StoneAI, Good Trouble Push IP Tracking to Shield Creatives from AI
Ai4
Frontier AI risks marginalizing 90% of working artists unless enterprises replace compliance-driven diversity with verifiable asset provenance and direct community governance.
## Background
Published on YouTube by the Ai4 channel on August 28, 2025, this video captures a fireside session featuring David Leighton (Co-Founder & CEO of WITI and StoneAI) and AJ Thomas (Founder & General Partner at Good Trouble Ventures). The presentation explores a practical playbook for human–AI collaboration, showing how inclusive design, cross-functional teams, and bias-checking transform AI into a sustainable driver of bottom-line growth.
## TL;DR
David Leighton and AJ Thomas discuss operationalizing inclusion across the entire AI lifecycle, arguing that true innovation stems from diverse cognitive perspectives rather than compliance-driven quotas. Drawing from experiences at Alphabet's Moonshot Factory and WITI, they highlight ventures like StoneAI and Good Trouble Ventures that protect intellectual property, provenance, and economic upside for working creators. Emphasizing the future of work as 'the how of now,' the speakers urge builders to involve diverse end-user communities early in product design and rigorously scrutinize data privacy policies in free AI tools.
## Atmosphere & style
The session feels like an energetic, fast-paced kitchen-table conversation between longtime friends and seasoned tech insiders who effortlessly balance high-level enterprise R&D with grassroots impact. The dynamic between AJ and David is warm, witty, and grounded, trading playful banter and self-deprecating humor while challenging the audience with pointed diagnostic questions. It carries an urgent, pragmatic tone that cuts through Silicon Valley AI hype to focus on creator protection, data provenance, and the lived experiences of everyday communities.
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=sApcIXpje1E&t=4s) True technological moonshots must directly engage and test with diverse everyday human populations rather than remaining confined to isolated research labs.
- [03:01](https://www.youtube.com/watch?v=sApcIXpje1E&t=181s) Compliance-driven diversity initiatives fail to create lasting value; breakthrough innovation requires diverse cognitive perspectives that challenge assumptions decades in advance.
- [03:01](https://www.youtube.com/watch?v=sApcIXpje1E&t=181s) WITI modernized its mission and acronym to Workforce Innovation, Trust, and Influence to align community initiatives with the AI era.
- [05:30](https://www.youtube.com/watch?v=sApcIXpje1E&t=330s) StoneAI combines AI storytelling tools with Cyberbear asset tagging to secure provenance and protect intellectual property for filmmakers and creators.
- [08:13](https://www.youtube.com/watch?v=sApcIXpje1E&t=493s) Operationalizing inclusion requires product teams to integrate lived experiences, cultural origin stories, and representative decision-makers into model training and governance.
- [08:13](https://www.youtube.com/watch?v=sApcIXpje1E&t=493s) Free AI services inherently commodify user inputs to train future models, underscoring the necessity of strict data privacy and access governance.
- [12:06](https://www.youtube.com/watch?v=sApcIXpje1E&t=726s) The future of work is the 'how of now'—the immediate daily workflows, ethical standards, and transparency practices adopted by organizations today.
- [14:36](https://www.youtube.com/watch?v=sApcIXpje1E&t=876s) While top-tier studios and celebrity creators access direct AI partnerships, 90% of working creatives face displacement and need tools that protect ownership and distribute economic upside.
- [17:07](https://www.youtube.com/watch?v=sApcIXpje1E&t=1027s) Bringing non-technical end users, artists, and underrepresented communities into the early design phase prevents market blind spots and broadens product utility.
- [17:35](https://www.youtube.com/watch?v=sApcIXpje1E&t=1055s) Enterprises and creators must scrutinize terms of service on emerging generative AI tools before inputting sensitive or proprietary data.
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=sApcIXpje1E&t=4s) Origin Stories: From Moonshots to Grassroots Connections
The session opens with lighthearted banter between David Leighton and AJ Thomas as they settle in for their nineteen-minute fireside chat, joking about how their private phone calls are always engaging enough that they belong on a public stage. AJ reflects on how their partnership began with a serendipitous personal choice: she snuck her young daughter into WITI’s (Women in Technology International) annual conference. That single event sparked a lasting relationship, leading to AJ serving on WITI’s board of directors for several years. David reflects on his own journey spanning back to the early 1980s PC software era and the dot-com web boom, noting how every major technology transition has required an ongoing fight for broader human inclusion.
AJ connects these technological waves to her prior leadership running people products and the global talent organization at Alphabet’s Moonshot Factory (X). During that period, deep learning and early generative AI capabilities were being actively incubated. To bridge cutting-edge research and societal impact, Alphabet created a dedicated organizational role titled "in touch with the real world." In this workflow, ambitious concepts progressed from rapid evaluation directly into hands-on testing with everyday human populations. As AJ emphasizes, a project could not qualify as a true moonshot unless it engaged with and served the diverse billions of people whose lives it intended to change. David contrasts this global tech bubble with his daily life in San Luis Obispo, California, observing that outside concentrated AI think tanks and tech hubs, many mainstream communities have yet to adopt or feel the benefits of these rapid advancements.
```mermaid
flowchart LR
accTitle: Alphabet Moonshot Factory Evaluation Pipeline
subgraph Lab["Incubation Phase"]
A["Emerging AI Technology"] --> B["Rapid Evaluation"]
end
subgraph RealWorld["Application Phase"]
B --> C["'In Touch with the Real World' Testing"]
C --> D["Global Moonshot Deployment\n(Billions of Users)"]
end
```
[03:01](https://www.youtube.com/watch?v=sApcIXpje1E&t=181s) WITI’s Evolution: Innovation Driven by Perspective, Not Compliance
Discussing the staying power of WITI since its founding in 1989, David explains that the organization originally emerged from partnerships with HP Labs to address the stark absence of women and underrepresented talent in computing. While the early mandate centered around raw representation numbers, the mission expanded across the dot-com boom and subsequent industry corrections. Three years ago, WITI merged and redefined its acronym to stand for **Workforce Innovation, Trust, and Influence**, deliberately aligning its community with the modern AI era.
David underscores a foundational philosophy: compliance-driven diversity programs inevitably fail to produce sustainable impact. Genuine breakthrough innovation occurs only when diverse cognitive perspectives challenge assumptions and anticipate trends fifteen to twenty years ahead—a discipline learned directly from long-term enterprise R&D environments. AJ points out that David’s leadership has built an expansive enterprise network, engaging top Chief Information Officers (CIOs) and leaders across tech titans including Apple, Microsoft, and Salesforce.
- **WITI Founding Year**: Established in 1989 out of collaborative initiatives with HP Labs.
- **Acronym Realignment**: Reframed as Workforce Innovation, Trust, and Influence to emphasize AI-era collaboration.
- **Enterprise Network**: Historic engagement with technical leadership across Apple, Microsoft, Salesforce, and global Fortune 500 enterprises.
[05:30](https://www.youtube.com/watch?v=sApcIXpje1E&t=330s) StoneAI: Empowering Creators with Provenance and IP Protection
Turning to David’s current venture as CEO of StoneAI, AJ asks him to break down the core mission of the company. David defines StoneAI as "storytelling with provenance," quickly making a self-deprecating joke about mistaking "provenance" for Providence, Rhode Island when first hearing the term. In practice, provenance establishes clear verifiable origins, ownership records, and intellectual property protection for creative assets. StoneAI works extensively with film studios, television networks, and legal verticals to protect human creators while deploying collaborative AI tooling.
David explains that the venture coalesced when veteran cinematographer and filmmaker Evan B. Stone began attending David’s weekly AI think tanks shortly after ChatGPT launched. Seeking to help production companies adopt AI ethically without stripping creators of their livelihoods, Evan partnered with David to combine creative production workflows with Cyberbear, a specialized asset-tagging technology. Rather than displacing human ingenuity in an already polarized entertainment industry, the platform aims to provide artists with "superpowers" to scale their creative expressions safely.
```mermaid
flowchart TD
accTitle: StoneAI Creator Protection and Asset Provenance Flow
subgraph Ingestion["Creative Production"]
A["Human Filmmakers & Creators"] --> B["AI-Assisted Storytelling Tools"]
end
subgraph Verification["Integrity Layer"]
B --> C["Cyberbear Asset Tagging"]
C --> D["IP & Provenance Verification Engine"]
end
subgraph Distribution["Protected Enterprise Delivery"]
D --> E["Studios, Networks & Legal Verticals"]
end
```
[08:13](https://www.youtube.com/watch?v=sApcIXpje1E&t=493s) Operationalizing Inclusion: An Audience Audit on Conscious Tech
AJ and David turn the conversation outward, running a live interactive poll with the audience to measure whether inclusion in AI is being operationalized or treated merely as theoretical corporate rhetoric. AJ prompts the room with several pointed diagnostic questions regarding their product design and governance models:
- **Lived Experiences**: How many builders design products actively shaped by the lived experiences of the communities most impacted?
- **Origin Stories in Data**: How many teams ensure that regional, cultural, and community origin stories directly inform AI training data?
- **Representation at the Table**: How many product, investment, and feature decisions are made by individuals who directly represent the target user base?
AJ cautions the audience about the operational pitfalls of overlooking access principles, citing recent vulnerabilities where OpenAI collaborative chat links became unintentionally discoverable on the public web. David adds an important reminder for practitioners: *"Nothing is free. If it says free, that means you're just giving them data to train new models."* The poll reveals strong representation of enterprise builders, capital allocators, and customer-facing operators in the room, while highlighting that formal internal governance structures—such as dedicated AI Centers of Excellence (CoEs)—are just beginning to emerge across modern organizations.
[12:06](https://www.youtube.com/watch?v=sApcIXpje1E&t=726s) The Future of Work as the "How of Now"
Transitioning to workplace operations, AJ challenges the conventional phrasing around the "future of work," arguing instead that the future of work is simply "the how of now." How teams construct their daily workflows, ethical boundaries, and toolsets today directly dictates the structural future of the workplace. She highlights Zapier and its Chief People Officer, Brandon Sammut, as a prime model of operational transparency for open-sourcing Zapier's entire internal AI tech stack across the full employee lifecycle from "hire to retire."
David shares that WITI is exploring using Google's NotebookLM to synthesize and open-source decades of historical WITI summit data directly to community builders. AJ references her work alongside fellow attendees on the JFFLabs (Jobs for the Future) Futurist Council, remarking humorously that while self-described "futurists" cannot predict every anomaly, their core responsibility is to identify early structural signals, interrogate default assumptions, and enforce first-principle design thinking when integrating AI into human systems.
[14:36](https://www.youtube.com/watch?v=sApcIXpje1E&t=876s) Good Trouble Ventures: Bridging the 90% Creator Divide
David highlights AJ’s ongoing impact and asks her to elaborate on her investment work, noting grassroots projects like that of Bruce Bryan—a man who spent 29 years in New York state prisons after being wrongfully convicted, received executive clemency, and is now leading programs to upskill at-risk youth and children of incarcerated parents in practical AI literacy. AJ explains that genuine inclusion requires building foundational infrastructure for access rather than leaning into corporate buzzwords around diversity.
Through Good Trouble Ventures, AJ focuses on backing companies at the crossroads of technology and human creativity to ensure artists can sustain viable long-term careers. She illustrates a stark disparity in the creative industries: while the top 10% of high-profile entertainers and legacy studios form direct partnerships with frontier AI incumbents, the remaining 90% of working creatives are systematically marginalized, lacking both access and IP protections as their crafts face automation. Good Trouble Ventures deliberately invests in platforms that democratize creative tooling, protect ownership rights, and distribute economic upside to this underrepresented majority.
[17:07](https://www.youtube.com/watch?v=sApcIXpje1E&t=1027s) Inclusive Product Design and Involving End-User Communities Early
The discussion turns toward practical methods for embedding inclusion into everyday organizational workflows, particularly when designing and developing new AI-driven tools. AJ encourages teams to look across their current organizations and ask what simple, actionable steps they can take to incorporate underrepresented viewpoints early in the development cycle. Specifically, teams should consider involving creators, artists, and diverse end-user communities who will ultimately interact with the technology. Even if these groups do not traditionally have hands-on access to the product during its initial technical architecture phase, deliberately integrating their feedback from the very beginning ensures that the final product serves a much broader audience rather than a narrow demographic.
- Product teams should identify simple mechanisms within their existing workflows to bring creators, artists, and non-technical end users into the ideation and design phases.
- Early community engagement helps prevent blind spots before products reach the general market.
---
[17:35](https://www.youtube.com/watch?v=sApcIXpje1E&t=1055s) Reading the Fine Print on AI Tools and Closing Thoughts
As the session nears its conclusion under a one-minute warning, David brings up a timely cautionary tale regarding emerging generative tools, pointing specifically to infographic generator Napkin AI (`napkin.ai`). While praising the platform as remarkable and acknowledging that users will quickly get hooked, he urges the audience to carefully read the fine print and terms of service on free AI applications. Many free tools require users to grant permission for their inputted data and proprietary information to be utilized by the provider. Although creators generally retain ownership of their creations, David warns against feeding sensitive, confidential details into external services without understanding data usage policies. He encourages the audience to dive deep into these questions over the coming days and connect with communities like WITI and Good Trouble Labs as technology shifts daily.
```mermaid
flowchart TD
accTitle: AI Tool Data Privacy and Inclusion Considerations
accDescription: Flowchart illustrating key steps when adopting new AI tools and designing inclusive products.
subgraph ToolAdoption["AI Tool Adoption & Privacy"]
A["Explore New AI Platforms<br>(e.g., Napkin AI)"] --> B["Review Terms of Service & Fine Print"]
B --> C["Safeguard Proprietary & Sensitive Data"]
end
subgraph InclusiveGovernance["Inclusive Strategy & Community"]
D["Engage Communities<br>(WITI, Good Trouble Labs)"] --> E["Address the 'How of Now'"]
E --> F["Global & Ethical Deployment"]
end
C -.-> D
```
AJ reinforces the importance of scrutinizing service agreements, noting that anyone in an organization who contributes to or oversees terms and governance must "ask the hard question." She stresses that shaping the future of technology is fundamentally about "the how of now"—how builders show up responsibly and ethically today while creating products for a global audience. David agrees, underscoring the necessity of designing for a global community with compliance and privacy standards in mind, before the session wraps up to a round of applause.
- **Data Privacy Awareness**: Free AI tools often stipulate broad data usage permissions in their fine print; users must exercise caution with sensitive corporate or personal data.
- **Community Engagement**: Continuous learning and ethical governance are supported through networks like WITI and Good Trouble Labs.
- **Accountability in Governance**: Organizational leaders writing terms and policies must actively interrogate how data is handled and how inclusive their design practices remain.
## Notable quotes
> [00:04](https://www.youtube.com/watch?v=sApcIXpje1E&t=4s) in touch with the real world — AJ Thomas
> [03:01](https://www.youtube.com/watch?v=sApcIXpje1E&t=181s) Workforce Innovation, Trust, and Influence — David Leighton
> [05:30](https://www.youtube.com/watch?v=sApcIXpje1E&t=330s) storytelling with provenance — David Leighton
> [05:30](https://www.youtube.com/watch?v=sApcIXpje1E&t=330s) superpowers — David Leighton
> [08:13](https://www.youtube.com/watch?v=sApcIXpje1E&t=493s) Nothing is free. If it says free, that means you're just giving them data to train new models. — David Leighton
> [12:06](https://www.youtube.com/watch?v=sApcIXpje1E&t=726s) the how of now — AJ Thomas
> [12:06](https://www.youtube.com/watch?v=sApcIXpje1E&t=726s) hire to retire — AJ Thomas
> [17:35](https://www.youtube.com/watch?v=sApcIXpje1E&t=1055s) ask the hard question — AJ Thomas
## Mentioned
David Leighton, AJ Thomas, WITI, StoneAI, Good Trouble Ventures, Alphabet's Moonshot Factory (X), HP Labs, Apple, Microsoft, Salesforce, Evan B. Stone, Cyberbear, OpenAI, ChatGPT, Zapier, Brandon Sammut, Google NotebookLM, JFFLabs, Bruce Bryan, Napkin AI, Good Trouble Labs
## Video description
David Leighton, Co-Founder & CEO, WITI, StoneAI
AJ Thomas, Founder & General Partner, Good Trouble Ventures
Abstracts
Peel back the hype and discover how inclusive design turns AI from a buzzword into a bottom‑line growth engine. In this rapid‑fire conversation, AJ Thomas (Founding Partner, Good Trouble VC) joins David Leighton (Co‑Founder & CEO, WITI) to explore the new playbook for human–AI teamwork:
* Pinpointing real‑world pain points where diverse perspectives unlock breakthrough solutions
* Embedding ethics and bias‑checks without killing velocity
* Building cross‑functional teams—operators, creatives, and data scientists—who mirror the users they serve
* Winning investor, customer, and talent confidence by proving that purpose and profit rise together
Attendees leave with a five‑step “inclusion acceleration checklist” they can start using the moment they leave the room.
StoneAI携手Good Trouble推进知识产权追踪，保护创作者免受AI冲击
前沿AI恐导致90%的从业艺术家被边缘化，除非企业以可验证的资产溯源和直接的社区治理取代合规驱动的多元化举措。
David Leighton 与 AJ Thomas 探讨了在整个 AI 生命周期中落实包容性的方法，认为真正的创新源于多元的认知视角，而非合规驱动的配额指标。结合在 Alphabet 登月工厂（Moonshot Factory）和 WITI 的从业经验，他们重点介绍了 StoneAI 和 Good Trouble Ventures 等旨在保护从业创作者知识产权、资产溯源和经济利益的机构。两位发言人强调未来的工作是“当下的践行方式（the how of now）”，敦促开发者在产品设计早期引入多元化的最终用户群体，并严格审视免费 AI 工具中的数据隐私政策。
Alphabet Moonshot Factory（X）前团队成员 AJ Thomas 指出，真正的“登月级”技术项目必须走出隔离的实验室，直接面向多元化的真实大众群体进行实地测试与验证。 (0:04)
WITI 将其愿景缩写重塑为“劳动力创新、信任与影响力”（Workforce Innovation, Trust, and Influence），强调仅靠合规驱动的多元化举措难以产生持久价值，真正的突破性创新需要提前数十年挑战既定假设的多元认知视角。 (3:01)
StoneAI 将 AI 叙事工具与 Cyberbear 资产标记技术相结合，为电影制作人和创作者构建可追溯的出处验证（provenance）与知识产权保护体系，确保人类创意不被技术浪潮边缘化。 (5:30)
Good Trouble Ventures 重点支持面临技术替代风险的 90% 普通从业创作者，致力于填补顶层 10% 头部工作室与大众创作者之间的鸿沟，提供保障所有权并分享经济收益的普惠工具。 (14:36)
针对 Napkin AI 等免费生成式工具，对话提醒“天下没有免费的午餐”，免费服务往往以使用用户数据训练新模型为前提，企业与创作者必须严格审查条款并建立负责任的日常数据治理规范。 (17:35)
Reading summary
David Leighton and AJ Thomas argue that sustainable artificial intelligence deployment requires operationalizing inclusion, safeguarding intellectual property, and engaging end users early rather than relying on compliance-driven diversity models.
Moving from Compliance to Direct Inclusion (0:04)
Leighton, drawing on the history of WITI since its 1989 founding with HP Labs, maintains that compliance-driven diversity programs fail to generate lasting impact. True technological innovation requires diverse cognitive perspectives capable of projecting 15 to 20 years forward. Reflecting this shift, WITI rebranded its acronym to Workforce Innovation, Trust, and Influence to align with modern AI collaboration across enterprise networks including Apple, Microsoft, and Salesforce.
Thomas connects this approach to her former role leading people products at Alphabet's Moonshot Factory (X), where projects integrated dedicated personnel tasked with staying 'in touch with the real world.' In that workflow, technical concepts were evaluated against the lived experiences of everyday populations, under the operating principle that an initiative does not qualify as a moonshot unless it directly engages the diverse billions of people it aims to impact.
Protecting Creator Provenance and Economic Viability (5:30)
In creative industries, Thomas observes an economic disparity: while the top 10% of high-profile entertainers secure direct partnerships with frontier AI incumbents, the remaining 90% of working creatives risk displacement and lack intellectual property protections. Through Good Trouble Ventures, Thomas invests in platforms designed to democratize creative tooling, secure ownership rights, and distribute economic returns to this broader majority, including grassroots programs such as Bruce Bryan's work teaching practical AI literacy to at-risk youth.
Addressing this vulnerability in entertainment and legal verticals, Leighton's venture, StoneAI—co-founded with cinematographer Evan B. Stone—combines storytelling workflows with Cyberbear asset-tagging technology. The platform establishes provenance through verifiable ownership and IP records, aiming to provide human creators with collaborative tools that protect their livelihoods rather than replacing them.
Designing the 'How of Now' and Product Governance (12:06)
Thomas reframes the 'future of work' as the 'how of now,' arguing that near-term operational choices, ethical boundaries, and tool adoption dictate long-term workplace structures. She points to Zapier as an example of operational transparency for open-sourcing its internal AI stack across the full employee lifecycle, while Leighton notes WITI is exploring Google's NotebookLM to synthesize decades of summit archives for public use.
Both speakers stress that inclusive product development must begin during the earliest technical design phases by bringing non-technical end users, creators, and affected communities into ideation. Diagnostic audience polling revealed that while formal structures like AI Centers of Excellence are beginning to emerge, technical teams frequently overlook whether target communities' origin stories and lived experiences are reflected in data governance and decision-making.
Scrutinizing Terms of Service and Data Costs (17:35)
Leighton issues a warning regarding ostensibly free generative applications, such as the infographic platform Napkin AI, noting that zero monetary cost typically means users are trading proprietary data to train third-party models. He cautions creators and enterprises against uploading sensitive or confidential information without reviewing terms of service and licensing agreements.
Thomas reinforces that governance leaders and developers must actively interrogate service terms and security defaults. She cites operational oversights, such as OpenAI collaborative chat links inadvertently surfacing on the public web, as evidence that access and privacy controls require continuous technical vigilance.
The discussion leaves open the question of how quickly enterprise AI Centers of Excellence and smaller creators can institute formal provenance and governance standards as third-party model providers continuously adjust their terms of service.
精炼阅读摘要
StoneAI 首席执行官 David Leighton 与 Good Trouble Ventures 的 AJ Thomas 提出，人工智能时代的技术包容性不能仅停留在企业合规层面，必须通过早期社区参与、确权保障以及审慎的数据治理转化为具体的日常开发实践。
超越合规指标：从登月项目到产业视角创新 (0:04)
Thomas 回顾了她在 Alphabet 登月工厂（X）负责人才组织时的实践，当时团队设立了“与现实世界接触”的专属岗位，规定任何雄心勃勃的 AI 概念在快速评估后必须直接面对普通人群测试；若无法服务并触及多样化的数十亿大众，便不能被称为真正的登月项目。Leighton 则以加州圣路易斯奥比斯波等非科技核心区为例指出，主流社区至今尚未充分感受到 AI 进步的红利。
针对长期推动多元化的机构 WITI（1989年与惠普实验室合作成立），Leighton 解释称该机构三年前将名称重新定义为“劳动力创新、信任与影响力”，以适应现代 AI 变革。他强调基于合规指标的多元化项目往往无法带来持久影响，真正的突破性创新源于认知层面的多样性，这种视角能够提前十五至二十年挑战既有假设并预判技术演进趋势。
内容确权与保护：防范创作者面临的技术脱钩 (5:30)
在创意产业应用方面，Leighton 将其创立的 StoneAI 定义为“带出处追溯的故事讲述”。该平台起源于资深电影制作人 Evan B. Stone 在 ChatGPT 发布后寻求保护创作者生计的诉求，通过将创意制作流程与资产标记技术 Cyberbear 相结合，为影视工作室、电视网络及法律行业提供明确的数字资产所有权记录与知识产权保护，旨在赋予创作者辅助能力而非取代人类独创性。
Thomas 指出创意生态中存在严重的结构性分化：处于顶端的 10% 知名艺人和传统制片厂可以直接与前沿 AI 巨头建立商业合作，而其余 90% 的普通创作者在面临工种自动化的同时，普遍缺乏技术获取途径与版权防护。Good Trouble Ventures 正聚焦于此类交汇点，投资旨在将工具使用权、版权控制和经济收益下沉给大多数劳动者的底层基础设施，例如支持曾被冤狱羁押 29 年、获特赦后致力于青少年 AI 素养培训的 Bruce Bryan 项目。
落地的包容性设计：反思数据来源与产品决策权 (8:13)
对谈中展示的现场调研显示，许多团队对 AI 包容性治理的认知仍停留在理论层面，而专属的 AI 卓越中心（CoE）等组织架构才刚刚开始建立。Thomas 提出诊断性准则，呼吁构建者检视三项核心问题：产品设计是否融入了受影响群体的真实生活经历、训练数据是否包含地域与文化渊源故事，以及产品决策席位上是否有目标用户群体的直接代表。
针对工作场所的构建，Thomas 提出“工作的未来即当下的实践”（the how of now），即日常工作流、伦理边界与工具选型的确立方式直接决定了未来的组织形态，并列举了 Zapier 首席人才官 Brandon Sammut 推动将其全员工生命周期的内部 AI 技术栈予以开源的透明化实践。团队应当在技术架构搭建的初期就将艺术家及非技术终端用户纳入设计流程，及早消除认知盲点。
免费工具背后的隐私风险与服务条款警示 (17:35)
围绕实际部署与使用风险，Thomas 提及近期 OpenAI 协作对话链接在公共网络上被意外检索的案例，警示忽视访问控制原则带来的操作隐患。Leighton 则特别以信息图表生成工具 Napkin AI 为例，提醒使用者注意“免费 AI 服务”协议中的细则条款。他明确指出，所谓免费往往意味着用户需授权平台使用所输入的数据来训练后续模型。
两位对话者警告，虽然部分生成平台保留了创作者对其作品的名义所有权，但从业者绝不可在未厘清数据使用政策的情况下，将敏感、专有或机密信息直接上传至外部公共工具中。企业内部参与制定条款和监管政策的人员必须主动质询底层数据流向，平衡产品敏捷采纳与全球隐私及合规标准。
当前生成式 AI 的商业化落地正在加剧技术所有者与一线创作者之间的资源鸿沟。尽管业界开始探索资产打标与开源治理等应对机制，但在企业实际落地中，如何在保持敏捷创新的同时防范数据资产泄露并真正吸收弱势群体的反馈，仍是留给技术构建者与监管人员的长期挑战。
Detailed notes — zh
## Background
该内容由 Ai4 于 2025 年 8 月 28 日发布在 YouTube 平台上。页面简介指出，本次对谈由 Good Trouble Ventures 创始人 AJ Thomas 与 WITI / StoneAI 首席执行官 David Leighton 共同呈现，探讨如何拨开技术炒作迷雾，通过包容性设计、人机协作、伦理纠偏与多元团队搭建，将 AI 转化为兼顾商业价值与社会效益的增长引擎。
## TL;DR
对话围绕包容性与伦理设计在 AI 时代的产品落地与投资逻辑展开。David Leighton 结合 WITI 的演进与 StoneAI 在内容溯源（Provenance）方面的探索，强调多元视角与创作者知识产权保护是推动长期创新的核心底座；AJ Thomas 则结合其在 Alphabet 登月工厂与早期投资的经验，主张打造“有意识的科技”，让边缘化社群与长尾创作者能够真正参与 AI 决策并获得技术准入。两人同时警示了免费 AI 工具隐藏的数据隐私与微小条款风险，呼吁行业以“第一性原理”将责任与包容融入当下的每一项开发与治理行动中。
## Atmosphere & style
整场对谈节奏明快且极具互动感，两位主讲人凭借多年的深厚交情在台上谈笑风生、默契十足。从调侃台下握有预算的决策者，到分享冤狱获释者通过 AI 技能重塑新生的真实故事，现场情感在轻松幽默与庄重深刻之间自如切换。充满启发的观众互动与犀利务实的行业警示，让整场活动散发着既具前瞻思辨又脚踏实地的感染力。
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=sApcIXpje1E&t=4s) 前沿技术不仅要在实验室快速评估，更需通过“接触真实世界”（in touch with the real world）机制在多元人群中检验与落地。
- [03:01](https://www.youtube.com/watch?v=sApcIXpje1E&t=181s) 合规驱动的多元化往往流于形式，多元化视角的真正价值在于直接催生突破性创新，在快速迭代的 AI 时代尤为关键。
- [03:01](https://www.youtube.com/watch?v=sApcIXpje1E&t=181s) WITI 经过重组将其使命重塑为“工作场所创新、信任与影响力”（Workforce Innovation, Trust and Influence），联合科技巨头高管推动企业 AI 转型。
- [05:30](https://www.youtube.com/watch?v=sApcIXpje1E&t=330s) StoneAI 提出“带有出处证明的叙事”（storytelling with provenance），通过透明追踪所有权和 IP 起源保护创作者权益，而非替代创作者。
- [08:13](https://www.youtube.com/watch?v=sApcIXpje1E&t=493s) 包容性不能停留在口号，衡量标准在于 AI 研发是否真正融入受影响最深边缘群体的亲身体验，并由其原生故事影响模型训练。
- [08:13](https://www.youtube.com/watch?v=sApcIXpje1E&t=493s) 警惕免费 AI 工具的商业本质：任何免费服务往往意味着在利用用户数据训练模型，开发者与用户必须注意隐私与安全隐患。
- [10:16](https://www.youtube.com/watch?v=sApcIXpje1E&t=616s) 打造“有意识的软件”要求团队明确数据来源与去向，同时决策层中必须包含能够代表所服务用户群体的多样化成员。
- [10:16](https://www.youtube.com/watch?v=sApcIXpje1E&t=616s) AI 卓越中心（AI CoE）正像过去的 PMO 一样，成为现代企业规模化治理、落地和管理 AI 创新的标准组织架构。
- [12:06](https://www.youtube.com/watch?v=sApcIXpje1E&t=726s) “工作的未来即是当下的行动”，如 Zapier 开源人力资源 AI 技术栈，企业应秉持第一性原理在当下的工作流中践行包容性与透明化。
- [14:36](https://www.youtube.com/watch?v=sApcIXpje1E&t=876s) 通过向曾服刑人员、弱势青少年等边缘化群体提供 AI 技能重塑与准入通道，打破技术鸿沟并扩大社会公平。
- [14:36](https://www.youtube.com/watch?v=sApcIXpje1E&t=876s) 创作者生态存在“二八定律”，90% 的长尾创作者缺乏巨头资源，需重点投资由艺术家主导的“科技与创意”复合团队以保护其职业生涯。
- [17:35](https://www.youtube.com/watch?v=sApcIXpje1E&t=1055s) 使用新兴 AI 工具（如 Napkin AI）时须仔细审查细微条款（fine print），防止在未察觉的情况下让渡敏感数据与私有信息权限。
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=sApcIXpje1E&t=4s) 缘起与背景：从登月工厂到多元视角的创新力量
对谈在轻松幽默的氛围中展开。AJ Thomas 与 David Leighton 调侃两人平时的通话内容本就精彩得应该有一群观众聆听。两人的结缘始于一段奇妙的经历——AJ 曾“偷偷”把女儿带进了 WITI（Women in Technology International）的年度大会，而这一决定最终促成了她在 WITI 董事会任职数年。David 回顾了自己的职业生涯，他早在 1980 年代初就进入了 PC 软件行业，亲历了从个人电脑普及、Web 互联网爆发到如今 AI 浪潮的数次技术周期。
AJ 分享了自己在 Alphabet 旗下“登月工厂”（The Moonshot Factory，即 X）主管全球人才组织与人事产品的经历。当时正值大量 AI 技术在组织内部孵化，许多旨在影响全球数十亿人的登月项目必须推向真实世界。Alphabet 甚至专门设立了名为“接触真实世界”（in touch with the real world）的机制，让技术从快速评估阶段走向与真实人类的互动。David 对此深表赞同，他提到自己住在加州圣路易斯-奥比斯波（San Luis Obispo）这样的小城镇，虽然过去三年自身深处 AI 智库与前沿网络，但走出圈子就会发现，许多地区和人群对新技术的采纳并没有那么迅速，这凸显了技术落地中连接真实人群的重要性。
```mermaid
flowchart LR
accTitle: 登月技术向真实世界推进的演进路径
subgraph Ideation ["技术孵化阶段"]
A["Rapid Evaluation<br/>快速技术评估"] --> B["Moonshot Incubation<br/>登月技术孵化"]
end
subgraph RealWorld ["现实世界触达"]
B --> C["In Touch with the Real World<br/>与真实人类互动检验"]
C --> D["Global Impact<br/>影响全球数十亿用户"]
end
```
---
### [03:01](https://www.youtube.com/watch?v=sApcIXpje1E&t=181s) WITI 的演进：超越合规，以多元视角驱动长期创新
谈及推动包容性与技术获取的历程，David 讲述了 WITI 自 1989 年成立以来的演化逻辑。WITI 创立之初源于惠普实验室（HP Labs）的合作项目，起初核心目标是解决女性在科技行业代表性不足与数量失衡的问题。随着经历互联网繁荣与泡沫破裂的洗礼，三年前机构进行了战略重组，将 WITI 的缩写内涵重新定义为“工作场所创新、信任与影响力”（Workforce Innovation, Trust and Influence），全面拥抱 AI 时代。
David 尖锐地指出，单纯依靠“合规驱动”（compliance-driven）的多元化项目往往流于形式且难以奏效。真正的核心在于——不同的视角能够直接催生颠覆性的创新。当年惠普实验室之所以能具备前瞻 15 至 20 年技术趋势的视野，正是得益于吸纳了拥有不同背景与经历的人才，而在当前技术迭代缩短至 6 到 12 个月的 AI 时代，这种多元视角更为关键。AJ 补充道，WITI 构建的庞大社区汇聚了 Apple、Microsoft、Salesforce 等顶级科技巨头的领导者与 CIO，共同塑造着企业级产品的未来。
- **创立背景与时间**：WITI 成立于 1989 年，早期灵感与合作深度依托惠普实验室（HP Labs）。
- **品牌与使命重塑**：重组后将 WITI 定义为 Workforce Innovation, Trust and Influence（工作场所创新、信任与影响力）。
- **行业生态覆盖**：社区涵盖 Apple、Microsoft、Salesforce 等跨国科技企业的核心技术高管。
---
### [05:30](https://www.youtube.com/watch?v=sApcIXpje1E&t=330s) StoneAI 与创作者权益：构建具有数字溯源能力的伦理叙事
作为 StoneAI 的首席执行官，David 深入阐释了公司的核心理念——“带有出处证明的叙事”（storytelling with provenance）。针对“Provenance”（溯源/出处）这一术语，David 澄清其核心在于清晰展示所有权归属与知识产权（IP）的起源链条。StoneAI 主要服务于娱乐影视制作网络、好莱坞片场以及法律服务等垂直领域，旨在通过建立透明的追踪体系，在高度分化且充满版权争议的环境中保护创作者，赋能人类创意而非取代创作者。
David 介绍了其联合创始人 Evan B. Stone（拥有 30 余年从业经验的资深摄影师与电影制作人）。Evan 在 ChatGPT 问世之初便深度介入 AI 探索，并在参与 David 主持的 AI 智库研讨时达成合作共识。面对各大制片厂和媒体网络迫切希望以符合伦理的方式应用 AI 的诉求，StoneAI 结合了名为 Cyberbear 的 AI 资产标记与追踪系统。AJ 指出，这与她创立 Good Trouble Ventures 并投资“科技与创意交叉领域”的初衷不谋而合，创作者的数字权益保护正是解决内容溯源与信任危机的关键抓手。
---
### [08:13](https://www.youtube.com/watch?v=sApcIXpje1E&t=493s) 现场调查：包容性是停留在理论还是已落地为操作？
为了检验行业现状，AJ 与 David 向现场观众抛出了互动调查，重点探讨“包容性”（Inclusion）究竟是停留在理念层面，还是已经真正转化为可执行的产品研发流程：
1. **第一道提问**：“有多少人真正认为，你们当前研发的 AI 产品是由受其影响最深的群体的‘亲身体验’（lived experiences）所塑造的？”
2. **第二道提问**：“有谁能够确信，终端用户在文化、地域或其他维度的‘原生故事’（origin stories），直接影响了你们底层 AI 模型的训练方式？”
AJ 强调，当前行业内关于“AI 向善”（AI for Good）与影响力投资的讨论很多，但最根本的衡量标准在于：产品团队是否真正识别出了受该解决方案影响最深或处于边缘地位的社群，并在构建过程中主动向他们收集真实体验与反馈。David 与 AJ 提醒道，“天下没有免费的午餐，任何标榜免费的工具，本质上都在利用用户的数据来训练新模型。”两人以此引出近期 OpenAI 因协作链接/共享对话（collaborative links/chats）被公开搜索引擎抓取收录所引发的安全与隐私争议，警示开发者在构建面向广大社群的 AI 工具时必须时刻保持警惕。
### [10:16](https://www.youtube.com/watch?v=sApcIXpje1E&t=616s) 有意识的科技：决策桌上的代表性与 AI 卓越中心
AJ Thomas 深入探讨了“有意识的科技”（conscious tech）与“有意识的软件”（conscious software）的构建原则。她强调，团队在打造 AI 工具时，必须清晰了解数据的来源与流向。随后，她向全场观众提出了一个直击痛点的问题：“在座各位的决策层中，究竟有多少决策者真正能够代表你们产品所服务的用户群体？”为了活跃气氛，AJ 与台下观众展开了即兴举手互动，依次调查了现场有多少人负责为企业的 AI 项目提供资金支持、有多少人亲自参与自研或外包技术构建，以及有多少人真正走到一线与客户及终端用户直接对话。看到资金负责人举手时，她幽默地调侃道：“大家看过来，钱都在他们手里，快把手再举高一点！”
David Leighton 随即跟进，将话题延伸到企业内部的组织架构，询问现场有多少组织已经正式建立了“AI 卓越中心”（AI Centers of Excellence，AI CoE）。他指出，正如过去的专案管理办公室（PMO）一样，AI 卓越中心正日益成为企业规模化落地 AI 的标配组织形态。
```mermaid
flowchart TD
accTitle: 构建有意识的 AI 产品决策流程
subgraph 决策与数据源头 ["1. 决策与数据源头"]
A["明确数据来源与真实性"]
B["决策层涵盖目标用户代表"]
end
subgraph 组织治理落地 ["2. 组织治理落地"]
C["设立 AI 卓越中心 (AI CoE)"]
D["链接资金、研发与一线用户反馈"]
end
A --> C
B --> C
C --> D
```
- **数据溯源与有意识设计**：构建 AI 系统不仅关乎算法本身，更关乎是否清楚训练数据的起源与潜在偏见。
- **决策层的代表性短板**：产品设计与投资决策团队往往缺乏与终端社群高度重合的多样化视角。
- **AI 卓越中心（AI CoE）的兴起**：企业正加速建立类似 PMO 的标准化 AI 治理与落地中心。
---
### [12:06](https://www.youtube.com/watch?v=sApcIXpje1E&t=726s) 工作的未来即是“当下的行动”：从 Zapier 开源到第一性原理
在谈到“工作的未来”（Future of Work）这一深耕了二十多年的领域时，AJ 举了 Zapier 人才团队的典范案例。Zapier 的人才负责人 Brandon Sammut 将公司在整个员工生命周期（从入职到退休，“from hire to retire”）中应用的完整 AI 技术栈全盘开源，展示了社群共建的强大力量。David Leighton 对此表示高度共鸣，并分享 WITI（女科技人国际组织）在举办“人类与 AI 的未来”峰会后的转变，他们正评估利用 Google NotebookLM 将 WITI 多年积累的历史数据公开开源，让更多社群受益。
回顾行业演进，David 感叹在 2020 年 3 月疫情暴发前，“工作的未来”几乎是每场大会的主题演讲套话，而疫情和 ChatGPT 的相继爆发彻底重塑了现实。AJ 则对此给出了精辟的定义：“工作的未来就是当下的行动（the future of work is about the how of now）”——企业今天如何行动，便在如何定义未来。她还幽默地点名了台下 JFFLabs（Jobs for the Future Labs）未来学家委员会的同僚 Alex，调侃道：“我们总自称是未来学家，但实际上谁也无法预知未来。我们所做的，无非是观察信息、捕捉趋势，并在涉及包容性时，坚持用‘第一性原理’提出那些看似简单却尖锐的硬核问题。”
- **Zapier 人力资源 AI 技术栈开源**：涵盖员工全生命周期的 AI 工具链被无保留公开，成为 2025 年开源趋势的缩影。
- **WITI 历史资产智能化**：计划通过公共 NotebookLM 将数十年组织沉淀转化为开放知识库。
- **未来学家的第一性原理**：包容性不是口号，而是在每个细微工作流中审视“当下我们如何做”。
---
### [14:36](https://www.youtube.com/watch?v=sApcIXpje1E&t=876s) 扩大准入底线：从司法减刑人员赋能到创作者经济投资
David Leighton 分享了一个极具震撼力的真实项目：他的好友 Bruce Bryan 曾被错误定罪并在纽约监狱服刑长达 29 年，两年前终获减刑出狱。目前 Bruce 正牵头推进一项重大计划，专门面向高危青少年以及服刑人员子女等弱势群体，提供前沿的 AI 技能培训与技能重塑（upskill）。David 与 AJ 一致认为，在全球范围内，为这些长期被忽视的边缘化社群提供 AI 准入通道至关重要。
AJ 随后犀利地指出了当下科技界对“多样性”（Diversity）一词的浮躁态度。她直言，真正的多样性入场券是“构建能够赋予准入机会并明确定义包容的基础设施”。她介绍了自己创立的 Good Trouble Ventures 的核心投资逻辑：专注投资处于科技与创意交叉领域的早期团队，确保创作者能获得可持续的职业生涯。她分析了创作者生态残酷的“二八定律”——头部 10% 的顶级艺人能够与大型科技巨头深度合作，而剩下 90% 的边缘创作者则缺乏资源；AI 本应赋能这 90% 的群体，因此 Good Trouble 坚持其投资标的必须至少有一位创始人本身就是创作者或艺术家。
```mermaid
flowchart LR
accTitle: 创作者生态的 AI 准入鸿沟与投资逻辑
subgraph 头部创作者 ["头部 10% 顶尖创作者"]
T1["坐拥巨头合作与独家 AI 资源"]
end
subgraph 边缘创作者 ["长尾 90% 边缘创作者"]
M1["面临工艺被自动化替代风险"]
M2["缺乏顶级技术准入与资本支持"]
end
subgraph 破局路径 ["Good Trouble 赋能路径"]
GT["投资'创作者/艺术家'主导的科技企业"]
GT --> M1
GT --> M2
end
```
- **司法减刑与 AI 技能重塑**：Bruce Bryan 冤狱 29 年获释后，发起面向弱势青年与服刑人员子女的 AI 赋能项目。
- **超越多样性口号**：将包容性（Inclusion）下沉至产品底层的系统准入与基础设施构建。
- **Good Trouble Ventures 投资哲学**：重点下注“科技 + 艺术”复合型团队，扭转 90% 创作者在 AI 自动化浪潮中被边缘化的处境。
---
### [17:35](https://www.youtube.com/watch?v=sApcIXpje1E&t=1055s) 警惕免费工具的微小条款与闭幕警示
在收到会议倒计时一分钟的提醒时，David 话锋一转，向在场观众给出了一个非常务实的警示。他以最近流行的信息图表生成工具 Napkin AI（napkin.ai）为例，提醒大家虽然这些新工具功能极其惊艳，但用户必须仔细阅读其“微小条款”（fine print / 隐私协议）。许多免费工具在条款中默认索取了使用用户输入数据的权限，即便用户保留所有权，也可能在无意中泄露敏感信息。
AJ 在最后总结中呼吁全场：如果你在组织中恰好负责制定或审核这些产品条款与数据规则，请务必提出那些艰难而审慎的问题。她再次强调，未来的形态完全取决于当下如何构建产品与制定规则。两位演讲者在全场掌声与超时提醒的轻松气氛中致谢散场，呼吁大家在瞬息万变的技术浪潮中，持续加入 WITI、Good Trouble Labs 等全球社群展开深度协作。
- **免费 AI 工具的数据安全隐患**：如 Napkin AI 等工具虽高效，但免费服务背后往往伴随着数据使用权限的让渡。
- **细读微小条款（Fine Print）**：严防在未察觉的情况下将私有或敏感信息输入到第三方模型中。
- **责任前置**：产品设计者与合规制定者应在最前端为全人类与全球社群的数字权益把关。
## Notable quotes
> [00:04](https://www.youtube.com/watch?v=sApcIXpje1E&t=4s) 接触真实世界（in touch with the real world） — AJ Thomas
> [03:01](https://www.youtube.com/watch?v=sApcIXpje1E&t=181s) 单纯依靠合规驱动的多元化项目往往流于形式且难以奏效。真正的核心在于——不同的视角能够直接催生颠覆性的创新。 — David Leighton
> [05:30](https://www.youtube.com/watch?v=sApcIXpje1E&t=330s) 带有出处证明的叙事（storytelling with provenance） — David Leighton
> [08:13](https://www.youtube.com/watch?v=sApcIXpje1E&t=493s) 有多少人真正认为，你们当前研发的 AI 产品是由受其影响最深的群体的‘亲身体验’所塑造的？ — AJ Thomas
> [08:13](https://www.youtube.com/watch?v=sApcIXpje1E&t=493s) 有谁能够确信，终端用户在文化、地域或其他维度的‘原生故事’，直接影响了你们底层 AI 模型的训练方式？ — AJ Thomas
> [08:13](https://www.youtube.com/watch?v=sApcIXpje1E&t=493s) 天下没有免费的午餐，任何标榜免费的工具，本质上都在利用用户的数据来训练新模型。 — AJ Thomas / David Leighton
> [10:16](https://www.youtube.com/watch?v=sApcIXpje1E&t=616s) 在座各位的决策层中，究竟有多少决策者真正能够代表你们产品所服务的用户群体？ — AJ Thomas
> [12:06](https://www.youtube.com/watch?v=sApcIXpje1E&t=726s) 工作的未来就是当下的行动（the future of work is about the how of now） — AJ Thomas
> [12:06](https://www.youtube.com/watch?v=sApcIXpje1E&t=726s) 我们总自称是未来学家，但实际上谁也无法预知未来。我们所做的，无非是观察信息、捕捉趋势，并在涉及包容性时，坚持用‘第一性原理’提出那些看似简单却尖锐的硬核问题。 — AJ Thomas
> [14:36](https://www.youtube.com/watch?v=sApcIXpje1E&t=876s) 构建能够赋予准入机会并明确定义包容的基础设施 — AJ Thomas
> [17:35](https://www.youtube.com/watch?v=sApcIXpje1E&t=1055s) 如果你在组织中恰好负责制定或审核这些产品条款与数据规则，请务必提出那些艰难而审慎的问题。 — AJ Thomas
## Mentioned
AJ Thomas, David Leighton, Evan B. Stone, Brandon Sammut, Bruce Bryan, Alex, StoneAI, Good Trouble Ventures, WITI, Alphabet, The Moonshot Factory (X), HP Labs, Apple, Microsoft, Salesforce, Cyberbear, OpenAI, Zapier, Google NotebookLM, JFFLabs, Napkin AI
## Video description
David Leighton, Co-Founder & CEO, WITI, StoneAI
AJ Thomas, Founder & General Partner, Good Trouble Ventures
Abstracts
Peel back the hype and discover how inclusive design turns AI from a buzzword into a bottom‑line growth engine. In this rapid‑fire conversation, AJ Thomas (Founding Partner, Good Trouble VC) joins David Leighton (Co‑Founder & CEO, WITI) to explore the new playbook for human–AI teamwork:
* Pinpointing real‑world pain points where diverse perspectives unlock breakthrough solutions
* Embedding ethics and bias‑checks without killing velocity
* Building cross‑functional teams—operators, creatives, and data scientists—who mirror the users they serve
* Winning investor, customer, and talent confidence by proving that purpose and profit rise together
Attendees leave with a five‑step “inclusion acceleration checklist” they can start using the moment they leave the room.
[Source](https://www.youtube.com/watch?v=sApcIXpje1E)
Source quotes
[0:04] AJ Thomas: 接触真实世界（in touch with the real world）
[3:01] David Leighton: 单纯依靠合规驱动的多元化项目往往流于形式且难以奏效。真正的核心在于——不同的视角能够直接催生颠覆性的创新。
[5:30] David Leighton: 带有出处证明的叙事（storytelling with provenance）
[8:13] AJ Thomas: 有多少人真正认为，你们当前研发的 AI 产品是由受其影响最深的群体的‘亲身体验’所塑造的？
[8:13] AJ Thomas: 有谁能够确信，终端用户在文化、地域或其他维度的‘原生故事’，直接影响了你们底层 AI 模型的训练方式？
[8:13] AJ Thomas / David Leighton: 天下没有免费的午餐，任何标榜免费的工具，本质上都在利用用户的数据来训练新模型。
[10:16] AJ Thomas: 在座各位的决策层中，究竟有多少决策者真正能够代表你们产品所服务的用户群体？
[12:06] AJ Thomas: 工作的未来就是当下的行动（the future of work is about the how of now）
[12:06] AJ Thomas: 我们总自称是未来学家，但实际上谁也无法预知未来。我们所做的，无非是观察信息、捕捉趋势，并在涉及包容性时，坚持用‘第一性原理’提出那些看似简单却尖锐的硬核问题。
[14:36] AJ Thomas: 构建能够赋予准入机会并明确定义包容的基础设施
[17:35] AJ Thomas: 如果你在组织中恰好负责制定或审核这些产品条款与数据规则，请务必提出那些艰难而审慎的问题。
Source quotes
[0:04] AJ Thomas: in touch with the real world
[3:01] David Leighton: Workforce Innovation, Trust, and Influence
[5:30] David Leighton: storytelling with provenance
[5:30] David Leighton: superpowers
[8:13] David Leighton: Nothing is free. If it says free, that means you're just giving them data to train new models.
[12:06] AJ Thomas: the how of now
[12:06] AJ Thomas: hire to retire
[17:35] AJ Thomas: ask the hard question
Original transcript
[0:04] Well, hope everyone's feeling great
[0:05] today. We've got some some good fireside
[0:08] chatting to do in
[0:10] >> the next 19 minutes and 43 seconds.
[0:12] >> Exactly. I know because every time we
[0:14] talk to each other on the phone, I'm
[0:15] like, there should be an audience for
[0:17] this or at least your part, AJ. I'm
[0:19] like, wow. Yes. Be on stage.
[0:22] >> Awesome. Um well I mean our talk today
[0:25] is really about uh how we think about
[0:28] access and inclusion in this new world
[0:30] that we're all living in which by the
[0:32] way has been the world that we've been
[0:34] living in for a while but just new data
[0:35] coming in new technologies being built.
[0:38] Um so speaking of origin stories how
[0:41] David and I had an opportunity to meet
[0:43] um was actually me sneaking my daughter
[0:46] into Witty's annual conference. Uh best
[0:50] decision I think I ever made.
[0:51] >> That's right. Um because then ended up
[0:53] serving on your board for I don't know
[0:56] three four years.
[0:57] >> Many years. Yeah. Many years.
[0:59] Absolutely. Right. And just you know
[1:01] when you talk about origin stories also
[1:04] coming you know I come from early 80s
[1:08] right and then PC software days where it
[1:12] was kind of a fight for people any types
[1:15] of people.
[1:16] >> Yeah.
[1:16] >> Then the web and a lot of things
[1:18] happened. Mhm.
[1:20] >> Now here we are in AI and just like what
[1:22] you were leading at Alphabet was so
[1:25] impactful for just how different
[1:27] perspectives lead to innovation.
[1:29] >> Yeah. A little bit about my story um
[1:32] before we kind of get into some of these
[1:34] different topics because we're going to
[1:35] have questions for everybody here in the
[1:36] room as well. Um, I had the opportunity
[1:38] to actually serve as uh heading up
[1:41] people products um and the global talent
[1:44] organization at the Moonshot Factory
[1:46] which was insane um given it was around
[1:49] the time that a lot of the AI
[1:50] technologies were being incubated within
[1:52] the organization as you guys all know
[1:55] and um a lot of the moonshots that were
[1:57] were having to push out into the real
[1:58] world were needing a lot more
[2:01] interaction with humans. we actually had
[2:03] a role within alphabet that was called
[2:06] um in touch with the real world. So you
[2:08] would go from rapid evaluation to being
[2:11] able to touch the real world with that
[2:13] technology. Otherwise it would not
[2:15] actually be considered a moonshot type
[2:17] of technology, right? So if you're
[2:19] affecting billions of people around the
[2:21] world with the technology that you're
[2:23] trying to bring to bear, you want as
[2:24] many people as possible being able to
[2:26] have conversations around it, too. So
[2:28] that's what you're talking about with my
[2:30] role there, which has been great.
[2:31] >> Yeah. No, totally. And just, you know,
[2:34] and I feel like that's true with AI. I
[2:36] live in I live in the St. Louis Abyspo,
[2:39] California. Kind of smi small townish,
[2:43] >> you know, and and then just been flooded
[2:45] with AI for the last three years and,
[2:48] you know, leading think tanks, just
[2:49] talking to people in my network. Then
[2:51] you go a little outside of that.
[2:54] >> Yeah.
[2:54] >> I don't think it's affecting,
[2:56] >> you know, some people aren't jumping on
[2:58] I feel as quickly as they could be in
[3:00] some areas.
[3:01] >> Yeah. Well, your work with Witty has
[3:02] been so fantastic. How how many decades
[3:04] now have you been
[3:06] >> 89 we started Witty? It looks maybe
[3:08] before many people on the planet in the
[3:10] audience, but you know, I started very
[3:13] young.
[3:13] >> You started very young. Yes. When he was
[3:16] two months old. Um so David I I do want
[3:19] to talk about the work of witty because
[3:22] I think when we think about access and
[3:24] inclusion specifically for women in
[3:26] technology right and this is again this
[3:27] is how we had had the opportunity to
[3:29] meet um what was the thing that was the
[3:32] driving force that allowed for you all
[3:34] to weather all of the technological
[3:36] transformations that were happening?
[3:38] Well, you know, at the beginning it was
[3:40] around
[3:42] numbers and just lack of women, right?
[3:44] So that's how it started. Always
[3:46] including everybody what we were doing
[3:49] 80s9s. A lot of it came from our work
[3:52] with HP Labs way back then. Um,
[3:57] so that kind of led to the starting of
[3:59] Witty as a community and then as it you
[4:01] know grew through the dot growth, the
[4:05] com leveling out we'll call it and then
[4:08] three years ago when we merged our
[4:10] companies changed the acronym of witty
[4:13] to workforce innovation, trust and
[4:15] influence really a lot of that you know
[4:17] was around our new world of AI and how
[4:21] to just how the different perspectives
[4:24] really lead to innovation.
[4:26] Compliance driven programs we always
[4:29] felt just weren't going to be effective.
[4:32] >> And it was like that's what with HP Labs
[4:36] who was like our biggest client before
[4:37] we started Witty was always about
[4:40] different people getting different
[4:43] perspectives and being having being able
[4:46] to look 15 to 20 years out on what
[4:48] technology would look like. right now is
[4:51] that 6 to 12 months kind of right it's
[4:54] you know
[4:55] >> David's being super humble here because
[4:56] I think the community that they've
[4:58] architected at Witty has been some of
[4:59] the foremost leading CIOS in some of the
[5:02] largest organizations can you talk about
[5:04] some of the organizations that are part
[5:06] of Witty and contributing their origin
[5:08] stories to help build products in their
[5:09] enterprises
[5:10] >> right I mean you know early ones Apple
[5:13] Microsoft you know Salesforce you know I
[5:17] kind of look at it everybody's been
[5:19] through witty at some point, right?
[5:22] >> But I think um yeah, so it's it's been
[5:24] amazing ride and now to come out with AI
[5:27] and just the power of all these tools.
[5:30] >> Yeah,
[5:30] >> crazy writing that for you. You are now
[5:33] the CEO of a company called Stone AI.
[5:36] Can you tell me a little bit about what
[5:37] that is?
[5:38] >> Sure. Stone AI is storytelling with
[5:41] provenence.
[5:42] See,
[5:43] >> can you define what provenence? Yeah, I
[5:45] know Providence, Rhode Island.
[5:46] >> No, I that's what I said when I first
[5:48] heard that word. No, but you know it's
[5:49] really about showing origin of ownership
[5:52] and IP which is critical. So our clients
[5:54] at Stone AI, we do a lot of work with
[5:57] networks and studios in the
[5:59] entertainment space, um, legal
[6:02] verticals,
[6:03] some others, but it's just how do you
[6:05] track everything? How do we protect the
[6:08] creators? How do we how do we create
[6:10] this human and AI collaboration?
[6:14] So I think especially in industries like
[6:16] entertainment where just things are very
[6:18] polarizing and we want to empower the
[6:21] people to just give them superpowers to
[6:23] use their human creativity to to give us
[6:27] all the gifts that we see every day.
[6:30] >> Yeah. This is I think what also brought
[6:32] you and I together in my work now with
[6:34] good trouble ventures, right?
[6:36] >> When we met I was in the people and
[6:38] talent space leading a prop tech
[6:41] company,
[6:41] >> right? um and now investing in the
[6:44] intersection of technology and
[6:46] creativity. I think what's really
[6:48] intriguing about what you're doing is
[6:49] protecting the creators rights and what
[6:51] that actually looks like. How did you
[6:53] come about that? Because in lots of
[6:56] stories there are different origin
[6:58] stories being built into the narratives
[7:00] that are out there.
[7:02] >> Totally. And right and I think you know
[7:03] my partner Evan Bstone
[7:06] is another Evan Stone. Don't forget the
[7:08] B. But he long story but he uh you know
[7:12] has been cinematographer, creator,
[7:14] filmmaker for 30 plus years. dove into
[7:19] AI right when Chat GPT was released and
[7:23] just started coming to my AI think tanks
[7:25] I lead a few times a week and we became
[7:27] tight and I was seeing what he was doing
[7:30] and all the you know as as a lot of the
[7:33] networks and and production companies
[7:36] just are trying to like just are getting
[7:38] AI and how they could use it to actually
[7:41] tell stories in an ethical way.
[7:45] He's like, "I need a partner." And I
[7:47] said, "Yeah, great timing." Because
[7:49] we're also working with Cyberbear.
[7:51] That's like a AI tag. It's a It's a
[7:53] asset tagging system.
[7:56] >> So anyway, brought the two together.
[7:57] That's what we do. But this isn't, you
[7:59] know, but I want I know we have
[8:01] questions for the audience too. And I
[8:03] think that, you know, and also what
[8:04] you're doing is so amazing with um
[8:07] >> with making trouble with good trouble
[8:09] >> with good trouble ventures, you know,
[8:11] which anyway we'll talk more about, but
[8:12] let's get
[8:13] >> Yeah. So, we kind of want to get a feel
[8:15] for everybody here. This is questions
[8:16] for everyone. So, please feel free to
[8:18] look around. We are going to ask you to
[8:19] raise your hands. You don't have to
[8:20] stand up unless you want to. Extra
[8:22] points for those that do. Um, so our
[8:25] first question is, how many of you
[8:26] actually believe that your AI products
[8:29] today are shaped by lived experiences of
[8:32] the communities most affected by them?
[8:35] Raise your hand.
[8:40] How many are building AI products today?
[8:44] Okay.
[8:45] >> Nice.
[8:46] >> Okay. So, the reason why we asked that
[8:47] is I think we want to know whether
[8:51] inclusion is actually operationalized
[8:54] versus a theoretical thing. There's a
[8:55] lot of things happening with AI for
[8:57] good. There's a lot of investments being
[8:59] made on the impact side for how you
[9:01] build the kinds of technology that will
[9:03] impact those communities. And when we
[9:04] talk about inclusion, we really want to
[9:06] think about who are the most
[9:08] marginalized communities that are
[9:09] affected by the solution that you're
[9:11] bringing. Are you even as simple as
[9:13] asking them for feedback around how
[9:14] they're experiencing your build? Right.
[9:17] Um the next one. So origin stories. So
[9:22] who here can confidently say that origin
[9:25] stories whether they're cultural,
[9:27] regional, or otherwise of your end users
[9:30] directly influence how your AI models
[9:33] are being trained.
[9:38] Okay. Although we may not have too many
[9:40] trainers of AI models.
[9:42] >> It's okay,
[9:42] >> right? But yeah,
[9:43] >> we all are in some way or another as
[9:45] we're using it.
[9:46] >> I think right so many of these products
[9:48] I always tell right people the classes
[9:51] nothing is free. It says free that means
[9:54] you're just giving them data to train
[9:55] new models.
[9:56] >> That's right. That's right. Well, you
[9:58] guys saw some of the um backlash that
[10:01] OpenAI is getting as well with the
[10:04] collaborative links that are being sent
[10:07] out, collaborative chats that are being
[10:08] sent out that are being discoverable now
[10:10] online, right? So, thinking about all of
[10:13] these different use cases as you're
[10:14] building for the communities that are
[10:16] using them. Um, massive exposure if
[10:18] you're not really thinking about what
[10:20] all those different principles are. Um,
[10:22] let's talk about access and design
[10:25] principles. So much of the work around
[10:27] being able to build what I like to say
[10:29] conscious tech or conscious software is
[10:32] really being able to have um
[10:36] AI tools that you are building with uh
[10:41] knowing the origin by which those the
[10:43] data where the data is coming from right
[10:46] and so if we think about this I think
[10:48] the question becomes you know how many
[10:51] have decision-making tables where if
[10:55] you're making a decision on a product or
[10:57] a feature or a build or a project or an
[10:59] investment, whatever it may be, whatever
[11:00] role you're playing in the room, right?
[11:04] How many of your decision makers
[11:06] actually represent the community of
[11:08] folks that you're building your product
[11:09] for? Right? These are just interesting
[11:12] questions to be able to to ask. I see a
[11:14] lot of you nodding. How many of you fund
[11:16] AI initiatives in your organization?
[11:20] All right, look around the room. They
[11:22] got the dollars. Raise them again.
[11:24] >> Right.
[11:25] >> Awesome.
[11:26] How many of you build the technology in
[11:28] house or out? Okay. Awesome. We got
[11:31] builders here as well. Awesome. And how
[11:35] many of you are out there talking to
[11:37] customers or users of your of your
[11:40] products that you're building?
[11:42] Great. That's amazing. That's amazing.
[11:46] David,
[11:46] >> no. I was just going to mention too like
[11:48] AI centers of excellence. Have Have you
[11:51] are any of those building those formally
[11:53] within your organizations
[11:56] yet? There's one. Excellent. Yeah, we
[11:58] just had a I'm seeing that more and
[12:00] more, right? Just like project
[12:02] management offices.
[12:03] >> Yeah.
[12:04] >> And I think we'll see that quite a bit
[12:05] more.
[12:06] >> We're seeing a lot. So, I want to talk a
[12:07] little bit about the future of work.
[12:08] It's a space that I've played in for the
[12:10] past two decades, which has been really
[12:12] amazing in where we're seeing a lot of
[12:15] innovation. How many of you probably or
[12:17] some of you here have heard of the work
[12:19] that Zapier has done?
[12:22] Yes. So, Brandon Sammit who is leading
[12:24] their people and talent function there
[12:26] has been really amazing in opensourcing
[12:28] the entire AI tech stock for what
[12:32] they've done in their employee
[12:33] experience life cycle. You go from hire
[12:36] to retire and you see everywhere they're
[12:39] being implemented. It's all open sourced
[12:40] as well. This is also the year of open
[12:42] source for a lot of things. I love that
[12:44] you can build out with a community a lot
[12:47] of like what Witty was doing during the
[12:48] future of humanity and AI.
[12:50] >> Yeah.
[12:51] >> Talk a little bit about that.
[12:52] >> No, totally. Well, oh, as far as the
[12:54] last summit, right, and just with the
[12:57] shift and I'm actually looking at
[13:00] notebook LM public notebook LM for a lot
[13:03] of the historical witty stuff just to
[13:06] open source it to communities. Um, I
[13:09] think that's interesting. But yes, that
[13:12] I mean what we spoke about at the summit
[13:14] was just positioning ourselves and for
[13:17] the future, right? Because if you think
[13:18] about leading up to the leading up to
[13:22] March 2020 when the pandemic happened, I
[13:25] it was feels like every conference I
[13:27] went to future of work was the keynote
[13:30] or this or that, right? And then it was
[13:32] like, okay, here's here's the pandemic
[13:35] and a new way we all work.
[13:37] >> Yeah. Yeah. Then after that once open
[13:39] AAI released chat GPT that kind of
[13:42] extended it
[13:44] >> how do we
[13:45] >> well I always like to say that the
[13:46] future of work is about the how of now
[13:50] >> right so if you think about how you're
[13:52] doing things now you're actually
[13:54] defining what the future of work looks
[13:55] like
[13:56] >> um it's you know I have the opportunity
[13:59] of serving here with Alex I'm just going
[14:01] to call you out because I see you there
[14:02] hi um on jobs for the future labs
[14:05] futurist council and we always joke
[14:07] around and say So funny, we call
[14:08] ourselves futurists, but we have we know
[14:10] nothing about what's going to happen in
[14:11] the future. Um, and a lot of it is
[14:14] really looking at the information, being
[14:17] able to spot trend, and being able to
[14:18] ask the hard questions, right? And so
[14:21] some of the questions we just asked
[14:22] earlier are probably some very simple
[14:24] and innocuous questions, but when you
[14:26] think about inclusion, it is a very much
[14:28] a first principle mindset that you need
[14:31] to be taking a look at.
[14:32] >> Totally.
[14:34] >> For sure. Um, yeah. And I was thinking
[14:36] too, AJ, you're doing so much phenomenal
[14:38] work
[14:40] that I'd love for you to share with
[14:42] everybody and you know it's it's
[14:43] different than Zoom calls and stuff,
[14:45] right? Or just even smaller venues we
[14:47] know. I mean, please hit us up on
[14:49] LinkedIn. There's so many projects
[14:51] going, you know, I have I have one uh
[14:54] project my friend Bruce Bryan's heading
[14:56] up. Did you meet Bruce? He was
[14:58] wrongfully convicted in New York who
[15:01] spent 29 years in in New York prisons
[15:05] was got a man got um
[15:09] clemency two years ago and we have a
[15:12] whole project now for like under
[15:14] reppresented you know for at risk youth
[15:16] and how do we upskill everybody on AI
[15:19] right and
[15:20] >> and and children of incarcerated parents
[15:22] and stuff like that right I mean there's
[15:24] just so many communities all around the
[15:27] globe
[15:28] Yeah, I think in the conversations that
[15:30] David and I tend to have, especially
[15:31] when we were uh when I was on the board
[15:33] at Woody, was how do we expand access,
[15:37] right? Um there's the dirty word of
[15:39] diversity that keeps going around. Some
[15:41] of you are smiling out there, but I
[15:43] think it's the price of admission for
[15:46] true diversity is really looking at the
[15:49] infrastructure of how you're able to
[15:52] give access and define inclusion as an
[15:55] organization. And you can do that as a
[15:57] product organization. And I think that
[15:59] the work that I'm doing now with Good
[16:00] Trouble, I mean, we are looking for
[16:02] companies to invest in at the
[16:04] intersection of tech and creativity that
[16:06] actually benefit the creators to live
[16:08] more um sustainable creative careers,
[16:12] right? How many of you out there are
[16:14] creatives?
[16:15] Oh, you got more hands up here than
[16:18] than we did uh look looking at who
[16:20] builds AI technology. So you you know
[16:23] what that feels like when you're looking
[16:25] at something and going, "Oh, well, the
[16:26] thing I used to do that I spent years of
[16:29] my craft doing is now being automated.
[16:31] What does that actually look like?"
[16:32] Right? The top 10% of creators and
[16:35] entertainers are collaborating with some
[16:37] of the big incumbent companies out there
[16:40] that are building some of these massive
[16:42] technologies. And I think there's the
[16:44] 90% that actually are the marginalized
[16:47] community in the creator community
[16:50] because they don't have that certain
[16:52] type of access. But now AI has kind of
[16:56] expanded that. And so I think for the
[16:59] work that we're doing with Good Trouble,
[17:00] we're really looking for number one, our
[17:02] thesis is about investing in companies
[17:05] that intersect technology but have at
[17:07] least a creator or a founder that is a
[17:11] creator, a creative or an artist, right?
[17:13] So little things that you can do. I
[17:15] would encourage if you look at your
[17:17] organization now, what are simple things
[17:20] that you can build in how you include
[17:25] other communities that may be an end
[17:27] user?
[17:28] >> Y
[17:28] >> right um but not necessarily be able to
[17:32] touch the product at the very very
[17:33] beginning when you're building it.
[17:35] >> Totally. Um we got the one minute
[17:38] warning. So I will say one thing we did
[17:41] re I just want to share with everybody
[17:43] any users of Napkin AI.
[17:47] Oh not too many napkin.ai for
[17:49] infographics but just a lot of these
[17:51] tools read read um the fine print with
[17:55] some of them. Napkin's great but when
[17:57] you a lot of these free tools they will
[18:00] you're kind of giving them permission
[18:03] to use info you put in. And so there's a
[18:06] lot of remarkable things out there. But
[18:08] anyway, just reach out to either of us
[18:10] if you need support with any of that.
[18:12] Like napkin.ai. Everybody will be hooked
[18:15] after I after I tell you now that I just
[18:17] told you about it, but just make sure
[18:20] you're not putting details in there that
[18:22] you don't want shared. You still own it.
[18:25] But I just think these are just some of
[18:26] the things that for the next three days,
[18:28] everybody should dive deep into or get
[18:31] involved in Witty Good Trouble Labs, all
[18:34] of our communities, right? things are
[18:37] changing on the daily. So I think we get
[18:38] us
[18:39] >> I think that's a good point David to
[18:40] just end on as we look at our wrap up
[18:43] our session is look at the fine print
[18:45] and if you happen to be in your
[18:47] organization that is responsible for
[18:49] input into the fine print ask the hard
[18:53] question. So much about the future is
[18:56] about the how of now and so how are you
[18:58] guys showing up for that as we're
[18:59] creating these products for everybody
[19:01] else.
[19:01] >> Thank you. Yes, we're a global
[19:03] community, right? It's everything's
[19:04] glodpr.
[19:07] Anyway, we got we're getting the boot
[19:10] >> you everybody
[19:11] >> back. Thank you guys.
[19:12] >> Thanks. That was really great.
