Puri Warns Big Tech AI Talent Monopoly Starves US Industrial Base

Democratizing AI Talent: Bridging Policy, Innovation, and Equity With RegulatingAI — Ai4

Ai4

Clustering 77% of AI jobs in ten coastal metros threatens critical healthcare and manufacturing sectors, requiring open-source tooling, community college training, and regional workforce mandates to avoid an entrenched economic divide.

AI-assisted summary of Democratizing AI Talent: Bridging Policy, Innovation, and Equity With RegulatingAI — original source. Check the original for full context.

Source published:

Detailed notes

## Background Published on YouTube by Ai4 on August 27, 2025, "Democratizing AI Talent: Bridging Policy, Innovation, and Equity With RegulatingAI" features a keynote address by Sanjay Puri, Chairman and Founder of RegulatingAI. The presentation examines how policy decisions, technical innovation, and an intentional focus on equity can dismantle barriers in AI talent development and ensure career opportunities are accessible to all rather than confined to a privileged few.

## TL;DR Sanjay Puri addresses the severe concentration of AI talent, research, and venture capital within a handful of elite universities and coastal tech giants. This geographic and corporate bottleneck leaves essential sectors like manufacturing, healthcare, and energy starved of the engineering expertise necessary to modernize. To prevent a damaging new societal divide, Puri outlines a framework centered on community college partnerships, public compute access via NAIRR, open-source advocacy, and remote work adoption. He calls on policymakers and industry leaders to cultivate regional AI ecosystems across every state, classroom, and community nationwide.

## Atmosphere & style The talk carries an urgent yet accessible and conversational tone, grounded by a relatable personal anecdote about Puri's own son defaulting to a Big Tech career path. Delivering a pragmatic wake-up call rather than an antagonistic critique, Puri combines sharp statistics with lighthearted humor about Silicon Valley perks to emphasize broad economic realities. The session feels like an earnest, constructive policy briefing that builds toward an inspiring call to action for widespread regional equity.

## Key takeaways - [00:04](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=4s) Next-generation AI talent is hyper-concentrated within a handful of Silicon Valley mega-corporations, narrowing the focus of technological innovation and depriving foundational sectors of diverse problem solvers. - [03:00](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=180s) Geographic and academic concentration remains severe: 40% of top AI researchers reside in four elite universities, 77% of AI jobs cluster in ten metro areas, and over 50% of venture capital flows exclusively to Silicon Valley. - [05:18](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=318s) Astronomical compensation packages and multi-billion-dollar acqui-hires by major tech companies create an elite tier that starves mid-tier institutions and legacy enterprises of vital technical talent. - [05:18](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=318s) Critical non-tech sectors such as manufacturing, healthcare, finance, and logistics face existential talent droughts that threaten their modernization and operational resilience. - [10:12](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=612s) Democratizing the talent pipeline requires expanding beyond four-year elite university degrees into community college curriculums, vocational credentials, and corporate reskilling programs. - [10:12](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=612s) Federal initiatives like the National Artificial Intelligence Research Resource (NAIRR) serve as crucial public utilities, offering free GPU compute and benchmark datasets to underfunded researchers and institutions. - [12:51](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=771s) State-level partnerships, such as Wisconsin's collaboration with Microsoft to train 100,000 residents, alongside K-12 programs like AI4ALL, are vital for cultivating early and diverse AI literacy. - [13:32](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=812s) Platforms like CAIO Connect highlight high-impact enterprise AI deployments in traditional industrial sectors to prove to emerging engineers that meaningful AI careers exist outside Big Tech. - [13:32](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=812s) State and local governments should leverage land and permitting approvals for over $200 billion in hyperscaler data center investments to require firm commitments for local AI workforce upskilling. - [15:59](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=959s) Protecting and advancing open-source AI infrastructure is essential to ensuring baseline foundation models remain accessible to everyone rather than locked behind proprietary corporate monopolies. - [15:59](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=959s) Embracing remote and distributed work models allows organizations to access overlooked talent pools across rural and secondary markets without forcing relocation to expensive coastal hubs. - [17:19](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=1039s) Failing to distribute AI education, talent, and career opportunities across all regions and industries risks spawning an entrenched AI divide far more consequential than the digital divide of past decades.

## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=4s) A Personal Awakening: The Concentration of AI Talent

Opening the morning session at the AI Policy Summit with a humorous remark about wondering "where is the AI talent this morning," Sanjay Puri sets an urgent stage for why artificial intelligence talent must be democratized across regions, industries, and demographic communities. For the United States to lead responsibly in the global AI race, innovation cannot remain confined to a privileged few. Puri explains that this mission is deeply personal: his own son recently graduated with a degree in AI. When asked where he and his classmates intended to look for work, his son replied "without batting an eyelid" that they were all targeting the same three or four global tech giants. While acknowledging that these are phenomenal companies—and noting with a chuckle that his son did end up taking a job at one of them—the realization struck a serious chord.

If the entirety of the next generation's AI talent converges on a handful of mega-corporations located within two or three zip codes, the broader ecosystem suffers an acute loss of intellectual and geographic diversity. AI is not an isolated consumer app; it is a foundational, general-purpose technology destined to transform disease treatment, childhood education, supply chain logistics, and global commerce. Puri poses a fundamental question to practitioners, founders, and policymakers: can the nation afford to let its technological future be shaped by a narrow group of people working in isolated geographic enclaves on an equally narrow set of problems?

```mermaid flowchart TD accTitle: The AI Talent Concentration Dilemma accDescr: Flowchart illustrating how talent concentration leads to narrowed innovation versus how democratization enables broad societal impact.

subgraph Narrow["Current Concentration Model"] A["Top AI Graduates & Capital"] --> B["3-4 Tech Giants in Silicon Valley Hubs"] B --> C["Narrow Focus & Monolithic Problem Solving"] end

subgraph Inclusive["Democratized AI Model"] D["Regional & Diverse Talent Pools"] --> E["Cross-Industry Distribution: Healthcare, Retail, Manufacturing"] E --> F["Broad Societal Solutions & Responsible Global Leadership"] end ```

---

[03:00](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=180s) The Anatomy of Concentration: Elite Universities, Metro Hubs, and Capital

To underscore the severity of the bottleneck, Puri presents a set of stark empirical metrics illustrating the hyper-concentration of knowledge, employment, and investment capital across the United States. Today, approximately 40% of the top AI researchers in the country are concentrated in just four elite universities. While these academic institutions are undeniably world-class, this centralization creates a self-reinforcing gravity well that siphons prospective talent away from other regional institutions and undercuts academic diversity nationwide.

The geographic disparity is equally pronounced in the workforce and venture landscape. Fully 77% of all AI jobs in the United States are clustered within just ten metropolitan areas, leaving the remaining 300 cities to divide a meager 23% share. Furthermore, during the first half of the year, over 50% of all venture and private AI investment was poured directly into Silicon Valley. Puri emphasizes that this critique is not an attack on Northern California, playfully offering his full blessing to their "free kombuchas" and "nap pods," but rather a pragmatic plea to ensure that the remaining 99% of the nation participates in and benefits from the AI-driven economy.

- **40% of Top Researchers:** Housed within just 4 elite universities nationwide. - **77% of AI Employment:** Clustered within 10 major metropolitan cities, marginalizing roughly 300 other American urban centers. - **50% of H1 Investment:** Flowed exclusively into Silicon Valley startups and corporate labs.

---

[05:18](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=318s) The Broader Impact: An Elite Class and the Industry Talent Crunch

This hyper-concentration has created an ultra-elite compensation tier, where tech conglomerates routinely spend hundreds of millions of dollars on individual researcher packages or deploy $14 billion to execute "acqui-hires" of fledgling AI teams. Puri points out that while high compensation is accepted in sports leagues like the NBA or NFL, artificial intelligence is vastly different because it governs critical societal infrastructure like diagnostic medicine and public education. When a tiny cohort of researchers captures all resources, mid-tier researchers and non-tech enterprises are left completely starved of human capital.

``` +--------------------------+------------------------------------------------+---------------------------------------------+ | Industry Sector | Key AI Applications & Growth Drivers | Consequences of the AI Talent Crunch | +--------------------------+------------------------------------------------+---------------------------------------------+ | Manufacturing | Predictive maintenance, supply chain optimization| Factory downtime and stagnant automation | | Healthcare | Clinical diagnostics, rural transcription notes | Slower medical delivery and regional bias | | Financial Services | Real-time fraud detection, algorithmic trading | Heightened security risk and legacy drag | | Retail & Logistics | Precision targeted marketing, inventory routing | Inefficient distribution and margin erosion | +--------------------------+------------------------------------------------+---------------------------------------------+ ```

Beyond the tech sector, foundational industries face an existential talent drought despite massive Gross Value Added (GVA) projections by 2035. Manufacturing plants require machine learning specialists for predictive maintenance and supply chain resilience; healthcare networks need developers for diagnostic models and transcription workflows; and regional finance needs algorithmic fraud detection. However, because top graduates gravitate almost exclusively toward Silicon Valley, legacy and regional enterprises cannot recruit the expertise required to modernize, creating deep economic vulnerabilities across non-coastal and rural America.

---

[10:12](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=612s) Rebuilding the Talent Pipeline: Alternative Education, Reskilling, and Policy Initiatives

To correct these structural imbalances, Puri outlines a strategic blueprint focused on expanding educational access points, corporate upskilling, and bipartisan policy partnerships. Democratizing talent requires looking beyond traditional four-year computer science degrees and establishing accessible pathways through community colleges, vocational certifications, and online platforms offering specialized credentials. He highlights Nvidia's statewide partnership with California community colleges as a prime model for bringing cutting-edge hardware and curriculum directly to two-year students, as well as corporate upskilling programs like Walmart's internal workforce reskilling initiatives.

At the legislative and federal level, public policy must act as a foundational equalizer. Puri points to the administration's AI Action Plan, which prioritizes national talent pipeline development, alongside the National Artificial Intelligence Research Resource (NAIRR). Because individual researchers, local universities, and small businesses cannot afford expensive GPU clusters or proprietary datasets, NAIRR serves as a vital public utility by providing free, democratized access to high-performance compute and benchmark datasets, enabling innovation outside the major tech corridors.

- **Multi-Tier Education:** Incorporating two-year community colleges, vocational programs, and online nano-degree platforms into mainstream AI training. - **Industry Initiatives:** Replicating programs like Nvidia's California community college partnership and Walmart's internal worker reskilling programs. - **Federal Compute Access:** Leveraging NAIRR (National Artificial Intelligence Research Resource) to supply free GPU compute capacity and rich datasets to under-resourced universities and small enterprises.

[12:51](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=771s) State Initiatives, Hyperscaler Partnerships, and K-12 Education

Individual states are increasingly stepping up to establish their own artificial intelligence ecosystems rather than waiting for federal guidance. A prime example is Wisconsin, which is actively working toward building a dedicated regional AI hub in collaboration with Microsoft. Through this initiative, the state plans to train 100,000 residents in AI skills, demonstrating the massive leverage state governments possess when negotiating and partnering directly with technology providers.

At the foundational end of the pipeline, efforts must start well before higher education. Programs such as the ORISE AI challenges introduce technical problem-solving to primary and secondary school students. Similarly, initiatives spearheaded by Stanford professor Fei-Fei Li—widely recognized as the "godmother of AI"—such as her AI4ALL program, are instrumental in embedding AI literacy directly into K-12 education across diverse and historically underrepresented student populations nationwide.

- **State workforce training:** Wisconsin's partnership with Microsoft aims to upskill 100,000 residents for the emerging AI economy. - **Early pipeline intervention:** Programs like AI4ALL and the ORISE AI challenge target K-12 students to cultivate early technical literacy and broad inclusion.

---

[13:32](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=812s) Showcasing AI Beyond Big Tech: The Role of CAIO Connect

To expand the perception of where AI careers exist, Puri highlights the work of his nonprofit, CAIO Connect (Chief AI Officers Connect). The core mission of the organization is to spotlight enterprise AI leadership outside the handful of massive tech giants or FAANG corporations. Groundbreaking, high-impact machine learning and automation projects are quietly taking place across traditional manufacturing, retail, logistics, and heavy industrial platforms. CAIO Connect brings these leaders onto public stages so they can showcase what their teams are building and explain why emerging engineers and graduates should consider working for them.

Major industrial corporations are already deploying cutting-edge AI in complex physical environments. Johnson Controls uses AI to manage advanced facility climate and temperature control systems; Schneider Electric leverages machine learning across massive electrical and battery infrastructures; Grainger applies predictive tracking and catalog search across millions of office and industrial supply products; and global enterprises such as Michelin and Emerson Electric are transforming their engineering workflows. Puri stresses that students need to realize that cool, transformative AI work is not the exclusive domain of Silicon Valley consumer tech platforms.

Furthermore, with hyperscalers projected to invest over $200 billion across the nation to build the next generation of critical AI data centers, state and local governments must leverage their permitting and land resources. States should require that these massive capital expenditures include firm commitments to train and upskill local workers in applied AI technologies.

```mermaid flowchart TD accTitle: Multi-Sector AI Ecosystem Beyond Big Tech accDescription: Flow diagram showing CAIO Connect bridging diverse traditional industries and tying data center infrastructure investments to local workforce training.

subgraph Industry["Traditional Enterprise AI Applications"] JC["Johnson Controls<br/>Facility & Temperature Automation"] SE["Schneider Electric<br/>Energy & Battery Optimization"] GR["Grainger<br/>Industrial Supply Tracking"] MC["Michelin & Emerson Electric<br/>Manufacturing & Heavy Industry"] end

subgraph Policy["Infrastructure & State Policy"] DC["$200B Hyperscaler Data Centers"] REQ["State Mandates:<br/>Local Workforce AI Training"] DC --> REQ end

subgraph CAIO["CAIO Connect Platform"] HUB["Chief AI Officers Connect"] end

Industry --> HUB REQ --> HUB ```

---

[15:59](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=959s) Actionable Paths: Community Colleges, Open Source, and Distributed Work

Addressing the broader challenge of democratizing AI requires simultaneous interventions across education, software accessibility, and employment practices. Puri places heavy emphasis on community colleges as the primary engine for accessible, affordable workforce retraining, urging strong public-private partnerships to fund curriculum design and lab access.

Equally vital is defending and supporting open-source AI development. Echoing the popular metaphor that "AI is the new electricity," Puri warns that proprietary monopolies risk restricting who gets access to the grid: "well open source makes sure the lights are on for everybody not just for a select few." Keeping foundation models and research tools accessible prevents the concentration of power among a handful of tech conglomerates.

Finally, organizations must intentionally revive remote and distributed work models. Although corporate mandates have pulled many workers back into offices following the pandemic, Puri notes that rigid in-person requirements needlessly alienate qualified individuals in rural areas and smaller towns who cannot or do not wish to relocate to expensive coastal hubs. Because AI development can be executed from anywhere with an internet connection, embracing distributed teams is an essential mechanism for tapping into overlooked geographic talent pools.

- **Community colleges:** Serve as the most effective local vehicle for democratized, low-cost technical education. - **Open-source ecosystem:** Ensures baseline technological access remains universally available rather than locked behind proprietary walls. - **Distributed employment:** Enables talent in rural regions and secondary markets to participate fully in the AI economy without geographic displacement.

---

[17:19](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=1039s) Reversing the AI Divide: A Distributed National Talent Base

Returning to his personal anecdote about his graduating son, Puri envisions a future where the next generation of top engineering and computer science graduates actively choose diverse sectors over the default allure of Big Tech. A healthy national ecosystem is one where graduates say, "we want to work for a manufacturing company in Michigan or we want to work for a energy company in Texas or we want to work for a healthcare company in Ohio." Distributing top-tier machine learning expertise across the Midwest, the South, and every regional industry creates broad-based national productivity rather than localized wealth extraction.

Puri closes with an urgent warning about the social stakes of the transition. Failing to distribute AI opportunities will spawn an "AI divide" far more consequential, damaging, and entrenched than the digital divide of past decades. True democratization demands that AI talent, education, and career paths exist "in every city, every classroom, and every community," urging the audience to actively engage in building an inclusive talent pipeline.

## Notable quotes > [00:04](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=4s) where is the AI talent this morning — Sanjay Puri

> [00:04](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=4s) without batting an eyelid — Sanjay Puri

> [03:00](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=180s) free kombuchas — Sanjay Puri

> [03:00](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=180s) nap pods — Sanjay Puri

> [05:18](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=318s) acqui-hires — Sanjay Puri

> [12:51](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=771s) godmother of AI — Sanjay Puri

> [15:59](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=959s) AI is the new electricity — Sanjay Puri

> [15:59](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=959s) well open source makes sure the lights are on for everybody not just for a select few — Sanjay Puri

> [17:19](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=1039s) we want to work for a manufacturing company in Michigan or we want to work for a energy company in Texas or we want to work for a healthcare company in Ohio — Sanjay Puri

> [17:19](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=1039s) in every city, every classroom, and every community — Sanjay Puri

## Mentioned Sanjay Puri, RegulatingAI, AI Policy Summit, Silicon Valley, Nvidia, California Community Colleges, Walmart, AI Action Plan, National Artificial Intelligence Research Resource (NAIRR), Microsoft, Wisconsin, ORISE AI, Fei-Fei Li, Stanford University, AI4ALL, CAIO Connect, Johnson Controls, Schneider Electric, Grainger, Michelin, Emerson Electric

## Video description Sanjay Puri, Chairman/Founder, RegulatingAI

Abstracts

Understand how to make AI talent development more inclusive and accessible to people from all backgrounds. The discussion will explore how policy decisions, technological innovation, and a strong focus on equity can work together to ensure that AI education, careers, and opportunities are not limited to a privileged few. This discussion aims to bridge the gap between who has access to AI opportunities today and who should, ensuring a future where AI talent development is fair, inclusive, and widespread.

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

中文简报

普里警告：科技巨头垄断AI人才正在削弱美国工业基础 77%的AI岗位集中在十大沿海都会区，对医疗和制造业等关键领域构成威胁；亟需开源工具、社区学院培训及区域劳动力措施以避免经济鸿沟固化。 桑杰·普里（Sanjay Puri）指出，AI人才、研究成果和风险投资正严重集中于少数精英高校与沿海科技巨头。这种地域和企业层面的瓶颈，导致制造业、医疗保健和能源等关键行业极度缺乏推进现代化所需的工程专业人才。为防止造成破坏性的新型社会鸿沟，普里提出了一个行动框架，重点包括开展社区学院合作、通过NAIRR提供公共算力支持、倡导开源工具以及推广远程办公。他呼吁政策制定者与行业领袖在全美各个州、课堂和社区中积极培育区域性AI生态系统。 全美 AI 人才、就业与资本正高度向少数中心集中：约 40% 的顶尖 AI 研究人员集中在 4 所顶尖高校，77% 的 AI 职位分布在 10 个都会区，上半年更有超过 50% 的风投资金全部涌入硅谷。 (3:00) 科技巨头高昂的薪酬与巨额并购引发了严峻的人才虹吸效应，导致制造、医疗、金融与物流等传统基础产业面临生存性的人才短缺，严重阻碍其现代化转型。 (5:18) 打破人才垄断需突破传统四年制精英大学的限制，转向社区大学课程与企业转岗培训，并依托国家人工智能研究资源（NAIRR）向欠发达机构免费提供 GPU 算力及基准数据集。 (10:12) 通过威斯康星州与微软合作培训 10 万名居民、AI4ALL 推进 K-12 教育普及，以及 CAIO Connect 平台展示传统工业领域的落地成果，正在向新生代工程师证明科技巨头之外同样具备深远的 AI 职业机遇。 (12:51)

Reading summary

Sanjay Puri argues that the extreme concentration of artificial intelligence talent and capital within a handful of elite universities and tech giants threatens foundational American industries, asserting that broad economic resilience requires democratizing AI access through alternative education, public compute infrastructure, and regional employment. Hyper-Concentration of Talent and Investment (3:00) According to metrics cited by Puri, 40% of top U.S. artificial intelligence researchers are concentrated in just four elite universities, creating an institutional gravity well that diminishes academic diversity nationwide. The geographic disparity is similarly pronounced across employment and venture financing: 77% of all domestic AI jobs are clustered in ten metropolitan areas, leaving the remaining 300 American cities to share a 23% portion. Furthermore, over 50% of all private and venture AI investment in the first half of the year flowed directly into Silicon Valley startups and corporate laboratories. Puri warns that when prospective graduates default exclusively toward three or four coastal tech conglomerates, the broader economic ecosystem suffers an acute loss of intellectual and geographic variety. Because AI is a foundational general-purpose technology rather than an isolated consumer application, confining development to a few geographic enclaves risks narrowing technical innovation to a limited set of corporate priorities. Critical Industry Talent Shortages Beyond Big Tech (5:18) The concentration of capital has generated an elite compensation tier where major tech companies routinely spend hundreds of millions on individual packages or deploy $14 billion on team acqui-hires. Consequently, non-tech enterprises and foundational sectors face an acute talent shortage despite strong long-term Gross Value Added projections. Puri notes that without machine learning talent, manufacturing plants face production downtime from stalled predictive maintenance, healthcare networks struggle to scale clinical diagnostics in rural regions, and regional financial firms face heightened fraud and security risks. To counteract the perception that transformative AI is exclusive to Big Tech, Puri highlights his nonprofit, CAIO Connect, which showcases machine learning leadership within traditional enterprises. Companies such as Johnson Controls (facility temperature management), Schneider Electric (battery and electrical optimization), Grainger (industrial supply tracking), Michelin, and Emerson Electric are actively deploying applied AI across physical infrastructure, demonstrating that impactful engineering careers exist outside consumer software giants. Policy Interventions and Multi-Tiered Education (10:12) Rebuilding the talent pipeline requires looking beyond four-year computer science programs by integrating community colleges, vocational credentials, and internal reskilling initiatives. Puri highlights Nvidia's statewide partnership with California community colleges to provide advanced hardware and curricula, Walmart's internal worker reskilling programs, and Wisconsin's state partnership with Microsoft to train 100,000 residents. Early-intervention programs such as the ORISE challenges and Fei-Fei Li's AI4ALL similarly seek to embed technical literacy directly into K-12 education for underrepresented student groups. At the federal level, public policy must serve as an equalizer against computational bottlenecks. Puri emphasizes the National Artificial Intelligence Research Resource (NAIRR), which provides free high-performance GPU compute and benchmark datasets to under-resourced universities and small enterprises. Additionally, with hyperscalers projected to invest over $200 billion nationwide in new data centers, Puri argues state and local governments must leverage zoning and permitting to mandate local workforce AI training commitments. Preserving Open Ecosystems and Distributed Employment (15:59) Democratizing AI also requires institutional support for open-source development and flexible work arrangements. Drawing on the comparison of AI to electricity, Puri warns that proprietary monopolies risk controlling who has access to the power grid, whereas an open-source ecosystem ensures foundational tools and research models remain accessible across the entire economy. Finally, Puri urges organizations to actively embrace distributed and remote work. Post-pandemic office mandates needlessly exclude qualified workers in rural areas and secondary markets who cannot or choose not to relocate to high-cost coastal tech hubs. Because machine learning development requires only internet connectivity, distributed hiring models offer an immediate mechanism for spreading technical opportunity across non-coastal communities. Puri concludes that failing to actively disperse AI opportunities will produce an "AI divide" far more damaging and entrenched than historical digital divides, reiterating that long-term national competitiveness depends on distributing technical talent across every state, classroom, and traditional industry.

精炼阅读摘要

Sanjay Puri在AI政策峰会上指出，美国人工智能领域顶尖人才、学术资源与风险资本正高度向少数科技巨头及沿海核心城市集中，直接导致医疗、制造等传统实体行业陷入严重的技能短缺。他呼吁通过扶持社区学院、提供国家级公共算力、捍卫开源生态以及推行分布式办公，实现AI人才与创新的全面民主化，防止技术分化演变为更深固的社会鸿沟。 学术、就业与资本的高度集中现状 (3:00) Puri列举的一组实证数据显示，美国顶尖AI研究人才中有约40%集中在仅4所精英大学内，形成自我强化的吸附效应并削弱了区域高校的学术多样性。在产业与就业层面，全美77%的AI岗位聚集在仅10个大都会区，其余约300座城市仅能分摊剩余的23%；而在上半年，超过50%的风险投资与私人AI资金直接涌入了硅谷。这种高度垄断构建出了超高薪酬壁垒，大型科技企业往往斥资数亿美元招募单个研究人员或动用高达140亿美元通过收购兼并吸纳初创技术团队，使非科技领域的企业及中层研发力量在人才竞争中处于极端劣势。 核心实体产业面临的人才枯竭危机 (5:18) Puri强调，AI是一项将重塑疾病诊疗、基础教育及供应链的通用型底层技术，而非单一的消费级应用。高额薪酬集中化在职业体育联盟或许可行，但AI掌管着关乎社会命脉的基础设施。尽管制造业、医疗、区域金融与零售物流行业到2035年具有庞大的总增加值（GVA）预期，却正因人才枯竭而面临严峻的现代化阻碍：工厂难以部署预测性维护和供应链优化算法，区域医疗缺少临床诊断与转录开发人员，金融业面临实时反欺诈技术滞后，从而加剧了非沿海及农村地区的经济脆弱性。 多元教育管道与公共算力基础设施 (10:12) 为拓宽技术培育路径，Puri主张摆脱单一依赖四年制计算机本科学位的模式，将两年制社区学院、职业资格认证与线上纳米学位纳入培养体系。实践范例包括Nvidia与加州社区学院合作引入硬件与前沿课程，以及Walmart内部开展的员工技能重塑计划。在基础教育阶段，ORISE AI挑战赛与李飞飞（Fei-Fei Li）发起的AI4ALL等项目正致力于在K-12学生及代表性不足群体中普及AI素养。在政策层面，联邦政府的《AI行动计划》与国家人工智能研究资源（NAIRR）被视作关键杠杆，NAIRR能够向财力有限的高校和小微企业免费提供急需的高性能GPU算力与基准数据集；同时，威斯康星州等地方政府也与Microsoft达成合作，计划为该州培训10万名具备AI技能的劳动者。 拓展非科技巨头场景与基建培训绑定 (13:32) 针对人才只认准少数大型科技公司的现象，Puri创立的非营利组织CAIO Connect致力于向公众推介传统行业的AI应用，打破“尖端技术仅存在于硅谷”的固有偏见。当前，Johnson Controls在楼宇温控自动化中引入AI，Schneider Electric在大型电力与电池系统中使用机器学习，Grainger在数百万工业耗材中应用预测检索，Michelin和Emerson Electric也在重工业制造流程中推行智能化。此外，面对超大规模云服务商未来预计超过2000亿美元的全国性AI数据中心基础设施投资，Puri主张州和地方政府应充分利用土地与许可审批权，明确要求企业承担相应的本地劳动力AI实操技能培训承诺。 开源生态与分布式办公消除地域壁垒 (15:59) 在技术获取与工作制度方面，Puri将开源生态视为防止垄断的技术基石。他引用“AI即新时代的电力”这一比喻强调，专有垄断可能会切断外部接入，而开源则确保基础大模型和研究工具对全社会保持开放，避免权力完全集中于少数巨头。另一方面，企业必须重新重视远程与分布式办公模式。尽管疫情后许多企业恢复了现场办公要求，但强行要求员工驻场无形中排斥了中西部、南部及农村地区不愿或无法搬迁的合格人才；鉴于AI研发只需依靠互联网即可完成，推行分布式工作机制是激活全美潜在人才库的有效途径。 Puri总结认为，如果顶尖工程人才的职业归宿始终局限于极少数科技巨头，全美将面临比以往数字鸿沟更为顽固的“AI鸿沟”；唯有引导顶尖机器学习人才主动流向密歇根的制造业、得克萨斯的能源业或俄亥俄的医疗体系，在每个城市、教室和社区建立普及性培训与就业路径，才能建立起兼顾区域平衡与负责任的全国性生产力网络。

Detailed notes — zh

## Background 本内容由 Ai4 于 2025 年 8 月 27 日发布在 YouTube 平台上。视频记录了 RegulatingAI 主席兼创始人 Sanjay Puri 的主题演讲，主要探讨如何通过政策制定、技术创新与对公平性的重视，打破 AI 人才与教育资源的阶层壁垒，推动 AI 人才发展的普惠与包容。 ## TL;DR Sanjay Puri 指出当前全美 AI 领域的顶尖人才、研发资源和风险资本过度集中在极少数科技大厂与沿海都会区，导致传统支柱产业与偏远地区面临严重的“AI 人才枯竭”。他强调 AI 是一项关乎全人类未来的通用技术，绝不能沦为少数群体的排他性工具。为此，他呼吁重塑多元化人才培养通道，大力扶持社区大学、K-12 早期启蒙与开源生态，并借助国家级算力资源（如 NAIRR）与政企深度合作打破垄断，让技术创新真正惠及医疗、制造等广阔的实体经济与每一个社区。 ## Atmosphere & style 演讲在轻松幽默与深刻警醒之间切换自如，开场以儿子求职科技巨头的真实生活趣事切入，瞬间拉近了与观众的距离。随着一组组触目惊心的垄断数据被抛出，现场氛围逐渐转为严肃而充满紧迫感，演讲者以饱含激情的呼吁直击“AI 鸿沟”的系统性危机。整场演讲叙事层层递进、发人深省，既有对现实困境的犀利剖析，又洋溢着对技术普惠和区域公平发展的坚定信念与感召力。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=4s) AI 人才的民主化具有高度紧迫性，必须防止顶尖技术能力和资源仅被少数科技巨头垄断在极少数几个邮政编码内。 - [03:32](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=212s) 全美 40% 的顶尖 AI 研究人员仅集中在 4 所大学，77% 的 AI 相关岗位扎堆于 10 个核心都会区，资本与人才呈现出极端的地理失衡。 - [03:32](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=212s) 今年上半年全美流向 AI 领域的风险投资中超过 50% 涌入了硅谷，导致美国其他广阔地区面临技术与资本红利的脱节。 - [05:18](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=318s) 科技巨头开出的巨额年薪与大规模人才并购正在形成超级溢价，对处于中间梯队的科研人员、跨区域企业和非科技机构产生严重的挤出效应。 - [05:18](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=318s) AI 是关乎医疗、教育与社会治理的通用基础技术而非竞技体育，必须坚持以包容性而非排他性来引领发展，注入多元视角构建模型。 - [07:36](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=456s) 制造业、零售业、基层医疗和金融等实体产业对 AI 赋能有巨大需求，但因顶尖人才流向单一而正经历严重的“AI 人才枯竭”。 - [10:11](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=611s) 打破唯名校论并依托社区大学、在线纳米学位（如 Coursera）及企业员工技能重塑（如沃尔玛），是拓宽多元化 AI 人才管道的关键。 - [10:11](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=611s) 国家人工智能研究资源（NAIRR）等公共政策项目致力于为全美高校和小微企业提供免费 GPU 算力与数据集，打破科技巨头的算力垄断。 - [12:51](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=771s) 地方政府正通过政企合作推进区域人才培训，例如威斯康星州携手微软计划培训 10 万名本地劳动力掌握 AI 技能。 - [12:51](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=771s) AI 启蒙教育需尽早下沉至 K-12 基础教育阶段，如李飞飞发起的 AI4ALL 项目致力于为多元背景的青少年播下科技探索的种子。 - [13:32](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=812s) 非营利组织 CAIO Connect 致力于展示江森自控、施耐德电气、固安捷等传统实体企业的 AI 实践，引导人才跳出科技大厂光环并投身实体创新。 - [13:32](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=812s) 随着超大规模云厂商在全美投入超 2000 亿美元建设数据中心，各州政府应设立硬性条件，要求其为本地社区提供长期的 AI 技术培训。 - [15:59](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=959s) 开源生态是防止技术垄断的基石，如同确保所有人都能用上电力的保障，能让 AI 技术红利普惠广大公众。 - [15:59](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=959s) 重新拥抱远程与分布式办公模式能够激活乡村及偏远地区的人才潜能，防止衍生出危害更深远的“AI 鸿沟”。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=4s) 个人经历与核心议题：AI 人才为何不能只留在“几个邮政编码” 会议伊始，Sanjay Puri 带着轻松幽默的语气开场，调侃自己大清早还在琢磨“AI 人才都去哪儿了”，随后切入本次演讲的核心主旨：推动 AI 人才普及具有极高的紧迫性，必须让顶尖技术能力延伸至更多地区、行业与社区。他强调，AI 人才的民主化（Democratizing AI talent）本质上是为了确保创新能够在全美乃至全球范围内全面铺开，从而让美国在负责任的前提下引领全球 AI 竞争。 这一宏大议题对 Puri 而言有着极其切身的体会。他的儿子不久前刚从大学获得 AI 学位毕业，当 Puri 询问儿子及其同窗计划去哪里找工作时，对方甚至连眼睛都没眨一下，脱口而出便是“那三四家全球科技巨头”。虽然这些企业确实非常优秀（他的儿子最终也确实入职了其中一家），但这番对话让 Puri 深感震动。如果全美顶尖的 AI 人才全部被吸纳进极少数科技大厂、集中在“几个邮政编码”（a few zip codes）之内，未来的技术生态将会变成什么样？ 如果 AI 正在成为重塑人类疾病治疗、下一代教育模式以及供应链物流调配的通用基础技术，那么社会绝不能允许这项决定未来的关键技术仅仅由生活在狭窄地域、视角高度单一的一小群人去定义和构建。面对台下的 AI 专家、创业者与政策制定者，Puri 正式抛出了这场关于包容性、人才多样性与国家竞争力的深刻反思。 ### [03:32](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=212s) 触目惊心的数据：高校、城市与资本的极端集中 为了直观展现 AI 领域的失衡现状，Puri 列举了一组极具冲击力的统计数据。这种高度集中不仅是简单的市场行为，更演变成了一种正在自我强化的壁垒。 全美 40% 的顶尖 AI 研究人员仅仅集中在 4 所基础大学之内；而在就业市场上，全美 77% 的 AI 相关岗位高度扎堆在仅仅 10 个核心大都市圈，其余 300 多个城市只能瓜分剩下不到四分之一的就业机会。在资本端，仅在今年上半年全美流向 AI 领域的风投资金中，就有超过 50% 直接涌入了硅谷。Puri 幽默地补充道，这绝非针对硅谷冷嘲热讽，“如果你们来自硅谷，祝贺你们，我完全支持你们享用免费的红茶菌（kombucha）和睡眠舱（nap pods）”，但他真正的诉求是确保美国其他广袤地区同样能够分享到 AI 时代的红利。 - **顶尖高校集中度**：40% 的顶尖 AI 研究学者来自全美仅 4 所顶尖大学。 - **就业岗位分布**：全美 77% 的 AI 职位被 10 大都会区垄断，全美其余 300 多个城市面临严重人才真空。 - **资本吸附效应**：上半年全美 AI 投资总额中，超过 50% 的资金完全沉淀在硅谷。 ### [05:18](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=318s) 顶层精英化 vs 普惠创新：高薪并购下的系统性失衡 资本与人才的极度集中直接拉大了产业间的断层，催生出一个孤立的“AI 精英阶层”。Puri 指出，当下科技巨头动辄开出数亿美元的年薪包，或是斥资数十亿甚至上百亿美元进行大规模“人才并购”（acqui-hire）。虽然顶尖从业者获得高薪无可厚非，但这种超级溢价正在对广大处于中间梯队（mid-tier）的科研人员、跨区域企业和非科技机构造成巨大的挤出效应。 有些人可能会辩解称这只是市场规律，如同体育界中 NBA 或 NFL 永远由少数超级巨星统治一样。但 Puri 严肃地反驳了这种观点：AI 绝不是竞技体育，它是一项直接关乎全人类治愈绝症、培育下一代以及社会公平治理的通用技术。一个健康的国家必须坚持“以包容而非排他来引领发展”（leading by inclusion not by exclusion）。如果在系统构建之初就缺乏来自不同区域、文化背景和垂直行业的思维多样性，开发出的 AI 模型必然陷入狭隘，难以解决现实世界中极其复杂多样的问题。 ### [07:36](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=456s) 行业断层与地理鸿沟：传统产业的 AI 人才荒 当前的讨论往往过度聚焦于科技行业（Tech Sector），但 AI 应用的广阔天地实际上存在于传统支柱领域。无论是制造业的预测性维护与供应链管理、零售业的精准营销，还是医疗领域的病理诊断与服务递送，亦或是金融业的反欺诈与算法交易，这些行业在 2035 年的总增加值（GVA）预期都将依赖 AI 的深度赋能。然而，恰恰是这些至关重要的民生与实体产业，正在经历绝对意义上的“AI 人才枯竭”（talent crunch）。 高学历的顶尖 AI 毕业生几乎极少流向制造业或中西部传统企业，导致这些行业空有巨大的转型需求却缺乏创新实施能力。与此同时，地理层面的鸿沟同样严峻。地图上象征 AI 机会的绿点几乎全部分布在东西海岸，中西部与广大乡村小镇被严重边缘化。偏远地区的基层医疗机构急需 AI 工具来转写医生病历、辅助诊断以缓解人手短缺，但当地企业和公共部门根本雇不到专业人才。正如 Puri 所总结的：“人才无处不在，但机会却并未平等眷顾每一个人。” ```mermaid flowchart TD accTitle: AI 发展失衡现状与传统产业人才荒 subgraph Imbalance["资源与人才过度集中"] A["资本与高薪垄断<br/>(硅谷 & 少数科技巨头)"] B["科研与岗位扎堆<br/>(4所名校 & 10大都市圈)"] end subgraph Shortage["实体产业与区域断层"] C["传统行业人才荒<br/>(制造/医疗/零售/金融)"] D["地理鸿沟<br/>(中西部、乡村及基层医疗)"] end Imbalance -->|"虹吸效应 / 缺乏包容"| Shortage ``` ### [10:11](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=611s) 破解之道：重塑人才管道与政策基础设施支持 面对人才与地理的双重脱节，Puri 提出必须从人才培养管道重构、政企战略合作以及算力基础设施普惠等多个维度同步发力。重塑人才管道的核心在于打破唯名校论，开辟多元化的人才准入路径（entry points），尤其是充分发挥全美庞大的两年制社区大学（Community Colleges）体系的作用。例如，英伟达（Nvidia）已经在加州与社区大学展开深度合作，定向培训 AI 技能；同时结合各类职业技能认证、Coursera 等在线纳米学位平台，实现广泛的技术普及与在职员工技能重塑（upskilling/reskilling，如沃尔玛正在推进的员工 AI 技能转型）。 在公共政策层面，必须依靠联邦与州政府的协同介入。Puri 提及联邦层面的 AI 行动计划（AI Action Plan），其重点正是构建覆盖全美的人才培养管网；此外，国家人工智能研究资源（NAIRR）的建立至关重要。NAIRR 致力于为全国缺乏预算的高校研究机构和小微企业免费提供急需的 AI 算力（GPU）和高质量数据集，从底层打破只有科技巨头才能负担高昂算力的资源垄断格局，各州政府也正在积极跟进并出台相关落地政策。 ```mermaid flowchart TD accTitle: 人才民主化与普惠生态建设路径 subgraph Pipeline["人才管道多元化"] A["社区大学与职教培训<br/>(如 Nvidia 加州合作项目)"] B["在线纳米学位与微证书<br/>(如 Coursera)"] C["在职员工技能重塑<br/>(如 Walmart 培训计划)"] end subgraph Policy["政策与基础设施普惠"] D["联邦 AI 行动计划<br/>(覆盖全美的教育管网)"] E["国家 AI 研究资源 NAIRR<br/>(提供普惠算力 GPU 与数据集)"] end Pipeline --> F["赋能传统行业与中西部偏远地区<br/>(医疗、制造、零售、金融)"] Policy --> F ``` ### [12:51](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=771s) 地方政策驱动与 K-12 早期 AI 启蒙教育 在推动 AI 人才普及的过程中，各州政府正发挥着至关重要的支柱作用。以威斯康星州为例，该州正积极打造区域性 AI 创新中心，并与微软（Microsoft）展开深度合作，计划在当地开展面向 10 万人的 AI 技能培训。这种由州政府牵头、科技巨头赋能的模式，展示了地方政策在重塑劳动力结构中的巨大潜力。与此同时，AI 人才培养必须将关口前移，尽早下沉到基础教育阶段。 通过像 ORISE AI Challenge 这类竞赛项目，K-12（基础教育）阶段的学生得以在早期接触并理解人工智能。演讲者特别提及被誉为“AI 教母”的斯坦福大学教授李飞飞（Fei-Fei Li）及其发起的 AI4ALL 项目。该项目致力于将前沿的 AI 启蒙教育全面引入 K-12 课堂，打破教育资源的壁垒，从少年时期便为背景多元的学生播下科技探索的种子。 - **威斯康星州与微软合作计划**：共同培训 100,000 名本地劳动者掌握 AI 关键技能，推动建设区域性 AI Hub。 - **AI4ALL 计划**：由李飞飞联合创立，旨在向 K-12 阶段及代表性不足的群体普及普惠性 AI 教育。 ```mermaid flowchart TD accTitle: AI人才民主化生态体系架构 subgraph S1["教育与早期普及"] A["K-12 启蒙 (如 AI4ALL / ORISE)"] --> B["社区大学与职业教育"] B --> C["地方高技能劳动力储备"] end subgraph S2["政企合作与基础设施"] D["超大规模云厂商 2000 亿美元数据中心投资"] --> E["州政府绑定本地人才培训协议"] E --> C end subgraph S3["跨行业落地与多元就业"] C --> F["传统与重工业创新应用<br>(制造/能源/医疗/零售)"] G["开源生态 (Open Source)"] --> F H["远程与分布式办公"] --> F end ``` --- ### [13:32](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=812s) CAIO Connect 与突破科技巨头光环的实体产业 AI 实践 为了打破“只有头部科技巨头才能做顶级 AI”的固有偏见，演讲者创办了非营利组织 CAIO Connect（首席 AI 官连接平台）。长期以来，公众和毕业生的目光过度聚焦在 FAANG 等少数三四家大型科技公司上，而忽视了制造业、零售业及工业互联网等广阔实体行业正在发生的惊人变革。CAIO Connect 的核心使命就是搭建舞台，邀请这些传统行业的首席 AI 官（CAIO）走到台前，向大众展示企业内部令人振奋的真实创新案例，吸引年轻人才加入。 会场中汇聚了众多跨界领军企业：工业控制先驱江森自控（Johnson Controls）正在利用 AI 算法精准调节大型设施的暖通与温控系统；电气与能源管理巨头施耐德电气（Schneider Electric）在电池与电力系统优化中深度落地 AI；工业品与办公用品分销巨头固安捷（Grainger）借助 AI 追踪并调度庞杂的供应链库存；此外还包括米其林轮胎（Michelin）、艾默生电气（Emerson Electric）等。演讲者强调，超大规模云厂商（Hyperscalers）未来将在全美数据中心建设上投入超过 2,000 亿美元，各州政府在引入这些关键基础设施投资时，必须设立硬性条件，要求科技巨头承诺为当地社区提供长期的 AI 技术培训，实现资本投资与本地人才生态的良性共赢。 - **参与实体产业 AI 创新的代表企业**：江森自控（Johnson Controls）、施耐德电气（Schneider Electric）、固安捷（Grainger）、米其林（Michelin）、艾默生电气（Emerson Electric）。 - **数据中心投资红利转化**：超大规模数据中心建设投资超 2,000 亿美元，各州应借机锁定本地劳动力技能提升承诺。 --- ### [15:59](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=959s) 民主化 AI 人才的行动纲领：弥合新时代的“AI 鸿沟” 在总结构建包容性 AI 未来的行动路径时，演讲者提出了一套多维度的实施方案。首先是大力投资社区大学（Community Colleges）并深化公私合作（PPP模式）；其次是坚定支持开源生态。开源是实现科技普惠的关键，正如演讲者所言：“如果说 AI 是新时代的电力，那么开源就是确保所有人都能点亮电灯的保障，而不是仅服务于少数特权阶层。”（"If people say AI is the new electricity, well open source makes sure the lights are on for everybody not just for a select few."）此外，行业必须重新拥抱远程与分布式办公模式，让生活在乡村和偏远地区、无法或不愿搬迁的技术人才同样能够参与到全国的 AI 建设中。 演讲者在尾声再次呼应了开场提到自己儿子的求职故事：试想一下，如果顶尖大学的毕业生不再千军万马挤进硅谷的三四家科技大厂，而是自豪地选择前往密歇根州的制造企业、德克萨斯州的能源公司、或俄亥俄州的医疗机构开发 AI 系统，整个国家将释放出怎样磅礴的发展动能。AI 人才不应被禁锢在沿海少数科技中心，而应该遍布在“每一个城市、每一间教室和每一个社区”。我们必须全力避免在传统数字鸿沟之外衍生出危害更大的“AI 鸿沟”（AI Divide），通过全社会的协同努力，让技术发展的红利真正造福所有人。 - **教育与开源普及**：倾斜资源扶持社区大学，将开源作为防止技术垄断的基石。 - **分布式人才布局**：利用远程办公激活偏远及乡村地区的潜在高技能人才。 - **多元产业赋能**：引导 AI 毕业生流向密歇根（智能制造）、德州（新能源）、俄亥俄（智慧医疗）等实体产业重镇。 - **社会公平使命**：防范“AI 鸿沟”加剧社会不平等，确保人人享有 AI 时代的机遇。 ## Notable quotes > [00:04](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=4s) AI 人才都去哪儿了 — Sanjay Puri > [00:04](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=4s) 如果全美顶尖的 AI 人才全部被吸纳进极少数科技大厂、集中在“几个邮政编码”（a few zip codes）之内，未来的技术生态将会变成什么样？ — Sanjay Puri > [03:32](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=212s) 如果你们来自硅谷，祝贺你们，我完全支持你们享用免费的红茶菌（kombucha）和睡眠舱（nap pods） — Sanjay Puri > [05:18](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=318s) 以包容而非排他来引领发展 — Sanjay Puri > [07:36](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=456s) 人才无处不在，但机会却并未平等眷顾每一个人。 — Sanjay Puri > [15:59](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=959s) 如果说 AI 是新时代的电力，那么开源就是确保所有人都能点亮电灯的保障，而不是仅服务于少数特权阶层。 — Sanjay Puri > [15:59](https://www.youtube.com/watch?v=mvIsxTTZSpk&t=959s) 每一个城市、每一间教室和每一个社区 — Sanjay Puri ## Mentioned Sanjay Puri, RegulatingAI, Ai4, Nvidia, Coursera, Walmart, NAIRR, Microsoft, ORISE AI Challenge, 李飞飞 (Fei-Fei Li), AI4ALL, CAIO Connect, 江森自控 (Johnson Controls), 施耐德电气 (Schneider Electric), 固安捷 (Grainger), 米其林 (Michelin), 艾默生电气 (Emerson Electric) ## Video description Sanjay Puri, Chairman/Founder, RegulatingAI Abstracts Understand how to make AI talent development more inclusive and accessible to people from all backgrounds. The discussion will explore how policy decisions, technological innovation, and a strong focus on equity can work together to ensure that AI education, careers, and opportunities are not limited to a privileged few. This discussion aims to bridge the gap between who has access to AI opportunities today and who should, ensuring a future where AI talent development is fair, inclusive, and widespread. [Source](https://www.youtube.com/watch?v=mvIsxTTZSpk)

Source quotes

[0:04] Sanjay Puri: AI 人才都去哪儿了 [0:04] Sanjay Puri: 如果全美顶尖的 AI 人才全部被吸纳进极少数科技大厂、集中在“几个邮政编码”（a few zip codes）之内，未来的技术生态将会变成什么样？ [3:32] Sanjay Puri: 如果你们来自硅谷，祝贺你们，我完全支持你们享用免费的红茶菌（kombucha）和睡眠舱（nap pods） [5:18] Sanjay Puri: 以包容而非排他来引领发展 [7:36] Sanjay Puri: 人才无处不在，但机会却并未平等眷顾每一个人。 [15:59] Sanjay Puri: 如果说 AI 是新时代的电力，那么开源就是确保所有人都能点亮电灯的保障，而不是仅服务于少数特权阶层。 [15:59] Sanjay Puri: 每一个城市、每一间教室和每一个社区

Source quotes

[0:04] Sanjay Puri: where is the AI talent this morning [0:04] Sanjay Puri: without batting an eyelid [3:00] Sanjay Puri: free kombuchas [3:00] Sanjay Puri: nap pods [5:18] Sanjay Puri: acqui-hires [12:51] Sanjay Puri: godmother of AI [15:59] Sanjay Puri: AI is the new electricity [15:59] Sanjay Puri: well open source makes sure the lights are on for everybody not just for a select few [17:19] Sanjay Puri: we want to work for a manufacturing company in Michigan or we want to work for a energy company in Texas or we want to work for a healthcare company in Ohio [17:19] Sanjay Puri: in every city, every classroom, and every community

Original transcript

[0:04] Thank you Jamie. I'm trying to figure [0:06] out where is the AI talent this morning [0:08] but anyway that's a whole separate [0:10] issue. I hope you guys had a good uh [0:13] conference yesterday [0:15] and u went through a lot of different [0:18] sessions. [0:20] U to me this topic has a lot of [0:23] resonance because [0:25] u we're going to talk about why there is [0:27] an urgency to make sure AI talent is [0:30] spread across regions, industries, [0:34] communities. [0:36] Um because democratizing AI talent is [0:39] about making sure innovation spreads [0:42] everywhere so that the United States can [0:45] lead responsibly in the global AI race. [0:49] I come to this um topic uh on a personal [0:54] basis. My son recently graduated with a [0:57] degree in AI. So I asked him basically [1:01] where was he and his friends going to be [1:05] looking for jobs and without batting an [1:08] eyelid he said [1:10] that same three four companies that are [1:12] the big u global giants in the tech [1:15] world and they're great companies some [1:16] of you might be working for them there's [1:18] nothing wrong with that but it really [1:21] for me that was a [1:23] moment that really struck me that if all [1:26] the AI talent ends up [1:30] in a few companies in a few zip codes. [1:33] Then what is it that we are creating? [1:37] What happens to the uh diversity of [1:40] thought? What happens to the diversity [1:42] of talent? [1:44] uh if AI is going to be the next [1:47] generation important [1:51] tool or technology that's going to [1:53] transform [1:55] how we treat diseases, how we educate [1:57] our children, how we move goods across [2:00] the board, then are we sure we want to [2:05] have this future being built in two zip [2:08] codes with a narrow set of people [2:10] focused on a narrow set of problems? So [2:13] that's where I thought it's something at [2:16] least important to talk about. This is a [2:18] conference of u AI people. My personal [2:21] background is in AI. By the way, my son [2:23] did take a job in one of those [2:25] companies, but that's a whole separate [2:27] issue. But [2:30] the question really for us is at least [2:33] as a country is making sure as we are [2:35] heading into this technology, we know [2:38] what we're doing. So I thought I'll [2:40] raise this issue to all of you who are [2:43] amazing practitioners, entrepreneurs, [2:46] policy leaders etc. This is part of uh [2:49] the AI policy summit. Uh it's uh to me [2:54] an important topic. So u I'm going to [2:57] basically lay out to you certain facts. [3:00] Some of them probably obvious to you but [3:03] from a personal standpoint it became a [3:05] little more obvious as I said listening [3:07] to my son. Then why does it really [3:09] matter? because [3:12] you know hey social media is uh being [3:15] concentrated uh cloud has been [3:17] concentrated why does it really matter [3:19] and then is there something that I you [3:23] us can do about it or should do about it [3:26] so I'll just walk you through a few [3:28] things um just to give you some [3:30] perspective [3:32] 40% of the top AI researchers in this [3:35] country are at four basic universities [3:40] and you would say hey there are great [3:41] universities and they are fantastic [3:43] universities but that also talks about [3:46] what's happening in terms of AI talent [3:49] in other places and that's a scary uh [3:53] thought from that perspective in terms [3:56] because talent is going to keep going [3:59] into those kind of uh educational [4:02] institutions [4:04] and then if you think about it 77% [4:07] of AI jobs are clustered istered in 10 [4:12] metro cities. There are 300 cities in [4:15] this country that have the balance of [4:17] those jobs. And again, you would say [4:20] again that's a tech technical problem. [4:22] But if you were here yesterday with some [4:25] of the sessions we had, AI is a very [4:28] different technology. It's a general [4:29] purpose technology and as I said, it's [4:32] going to impact everything that we do. [4:35] And if that is the case and then finally [4:39] think about the amount of money that is [4:42] getting invested in AI just in the first [4:44] half of the year and 50% of it was [4:48] basically spent in Silicon Valley [4:52] and this is not to take any shots at [4:55] Silicon Valley. If some of you are uh [4:57] from Silicon Valley, you know, power to [5:00] you. I support your free kombuchas, nap [5:04] parts, etc. I just want to make sure the [5:06] rest of the country also gets its share. [5:10] So that's just kind of a little bit of a [5:13] ground framing of the situation that we [5:16] are in. [5:18] So why is does it matter? And I kind of [5:22] lay out that it's basically it leads to [5:24] concentration of talent [5:26] from a competitiveness also it does [5:30] impact our competitiveness and then [5:32] there's an industry and a geographical [5:34] gap uh that happens because of this. So [5:38] if you really look and I don't know if [5:41] you're in the world of AI I'm sure [5:43] everybody kind of knows what's happening [5:46] in terms of salaries and packages that [5:49] are being handed out u to the top tier [5:52] of AI researchers and I'm all for uh [5:56] folks getting them. I hope my son one [5:58] day gets uh one of those packages too. [6:02] But it does have implications when you [6:05] give 14 billion to acquire aqua hire a [6:08] company or you pay hundreds of millions [6:10] of dollars. What happens to mid-tier [6:13] researchers across companies in [6:16] different regions and I think that's [6:18] really the hyper concentration and an [6:21] elite class of AI that is getting [6:24] generated. Now, if you're in that elite [6:25] class, you say, "Hey, what's the big [6:28] deal? You look at the NBA, there's [6:30] always best players in the NFL, we've [6:33] had the best players, etc." [6:36] But as as I said, this technology is [6:39] going to be working on curing diseases [6:41] for you, educating your children in many [6:45] many other aspects. So, we need to make [6:48] sure that what we are getting into. [6:52] So again my perspective is that as a [6:56] country we need to be leading by [6:58] inclusion not by exclusion. And why is [7:01] that important? If a lot of you are [7:05] building AI systems, [7:07] if there is not a diversity of thought [7:10] from a regional standpoint, from a [7:13] people standpoint, from an industry [7:15] standpoint, you're really not going to [7:18] have strong AI systems that can solve [7:21] the kinds of problems that we are trying [7:23] to solve. Whether it is in medicine, [7:25] whether it is in manufacturing, whether [7:27] it is in retail, etc. it is going to [7:31] have a very very narrow focused area. [7:36] So just think about what's happening [7:40] across industries and this is something [7:42] I work a lot on is you know people [7:46] always talk about the tech industry and [7:49] most of you I would presume a lot of you [7:52] work in the tech sector but there is [7:55] something beyond the tech sector. Look [7:57] at where the AI adoption rate is [8:00] happening in things like manufacturing, [8:03] happening in retail, healthcare, [8:06] financial services. Look at the GVAS of [8:09] some of these industries that are [8:11] projected by 2035. [8:14] But how do you think these industries [8:17] are going to achieve the things that [8:19] they need to achieve? [8:21] manufacturing with predictive [8:24] uh maintenance, supply chain, retail for [8:28] targeted marketing and then when you [8:31] look at healthcare diagnostics, delivery [8:35] of healthcare, [8:36] finance for fraud detection, algorithmic [8:40] buying etc. I mean those are important [8:43] things that happen but each of these [8:46] industries is suffering from a absolute [8:48] talent crunch because [8:51] if you are coming out of school or if [8:54] you are an AI uh top tier person you [8:58] probably are not ending up in some of [9:01] these industries. So if you see [9:04] what is happening is [9:07] these industries are really struggling [9:10] to get the talent that they need and [9:12] that is a very very important issue [9:15] because when these industries struggle [9:18] we're going to lack the innovation that [9:20] is needed for these industries from that [9:23] standpoint. So we have an industry gap [9:26] and then there is a geographical gap [9:27] that exists. Think about it. 77% of AI, [9:31] when I said AI in metro jobs, what's [9:34] happening in rural or small towns? What [9:37] happens to small companies in those [9:40] areas? What happens to, you know, health [9:43] care systems in rural areas? They need [9:46] to implement AI to transcribe what [9:48] physicians are doing from a diagnostic [9:51] standpoints, etc. But when you look at [9:54] this map, [9:56] you see all the greens. They're [9:57] generally heavily towards the coastal [10:01] areas and talent happens to be [10:04] everywhere but in many cases opportunity [10:07] is not for them. [10:10] So [10:12] what do you think we can do? I mean this [10:14] is some things that as I said I've been [10:16] thinking about is we really need to [10:18] rebuild the talent pipeline that we have [10:21] in this country and we need to have some [10:24] policy partnerships whether at the [10:26] federal and the state level and then we [10:29] have to highlight champions unsung [10:32] champions and then also we need to [10:34] engage hyperscalers who are investing in [10:36] data centers around our country. [10:41] So [10:43] in terms of building or rebuilding the [10:45] talent pipeline, we really need to focus [10:48] on different different entry points for [10:51] education whether it's community [10:53] colleges. [10:54] A lot of people in our country go into [10:57] two-year community colleges and we need [11:00] to make sure there is strong engagement. [11:02] Nvidia has set up a partnership in [11:05] California with community colleges to [11:07] train people in AI and then vocational [11:11] programs and online courses too so that [11:14] it can be spread. Now obviously we see [11:17] Corsera and some of these other things [11:19] that are doing nano degrees and uh [11:22] things of that nature but also we need [11:26] to make sure that we are upskilling the [11:28] workforce. It's great to get talent in [11:31] but what about the people who are [11:33] existing in those companies in that part [11:35] of the uh country. Walmart is doing a [11:39] reskilling of its employees and some of [11:41] these other employers have taken it on [11:44] to reskill the employees because that is [11:47] going to be an important aspect of what [11:49] we do. So that rebuilding our talent [11:53] pipeline is very very critical [11:56] and then what can we do at a policy [11:58] level and this is where federal and [12:00] state both come into the play. The [12:03] federal government with the uh the Trump [12:05] administration recently announced the AI [12:07] action plan. We're going to have a [12:10] session uh later this afternoon. If you [12:12] have the time come we have some great [12:14] speakers on that. But a big part of that [12:17] is in terms of education and creating a [12:20] talent pipeline across the country. And [12:24] the other thing is uh NAR and what is [12:26] NAR? It's a national AI resource [12:29] research facility which gives AI compute [12:33] as well as data to academic institutions [12:37] and small companies around the country [12:39] because everybody can't afford a GPU or [12:43] people don't have access to data sets. [12:45] So this gives you a free access to that. [12:48] States have gotten into the act too. A [12:51] lot of states now are working. Wisconsin [12:53] is working towards how can we build an [12:56] AI hub. They work with Microsoft to [12:59] train 100,000 people uh in Wisconsin [13:03] towards that. So I think states have a [13:05] big big role towards that. And then [13:08] finally there is the orise AI challenge [13:12] which is helping even at K12. [13:14] We can start early on at K12. So when [13:19] you look at what uh the godmother of AI [13:22] Fifi Lee she has AI for all program [13:25] which really gets into K12 education [13:28] early on across the board. [13:32] What this is something that we are [13:34] doing. I run a nonprofit called Kaio uh [13:38] connect which means the chief AI [13:39] officers connect and our focus really [13:42] why do we do this is there are [13:45] organizations besides the three four big [13:47] tech companies or fang or whatever the [13:50] acronym is right now that are doing [13:52] incredible work whether it is in [13:55] manufacturing whether it is in retail [13:58] whether it's in industrial uh you know [14:02] platforms [14:03] we basically bring them over and tell [14:06] them tell our audience and tell the [14:09] people what innovative stuff are you [14:11] guys doing within your companies and [14:14] what why should people come in and work [14:17] for you. So when you look at some of the [14:19] things that companies these are just a [14:21] few but some of them are here by the way [14:23] you know Johnson Controls Industrial [14:25] Control Company Schneider Electric huge [14:29] again in terms of uh batteries and [14:32] electric stuff Granger office supplies [14:34] their chief AI officer is here Michelin [14:37] tires [14:38] Emerson Electric and many many others [14:42] and why do we do that because I don't [14:45] want the next kids coming out of school [14:48] to say, "Hey, these are the only [14:50] companies that are doing cool AI [14:52] things." Because when you listen to [14:53] these folks, they are doing they're [14:56] implementing AI for incredible stuff for [15:02] temperature control in this uh facility. [15:05] Johnson Controls and companies like that [15:07] are using AI to do that stuff. And then [15:10] again, Granger from Office Supply to [15:13] track all the different kinds of things [15:15] that they have within them. So there's [15:17] some amazing creative things that are [15:19] happening and we need to highlight and [15:21] showcase uh you know these kinds of [15:24] companies [15:26] and then again 200 billion is going to [15:28] be spent by these hyperscalers around [15:31] the country in data centers. They need [15:34] these data centers because AI data [15:36] centers are critical. some of the [15:39] companies and some of the states states [15:41] need to say hey if you want to invest [15:43] and they need these investments you need [15:46] to commit to training locally out here [15:50] in AI and AI technologies and I think [15:53] that's very very important for states to [15:56] be doing [15:57] so [15:59] what can we do together [16:02] well I told you what we are doing I [16:06] would say investing education Community [16:08] colleges to me are very very important. [16:12] Supporting public private partnerships I [16:14] think are critical. [16:17] The other thing I think is supporting [16:19] open source. Open source is really [16:22] critical. If people say hey AI is the [16:25] new electricity well open source make [16:29] sure the lights are on for everybody not [16:31] just for a select few. So I would say [16:34] support open-source because it gives [16:37] access to everybody and then [16:39] spotlighting uh companies that are not [16:43] part of this genre. There are so many [16:45] other companies that are doing some [16:47] creative things highlight them and then [16:51] promote remote and distributed work. [16:53] kind of right now it's kind of not in [16:55] vogue after co we went remote then now [16:58] we are back in there but there are [17:00] people in rural areas in other parts of [17:03] the country that have the talents that [17:05] don't want to move can't move and we [17:07] don't tap into those skills AI work can [17:10] be done everywhere so I would say is you [17:13] know let's give that an opportunity too [17:17] so finally I would say is I'll come back [17:19] to my son's story is imagine when he was [17:23] graduating out. He and his friends said [17:24] that, you know, instead of working for [17:27] these three four great companies, we [17:30] want to work for a manufacturing company [17:32] in Michigan or we want to work for a [17:35] energy company in Texas or we want to [17:37] work for a healthcare company in Ohio. [17:40] Think about the benefits that we as a [17:43] nation would get because AI talent [17:46] really should be everywhere in every [17:49] city, every classroom, and every [17:51] community. I think if that happens only [17:55] if that happens we as a country are [17:57] going to really benefit and as a [18:00] community because this has a social [18:04] impact too what we don't want to do is [18:07] create an AI divide just like we call it [18:10] a digital divide this has even a bigger [18:13] consequences in doing that so I hope um [18:17] you folks will do your part we are doing [18:20] our part in there so thank you so much [18:22] for coming and listening to [18:25] democratizing AI talent and all of you [18:27] have great talent and I hope that uh we [18:30] get a chance to engage with you. Thank [18:32] you so much. [18:34] [Applause]