable.
[16:59] The NSF has said, "Hey, up to 10% of the
time on the national research platform can be
[17:06] can be dedicated to this work that I'm talking
about." So, you could be leveraging free AI GPUs
[17:13] on projects that help students, universities,
and your businesses. The micro internship model
[17:18] is a way to scale up and be effective instead
like that high risk internship thing that I was
[17:23] talking about. I'm going to owe my my my daughter
an apology after this for bringing up her story.
[17:29] And then finally an experienced intermediary
is what you're going to need to be successful
[17:34] because they'll manage all that hard work for
you. Managing students, payments, relationships,
[17:39] and projects. And um I want to end there so that
at least there's a couple minutes for questions.
[17:46] Thank you very much. If you have any questions you
want to reach out to me, that's my email address.
[17:53] Uh thank you so much for an exceptional
presentation. I spoke earlier, I work for
[17:58] Columbia. My question to you is and I love what
you guys are doing with helping students get ready
[18:04] for work and finding them employment. The question
I have is how do you handle your international
[18:09] students? Yes, well, thank you for the question.
The question about international students,
[18:15] one thing that has happened at at many higher
ed education inst- you work at Columbia? So,
[18:21] you may or may not be seeing this the same
the pattern we're seeing at Cal State is a
[18:25] lot of our international students, we've seen a
big decline in the numbers that are coming over
[18:30] and the the the good thing about the sprint
internship model is that they can participate
[18:36] in it, you know, and and and and get the
experiences that, you know, you would want,
[18:41] but um some of those sprint internships can
be unpaid. It depends on how the financial
[18:45] aid impact is and all that, but um are you talking
about the financial implications or which parts?
