{"id":28976,"date":"2026-01-29T08:49:07","date_gmt":"2026-01-29T16:49:07","guid":{"rendered":"https:\/\/pcvdevel.wpenginepowered.com\/?page_id=28976"},"modified":"2026-05-11T12:11:55","modified_gmt":"2026-05-11T19:11:55","slug":"%e4%ba%ba%e5%b7%a5%e6%99%ba%e8%83%bd%e6%94%bf%e7%ad%96","status":"publish","type":"page","link":"https:\/\/www.pacificcommunityventures.org\/zh\/ai-policy\/","title":{"rendered":"\u4eba\u5de5\u667a\u80fd\u653f\u7b56"},"content":{"rendered":"<h1>\n\t\t\t\u5e38\u89c1\u95ee\u9898\u89e3\u7b54\t<\/h1>\n<h2>\n\t\t\t\u592a\u5e73\u6d0b\u5171\u540c\u4f53\u98ce\u9669\u6295\u8d44\u516c\u53f8\uff08PCV\uff09\u5173\u4e8e\u8d1f\u8d23\u4efb\u4e14\u5408\u4e4e\u9053\u5fb7\u5730\u4f7f\u7528\u4eba\u5de5\u667a\u80fd\u548c\u673a\u5668\u5b66\u4e60\u7684\u653f\u7b56\t<\/h2>\n\t<p>2026\u5e743\u6708<\/p>\n\t<p>PCV\u7684\u5ba2\u6237\u3001\u987e\u5ba2\u3001\u540c\u4e8b\u3001\u884c\u4e1a\u5408\u4f5c\u4f19\u4f34\u548c\u652f\u6301\u8005\uff0c,\u00a0<\/p>\n<p>Human civilization has entered a new era, the 4th\u202findustrial revolution, of technological\u00a0innovation\u00a0as artificial intelligence (AI) and machine learning (ML)\u00a0bring\u00a0the power and potential to streamline our business operations, generate ideas and content, and process vast information at speeds beyond human capacity.\u00a0<\/p>\n<p>PCV has long been a data-driven, &#8220;high tech and high-touch&#8221; organization founded as one of the country&#8217;s first impact investing organizations. PCV is dedicated to investing in small business entrepreneurs&#8217; passion and resilience, helping them create\u00a0good quality\u00a0jobs and advance economic mobility, climate\u00a0resilience,\u00a0and financial wellbeing outcomes, while uplifting communities that have been historically underserved.\u00a0PCV is also a federally and state-certified community development financial institution (CDFI),\u00a0representing\u00a0a field born out of the Civil Rights movement as\u00a0its\u00a0economic justice pillar, more than 30 years ago following a federal acknowledgment of\u00a0historical\u00a0redlining in the formal financial industry.\u00a0<\/p>\n<p>We acknowledge the potential ethical issues that AI poses, and the impact, both positive and negative, that it can have on our employees,\u00a0clients,\u00a0customers, and communities. We are\u00a0committed to\u00a0continually\u00a0researching\u00a0and understanding\u00a0the algorithmic bias perpetuated by AI\u00a0as its capabilities advance, particularly in sensitive industries related to ours &#8211; such as financial decisions, hiring, and criminal justice.\u00a0<\/p>\n<p>Therefore, we\u00a0are\u00a0developing,\u00a0implementing, and\u00a0regularly updating\u00a0a comprehensive policy around the ethical use of AI and ML that aligns with our company&#8217;s values, mission, and goals.<b>\u6211\u4eec\u8ba4\u4e3a\uff0c\u5229\u7528\u4efb\u4f55\u4eba\u5de5\u667a\u80fd\u6216\u673a\u5668\u5b66\u4e60\u5e94\u7528\u3001\u5de5\u5177\u3001\u65b9\u6cd5\u6216\u7cfb\u7edf\u7684\u51b3\u7b56\u90fd\u5fc5\u987b\u7ef4\u62a4\u4e2a\u4eba\u548c\u793e\u533a\u7684\u540c\u610f\u3001\u6570\u636e\u81ea\u4e3b\u6743\u548c\u9690\u79c1\u6743\uff0c\u4ee5\u53ca\u5bf9\u4eba\u7c7b\u80fd\u52a8\u6027\u548c\u5c0a\u4e25\u7684\u5c0a\u91cd\uff0c\u540c\u65f6\u6700\u5927\u9650\u5ea6\u5730\u51cf\u5c11\u5bf9\u6c14\u5019\u7684\u4e0d\u5229\u5f71\u54cd\u3002.<\/b>\u00a0<\/p>\n<p>\u56e0\u6b64\uff0c\u5728\u4f7f\u7528\u4eba\u5de5\u667a\u80fd\u548c\u673a\u5668\u5b66\u4e60\u6280\u672f\u65f6\uff0c<b>PCV\u4e25\u683c\u9075\u5b88<\/b>\u63a5\u4e0b\u6765\u7684\u516d\u4e2a<b>\u00a0\u6838\u5fc3\u539f\u5219<\/b>:\u00a0<\/p>\n<ol>\n<li>PCV\u00a0must ensure that the benefits of the AI used minimize the harm created, and it must not harm any\u00a0particular group\u00a0or individuals disproportionately.<\/li>\n<li>PCV\u00a0must select or develop AI that has been trained on demographically disaggregated data to prevent discrimination or disproportionate harm on any person or group of persons\u00a0by the use of\u00a0that AI system. We acknowledge that algorithms trained without proportionally\u00a0balanced\u00a0demographics\u00a0nearly always\u00a0perform worse for low-income, underrepresented, or\u00a0underserved groups.<\/li>\n<li>The AI system must never be used to discriminate against or exclude individual\u00a0clients,\u00a0customers,\u00a0or groups of people during the course of business.\u00a0AI systems include, but are not limited to, predictive models, matching algorithms, support desk chat\u00a0bots, recommendation engines, auto-generated content for marketing and communications, tools, and scripts to automate administrative tasks (e.g., like finance, loan processing, and reporting) and others.<\/li>\n<li>AI systems used at PCV must\u00a0at all times\u00a0prioritize and safeguard clients&#8217; and customers&#8217; privacy and data rights, both, to protect data from abuse or monopolization\u202fand\u202fto ensure we are abiding by loan-level privacy legislation as we are\u00a0obligated\u00a0as\u00a0a CDFI. Clients include\u00a0small\u00a0business entrepreneurs\u00a0and their workers. Customers include other CDFIs or mission-driven organizations that PCV supports in its role as a capacity builder in data, analytics, ML, and AI. PCV commits to strict protection of customer and client data,\u00a0at all times,\u00a0maintaining\u00a0this data within a firewalled environment separated from other program data.\u00a0Clients and customers\u00a0shall\u00a0maintain\u00a0the right to request access to their data and remove it from PCV&#8217;s systems.\u00a0PCV will undertake\u00a0the best\u00a0efforts to\u00a0fulfill such a request.\u00a0<\/li>\n<li>\u4eba\u5de5\u667a\u80fd\u7cfb\u7edf\u5e94\u8be5\u7528\u4e8e<i>\u544a\u77e5\u548c<\/i>\u202f<i>\u652f\u6301<\/i>\u202fhuman critical analysis and decision-making and should never\u00a0replace\u00a0human decision-making, nor\u00a0be unsupervised\u00a0in any scenario, nor\u00a0act as a\u00a0fully\u00a0autonomous system.\u00a0\u00a0An AI system must always\u00a0be designed to keep a human in the\u00a0decision-making\u00a0loop.\u00a0\u00a0TAs AI and ML systems become increasingly autonomous and agentic; the following examples are not exhaustive but intend to guide PCV staff:\u00a0\n<ul>\n<li><strong>Coding agents (Claude Code, Copilot, Antigravity, Jules, etc.)<\/strong><strong>:<\/strong>\u00a0heavy\u00a0and unsupervised\u00a0reliance on coding agents is forbidden at PCV. Coding agents may be\u00a0leveraged\u00a0during\u00a0algorithm-building,\u00a0assembling data engineering workflows,\u00a0software development, testing, and deployment,\u00a0only\u00a0as checks\u00a0on human-authored code,\u00a0as\u00a0tools to\u00a0assist\u00a0troubleshooting,\u00a0as aids to generate test scripts, and as guides to advise on efficient deployment.\u00a0Coding\u00a0agents\u00a0must never be\u00a0activated on tasks that generate new, unsupervised code,\u00a0access data,\u00a0\u00a0folders,\u00a0databases,\u00a0or other repositories without explicit\u00a0human\u00a0permission,\u00a0scrape\u00a0any preexisting content without explicit human permission, or\u00a0be\u00a0allowed to\u00a0autonomously run scripts\u00a0or actions\u00a0of any kind without a two-person review, testing, and\u00a0formal move\u00a0of the codebase\u00a0into production.\u00a0At any point of\u00a0a technical project lifecycle, a developer must be able to explain one or more lines of code when\u00a0questioned. An inability to provide\u00a0a rationale or\u00a0appropriate levels\u00a0of transparency\u00a0by a technical human resource\u00a0will be assumed\u00a0as\u00a0an indication\u00a0of\u00a0unauthorized AI use.\u00a0\u00a0\u00a0<\/li>\n<li><strong>\u6cd5\u5b66\u7855\u58eb\u8f85\u52a9\u5199\u4f5c<\/strong><strong>:\u00a0<\/strong>fully autonomous writing (for example,\u00a0leveraging\u00a0the outputs of OpenAI, Claude, and other\u00a0LLMs, and representing\u00a0them as one&#8217;s own writing) is\u00a0strictly forbidden\u00a0at PCV. Staff may use such tools\u00a0to support the writing process &#8211; such as to brainstorm\u00a0new program ideas, conduct research on primary sources, assist with an outline for a well-structured proposal,\u00a0conduct grammar and spell-checking on final content,\u00a0and similar\u00a0light editing functions,\u00a0but the actual written content must be\u00a0one&#8217;s own writing and represent one&#8217;s own original thinking.\u00a0When presenting a final deliverable to one&#8217;s manager\u00a0or other stakeholder,\u00a0PCV staff must\u00a0disclose\u00a0AI tool\u00a0use\u00a0either verbally, in email, or as a footnote on\u00a0the writeup. The disclosure should include the AI tool used, the approximate dates\u00a0and activities for which\u00a0it was used.\u00a0<\/li>\n<li><strong>\u4eba\u5de5\u667a\u80fd\u751f\u6210\u7684 <\/strong><strong>\u62b5\u62bc\u54c1\uff1a<\/strong><strong>\u00a0<\/strong>fully autonomous content creation (for example, websites,\u00a0slides, logos,\u00a0graphics, social media posts, photos, videos, landing pages,\u00a0style guides,\u00a0user interfaces, user experience workflows,\u00a0and\u00a0branding\u00a0collateral, as a few examples)\u00a0without thorough human review\u00a0and without proper attribution,<strong>\u00a0<\/strong><strong>\u5728PCV\uff0c\u8fd9\u662f\u4e25\u683c\u7981\u6b62\u7684\u3002.<\/strong>\u00a0Staff may use\u00a0tools such as Base44, Miro, Figma, Canva, Framer,\u00a0Tailwind &#8211; and others &#8211; to generate\u00a0prototypes whose purpose is to communicate their ideas and vision,\u00a0assist\u00a0them and their colleagues with brainstorming, and communicate\u00a0final\u00a0specifications. Such AI-generated collateral, however, should\u00a0never be considered\u00a0final deliverables, nor be pushed into production, without\u00a0a thorough code review (for\u00a0example,\u00a0new\u00a0landing pages, web-based tools), and management sign-off to ensure brand consistency, security, and intellectual property considerations.\u00a0\u00a0<\/li>\n<\/ul>\n<\/li>\n<li>PCV will\u00a0not use\u00a0autonomous AI to make lending decisions\u00a0or other decisions of consequence without a change to this policy\u00a0and CEO approval. Consequential decisions include but are not limited to the following:\u00a0\n<ul>\n<li>Determining\u00a0who will receive and not receive a loan.\u00a0<\/li>\n<li>Determining\u00a0who will receive and not receive business advisory support.<\/li>\n<li>Determining\u00a0who we will inform of our products, services, and new opportunities, such as\u00a0participating\u00a0in a study, a new pilot program, or incentives of any kind.\u00a0<\/li>\n<li>If a particular AI-supported decision poses any doubt,\u00a0PCV will ensure\u00a0that no group is barred from a PCV product or service without\u00a0citing specific\u00a0programmatic\u00a0or loan eligibility criteria\u00a0or legal rationale. Always ensure that no information is withheld from a particular group or groups, again, without\u00a0a\u00a0previously\u00a0communicated programmatic or loan eligibility criteria or legal rationale.\u00a0<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n<p>Because algorithms are trained on large data sets of lending, hiring, and criminal justice decisions that\u00a0often employ\u00a0a definition of &#8220;risk&#8221; that has historically excluded low-income and\u00a0underserved\u00a0communities,\u00a0PCV\u00a0has to\u00a0be particularly intentional about the use cases and\u00a0guardrails\u00a0puts\u00a0in place to test them within PCV and our\u00a0community finance\u00a0industry. We recognize that these\u00a0new technologies\u00a0can easily\u00a0reverse\u00a0the gains\u00a0made to date with\u00a0PCV&#8217;s\u00a0high-tech high-touch approach and\u00a0risk\u00a0accelerating\u00a0bias\u00a0in community\u00a0finance\u00a0field at a scale and speed that we may never be able to address before adverse impacts get deeply reinforced.\u00a0<\/p>\n<p>PCV recognizes\u00a0that what we do as a CDFI, an impact investor, a community development organization &#8211; in terms of\u202f<i>\u5982\u4f55<\/i>we deploy AI now -may\u00a0set a precedent for the sector for years to come. As we increase our technology and AI capacities within PCV, we commit to sharing learnings with the CDFI and impact ecosystem, advocate for evidence-backed practices that center clients and community voices, protect against bias, curtail market drift, prevent exclusion, and ensure we seek the explicit consent of those we and our peers serve.\u00a0<\/p>\n<p>\u5728\u4e0e\u540c\u884c\u793e\u533a\u53d1\u5c55\u91d1\u878d\u673a\u6784 (CDFI) \u548c\u5176\u4ed6\u4ee5\u4f7f\u547d\u4e3a\u5bfc\u5411\u7684\u7ec4\u7ec7\u5408\u4f5c\u65f6\uff0c\u6211\u4eec\u5fc5\u987b\u5206\u4eab\u6211\u4eec\u7684\u4ef7\u503c\u89c2\u4ee5\u53ca\u6211\u4eec\u8ba4\u4e3a\u53ef\u63a5\u53d7\u7684\u4eba\u5de5\u667a\u80fd\u4f7f\u7528\u89c4\u8303\uff08\u8be6\u89c1\u4e0b\u6587\uff09\u3002\u5982\u679c\u5728\u5408\u4f5c\u8fc7\u7a0b\u4e2d\uff0c\u6211\u4eec\u53d1\u73b0\u5f7c\u6b64\u5728\u8fd9\u4e9b\u4ef7\u503c\u89c2\u548c\u89c4\u8303\u65b9\u9762\u5b58\u5728\u5206\u6b67\uff0c\u6211\u4eec\u5c06\u6709\u6743\u7ec8\u6b62\u5408\u4f5c\u3002\u5bf9\u4e8e\u6211\u4eec\u8fbe\u6210\u5171\u8bc6\u7684\u7ec4\u7ec7\uff0c\u6211\u4eec\u5c06\u81f4\u529b\u4e8e\u6784\u5efa\u4e00\u4e2a\u5b9e\u8df5\u793e\u533a\uff0c\u4ee5\u5e2e\u52a9\u63a8\u5e7f\u80fd\u591f\u63d0\u5347\u6240\u6709\u4eba\uff08\u5c24\u5176\u662f\u670d\u52a1\u4e0d\u8db3\u4eba\u7fa4\uff09\u4eba\u7c7b\u5c0a\u4e25\u7684\u4eba\u5de5\u667a\u80fd\u6807\u51c6\u3002.\u00a0<\/p>\n<p><b>\u4ee5\u4e0b\u662f\u66f4\u5177\u4f53\u7684\u6307\u5bfc\u539f\u5219\uff1a<\/b><\/p>\n<ol>\n<li><strong>\u59cb\u7ec8\u6ce8\u660e\u4eba\u5de5\u667a\u80fd\u7684\u4f7f\u7528\uff1a<\/strong> Whenever AI assists in the creation of content to co-author a piece of writing or to\u00a0wholly create\u00a0the content, the employee must ensure that proper credit is given to the AI system or tool used. This must be done by adding a footnote or other indications\u00a0in\u00a0the document or publication.<\/li>\n<li><strong>\u786e\u4fdd\u6570\u636e\u9690\u79c1\uff1a<\/strong> PCV recognizes the importance of protecting sensitive customer data and confidential company information when feeding material into AI applications. Before using any AI tool, the employee must review the accompanying language on data security and obtain approval from the Chief Data Officer and CEO before\u00a0initiating\u00a0use (see below). PCV will ensure that\u00a0appropriate security\u00a0measures are in place, such as data anonymization, encryption, plug-in tools, and access controls (to mitigate third-party tools), to\u00a0prevent unauthorized access or use of such information.<\/li>\n<li><strong>\u51b3\u7b56\u65f6\u8981\u8c28\u614e\uff1a<\/strong> PCV recognizes that AI can automate standard processes that lead to decision-making. However, we must exercise caution and ensure that human oversight is an intentional part of the process to avoid biased or discriminatory decisions. PCV and its employees commit to regularly reviewing the AI systems and tools used to ensure their accuracy, fairness, and transparency. PCV will provide training to employees on the ethical use of AI &#8211; to ensure it is\u00a0leveraged\u00a0to inform and support human decision-making, not\u00a0displace it.<\/li>\n<li><strong>\u4e0e\u503c\u5f97\u4fe1\u8d56\u7684\u673a\u6784\u5408\u4f5c\uff1a<\/strong> PCV&#8217;s senior leadership is working to\u00a0establish\u00a0partnerships with trusted institutions, such as non-profit organizations, industry associations, and research institutions, who have\u00a0expertise\u00a0in the proper and ethical use of AI. We will regularly seek\u00a0their guidance\u00a0on ethical AI practices and incorporate their recommendations into our policies and procedures.\u202f(<em>\u4f8b\u5982\uff0cCode for America\u3001Black Wealth Data Center\u3001Data Equity \u7b49\u3002.<\/em>)<\/li>\n<li><strong>\u575a\u6301\u900f\u660e\u539f\u5219\uff1a<\/strong> PCV and its employees must ensure that the use of AI is transparent to our stakeholders,\u00a0clients,\u00a0customers, and communities. We will clearly communicate the intended use of AI systems and tools, their limitations, and any potential\u00a0risks to\u00a0stakeholders. We will also provide an avenue for feedback and concerns related to AI use.<\/li>\n<li><strong>\u81f4\u529b\u4e8e\u6301\u7eed\u6539\u8fdb\uff1a<\/strong> PCV\u5c06\u5b9a\u671f\u5ba1\u67e5\u5e76\u66f4\u65b0\u6211\u4eec\u7684\u4eba\u5de5\u667a\u80fd\u4f26\u7406\u4f7f\u7528\u653f\u7b56\uff0c\u4ee5\u53cd\u6620\u4e0d\u65ad\u53d8\u5316\u7684\u8d8b\u52bf\u3001\u6700\u4f73\u5b9e\u8df5\u548c\u65b0\u51fa\u73b0\u7684\u98ce\u9669\u3002\u6211\u4eec\u5c06\u786e\u4fdd\u6211\u4eec\u7684\u653f\u7b56\u7b26\u5408\u884c\u4e1a\u6807\u51c6\u3001\u76d1\u7ba1\u8981\u6c42\u4ee5\u53ca\u6211\u4eec\u516c\u53f8\u7684\u4ef7\u503c\u89c2\u548c\u4f7f\u547d\u3002.\u00a0<\/li>\n<li><strong>\u6d4b\u91cf\u504f\u5dee\uff1a<\/strong> PCV will use\u00a0household income, ZIP code designations, and demographic data\u00a0to measure and\u00a0mitigate\u00a0potential bias.\u00a0<\/li>\n<li><strong>\u6570\u636e\u91cd\u5851\u5de5\u4f5c\u7ec4\uff1a<\/strong> \u00a0PCV will apply the same data governance practices to\u00a0machine learning methodologies and\u00a0generative AI tools that it does with\u00a0vendor\u00a0platforms like\u00a0Lenderfit,\u00a0Downhome, and\/or\u00a0Qooper. Any concerns\u00a0regarding\u00a0bias, security, privacy, or mission drift\u00a0will be surfaced in\u00a0regularly scheduled\u00a0internal\u00a0Data Governance\u00a0Working Group\u00a0meetings\u00a0to discuss\u00a0and decide\u00a0with\u00a0internal\u00a0stakeholders.\u00a0Any PCV staff may also raise concerns directly with PCV&#8217;s Chief Data Officer and\/or CEO at any time.<\/li>\n<li><strong>Approvals for New Tools:<\/strong> Staff\u00a0field-testing new AI and ML tools need to submit a request for permission to the Chief Data Officer and CEO, clarify the use case they want to test, what the intended result or value-add will be to PCV, and how they are adhering to the above guidelines. They also need to clarify this policy with any external consultants, incorporate into external MOUs, and receive a signature of confirmation.\u00a0If new tools carry\u00a0additional\u00a0cost,\u00a0the staff must research all cost implications,\u00a0identify\u00a0the\u00a0tier\u00a0or package that is most\u00a0appropriate for\u00a0tool use, and obtain explicit permission from the CFO, CDO, and\/or CEO before\u00a0initiating\u00a0use.\u00a0Every\u00a0additional\u00a0user or license that is requested requires\u00a0approval through the same channels.\u00a0The staff that proposes\u00a0the new tool is also responsible for\u00a0identifying\u00a0any changes to this AI Policy that may arise\u00a0as a result of\u00a0using the tool. The new policy guidelines must be\u00a0discussed\u00a0preferably at the Data Governance Working Group forum, or at the very least with the Chief Data Officer for approval and inclusion in the policy.\u00a0<\/li>\n<li><strong>\u7528\u4e8e\u4f1a\u8bae\u8f6c\u5f55\u548c\u5f55\u5236\u865a\u62df\u7535\u8bdd\u4f1a\u8bae\u7684AI\u5de5\u5177\uff1a<\/strong>PCV\u00a0permits\u00a0its staff\u00a0to use a company-approved virtual call\u00a0transcriber\/recorder\u00a0<strong>\u4ec5\u6709\u7684<\/strong>\u00a0under\u00a0the following\u00a0conditions:\n<ul>\n<li>Staff must only use PCV-approved applications (for example: Fathom, as of January 2026).\u00a0<a href=\"https:\/\/fathom.video\/customize\" target=\"_blank\" rel=\"noreferrer noopener\">Fathom,<\/a>\u202fat\u00a0the time of\u00a0writing,\u00a0is\u00a0the\u00a0one\u00a0service that stands out in terms of its privacy and security features. It is Soc 2 Type II, HIPAA, CDPR, and CCPA compliant. It also has a feature that collects electronic consent from meeting attendees prior to the meeting before a recording is authorized.\u00a0If staff wish to propose an\u00a0alternate\u00a0solution, it must meet these standards at a minimum. The solution must also\u00a0be compatible with a\u00a0Microsoft\u00a0environment and\u00a0must receive explicit approval from the Chief Data Officer and CEO.\u00a0<\/li>\n<li>Staff must collect explicit consent from every\u00a0participant on the virtual call, including all internal or external stakeholders prior to\u00a0the start of recording\u00a0and\/or transcribing. If someone objects,\u00a0staff\u00a0will turn the\u202frecorder\u202foff and discuss other means to collect the information\u00a0needed.\u00a0<\/li>\n<li>Staff must disable all automated features in the recording application that\u00a0a)\u00a0involve a recorder entering a meeting without explicit\u00a0permission provided by a human\u00a0at its start, and \/or\u00a0b)\u00a0automatically sending the recording and transcripts to all meeting participants\u00a0at the close of the meeting.\u00a0The recording must be sent manually to participants who request it, and only those who attended the original meeting can\u00a0receive access to the\u00a0recording\u00a0by\u00a0PCV\u00a0staff.\u00a0\u00a0<\/li>\n<li>Staff must\u00a0maintain\u00a0the\u00a0highest level of\u00a0security of each recording\u00a0output and\u00a0treat each recording as\u00a0a PII data artifact. This means that the\u00a0staff member who initiated the recording is by default the only person who has access to the recording, and s\/he provides access to others based on the principle of least privilege, determined first and foremost by internal staff who were present at the meeting at the time of recording, and secondly by external stakeholders who were present at the meeting.\u00a0If any other person requests access to the meeting recording, PCV staff must obtain permission from their manager and\/or the Chief Data Officer\u00a0prior to sharing.\u00a0<\/li>\n<li>By default, no recorders or transcribers will be\u00a0permitted\u00a0in internal PCV calls unless determined ahead of\u00a0time and\u00a0agreed upon by all participants.\u00a0<\/li>\n<li>PCV staff must exercise caution and vigilance in calls with all external parties. If there is\u00a0a chance &#8211; however small &#8211; that sensitive information will be shared during the call, PCV staff must\u00a0request\u00a0that the counterparty deactivate its recorders and notetakers\u00a0if a mutual NDA\u00a0(MNDA)\u00a0is not already in place.\u00a0If\u00a0an MNDA is in place, the counterparty can\u00a0proceed\u00a0with the recording, provided that the PCV team clarifies that the recording will only be accessible to\u00a0participants in the call. If an MNDA is\u00a0not in place, and\u00a0if the counterparty insists\u00a0on recording, the meeting can\u00a0proceed\u00a0only with the\u00a0caveat\u00a0provided by PCV staff that no sensitive information will be shared, and\u00a0the discussion will be limited to generalities.\u00a0Sensitive information includes any client and\/or customer PII, details on PCV&#8217;s programs that are not publicly available, and any proprietary\u00a0methodology, tool, application, technology,\u00a0and\/or\u00a0process\u00a0developed by PCV.\u00a0<\/li>\n<\/ul>\n<\/li>\n<li><strong>Usage of AI tools for\u00a0Auto-Completion<\/strong><strong>:<\/strong> AI tools are increasingly becoming integrated into day-to-day applications and software such as email, Microsoft Office applications, and coding platforms. Auto-completion features, or &#8220;intelligent recommendations,&#8221; now\u00a0accompany\u00a0most applications and are intended to create efficiency. PCV employees are\u00a0permitted\u00a0to\u00a0utilize\u00a0auto-complete functionality in Word, Excel, PowerPoint, email, and other\u00a0frequently\u00a0used applications but must review the auto-completed content for accuracy. Auto-completed code (for example: while writing Python or R while performing data analysis or building algorithms) is discouraged, as the auto-generated code often suggests shortcuts, does not fully comprehend mission-based use cases, and creates a series of steps whose logic can be difficult to interpret during code reviews. If code\u00a0recommendation\u00a0tools are used, the developer must check all logic, syntax, and outputs for accuracy, AND provide comments in the code where auto-complete was used. In cases where inaccurate, misleading, or &#8220;hallucinated&#8221; content is delivered &#8211; whether to internal or external stakeholders &#8211; blaming auto-completion tools\u00a0or AI\u00a0recommendations\u00a0will not be tolerated. All PCV employees take personal responsibility for the content they as individuals, with or without\u00a0auto-completion, produce.\u00a0<\/li>\n<li><strong>Balancing AI\u00a0utilization\u00a0with\u00a0<\/strong><strong>\u6c14\u5019\u5f71\u54cd\uff1a<\/strong>\u00a0PCV staff must use AI judiciously, always\u00a0being mindful of its climate impact. A good rule of thumb is to assume\u00a0that\u00a0a typical\u00a0working session with\u00a0ChatGPT\u00a0(involving between 20-50 prompts)\u00a0uses\u00a016 ounces\u00a0of water, or the amount in a standard water bottle\u00a0one\u00a0buys\u00a0from a convenience\u00a0store. Before\u00a0a tool like ChatGPT, Claude, or other,\u00a0is\u00a0used,\u00a0staff must\u00a0always\u00a0consider\u00a0whether it is worth\u00a0essentially pouring\u00a0a bottle of water down the drain.\u00a0A few other considerations:\u00a0\n<ul>\n<li>When\u00a0PCV employees\u00a0deploy\u00a0their AI-supported tools (e.g.,\u00a0AIKKA)\u00a0and build machine learning algorithms,\u00a0they must\u00a0analyze the trade-offs of performing multiple runs versus optimizing for accuracy. For\u00a0example,\u00a0if\u00a0multiple\u00a0additional\u00a0runs\u00a0(for example, for hyperparameter tuning)\u00a0are needed to increase predictive accuracy by 3% from 92% to 95%\u00a0for a machine learning algorithm,\u00a0employees\u00a0should\u00a0not undertake\u00a0the\u00a0additional\u00a0processing for the energy it would\u00a0expend\u00a0for limited benefit.\u00a0<\/li>\n<li>When\u00a0PCV\u00a0selects\u00a0systems and technology vendors,\u00a0it will\u00a0choose those who have\u00a0demonstrated\u00a0a commitment to sustainability in their data center infrastructure.\u00a0<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n<p>By adopting this policy, PCV and its employees aim to promote the ethical use of AI and to build trust among all our stakeholders. We believe that responsible and ethical AI use\u00a0generate\u00a0value for our customers,\u00a0bring\u00a0efficiencies to our workflows, and\u00a0contribute\u00a0to the innovation, growth, and outcomes for underserved communities.\u00a0<\/p>\n<p>PCV\u5de5\u4f5c\u4eba\u5458\u5728\u5236\u5b9a\u8fd9\u9879\u653f\u7b56\u65f6\u53c2\u8003\u4e86\u4ee5\u4e0b\u6846\u67b6\u3001\u6587\u7ae0\u548c\u4e13\u5bb6\u610f\u89c1\uff1a\u00a0<\/p>\n<ul>\n<li>\u201c\u4eba\u5de5\u667a\u80fd\u4e0e\u516c\u6c11\u6743\u5229\u4ece\u6839\u672c\u4e0a\u6765\u8bf4\u662f\u4e0d\u76f8\u5bb9\u7684\u201d\uff0c\u8587\u8587\u5b89\u00b7\u660e\u535a\u58eb\uff0c\u57ce\u5e02\u7814\u7a76\u6240Socos Labs\uff0c2017\u5e74,\u00a0<a href=\"https:\/\/www.youtube.com\/watch?v=Cm7IJiokqz8\" target=\"_blank\" rel=\"noopener\">https:\/\/www.youtube.com\/watch?v=Cm7IJiokqz8<\/a><\/li>\n<li>\u201c\u8fd9\u4e0d\u662f\u5de5\u4e1a\u9769\u547d\u201d\uff0c\u8587\u8587\u5b89\u00b7\u660e\u535a\u58eb\uff0cSocos\u5b66\u9662\uff0c2018\u5e74,\u00a0<a href=\"https:\/\/academy.socos.org\/not-the-ir\/\" target=\"_blank\" rel=\"noopener\">https:\/\/academy.socos.org\/not-the-ir\/<\/a><\/li>\n<li>\u201c\u6211\u4eec\u4e0d\u9700\u8981\u4eba\u5de5\u667a\u80fd\u5ba3\u8a00\u2014\u2014\u6211\u4eec\u9700\u8981\u4e00\u90e8\u5baa\u6cd5\u201d\uff0c\u8587\u8587\u5b89\u00b7\u660e\u535a\u58eb\uff0c\u300a\u91d1\u878d\u65f6\u62a5\u300b\uff0chttps:\/\/www.ft.com\/content\/b16fab3e-7f19-49ab-9bbb-9bfeccbaf063<\/li>\n<li>&#8220;Technologist Vivienne Ming: AI is a human right&#8221;, Dr. Vivienne Ming, The Guardian, 2018, https:\/\/www.theguardian.com\/technology\/2018\/dec\/07\/technologist-vivienne-ming-ai-inequality-silicon-valley<\/li>\n<li>\u201c\u4ee5\u4eba\u4e3a\u672c\uff1a\u7528\u4eba\u6027\u5316\u7684\u65b9\u5f0f\u5efa\u7acb\u5bf9\u4eba\u5de5\u667a\u80fd\u7684\u4fe1\u4efb\u201d\uff0cSalesforce \u9053\u5fb7\u4e0e\u4eba\u9053\u4f7f\u7528\u529e\u516c\u5ba4\uff0c2024 \u5e74\uff0chttps:\/\/humanatthehelm.splashthat.com\/<\/li>\n<li>\u201c\u6570\u636e\u516c\u5e73\u6846\u67b6\u201d\uff0cWe All Count\uff0c2022\uff0chttps:\/\/weallcount.com\/the-data-process\/<\/li>\n<li>\u201c\u300a\u8d1f\u8d23\u4efb\u5730\u4f7f\u7528\u6280\u672f\u300b\uff0c\u4e16\u754c\u7ecf\u6d4e\u8bba\u575b\uff0c2019\u5e74\uff0chttps:\/\/www3.weforum.org\/docs\/WEF_Responsible_Use_of_Technology.pdf<\/li>\n<li>\u201c\u7b2c\u56db\u6b21\u5de5\u4e1a\u9769\u547d\uff1a\u5176\u610f\u4e49\u53ca\u5e94\u5bf9\u4e4b\u9053\u201d\uff0c\u514b\u52b3\u65af\u00b7\u65bd\u74e6\u5e03\uff0c\u4e16\u754c\u7ecf\u6d4e\u8bba\u575b\uff0chttps:\/\/www.weforum.org\/stories\/2016\/01\/the-fourth-industrial-revolution-what-it-means-and-how-to-respond\/<\/li>\n<li>\u201c\u300a\u4eba\u5de5\u667a\u80fd\u6743\u5229\u6cd5\u6848\u84dd\u56fe\u300b\uff0c\u767d\u5bab\uff0c2023\u5e74\uff0chttps:\/\/www.whitehouse.gov\/ostp\/ai-bill-of-rights\/<\/li>\n<li>\u201c\u4eba\u5de5\u667a\u80fd\u98ce\u9669\u7ba1\u7406\u6846\u67b6\u201d\uff0c\u7f8e\u56fd\u56fd\u5bb6\u6807\u51c6\u4e0e\u6280\u672f\u7814\u7a76\u9662\uff0c2023 \u5e74\uff0chttps:\/\/www.nist.gov\/itl\/ai-risk-management-framework<\/li>\n<li>\u201c\u5fae\u8f6f\u63a8\u8fdb\u8d1f\u8d23\u4efb\u7684\u4eba\u5de5\u667a\u80fd\u521b\u65b0\u7684\u81ea\u613f\u627f\u8bfa\u201d\uff0c\u5fae\u8f6f\uff0c2023 \u5e74\uff0chttps:\/\/blogs.microsoft.com\/on-the-issues\/2023\/07\/21\/commitment-safe-secure-ai\/<\/li>\n<li>https:\/\/www.ie.edu\/insights\/articles\/from-cloud-to-cup-how-much-water-does-your-chatgpt-drink\/\u00a0<\/li>\n<\/ul>","protected":false},"excerpt":{"rendered":"<p>\u5e38\u89c1\u95ee\u9898\u89e3\u7b54 \u592a\u5e73\u6d0b\u793e\u533a\u98ce\u9669\u6295\u8d44\u516c\u53f8 (PCV) \u5173\u4e8e\u8d1f\u8d23\u4efb\u4e14\u5408\u4e4e\u9053\u5fb7\u5730\u4f7f\u7528\u4eba\u5de5\u667a\u80fd\u548c\u673a\u5668\u5b66\u4e60\u7684\u653f\u7b56 2026 \u5e74 3 \u6708 PCV \u7684\u5ba2\u6237\u3001\u5408\u4f5c\u4f19\u4f34\u3001\u540c\u4e8b\u3001\u884c\u4e1a\u4f19\u4f34\u548c\u652f\u6301\u8005\uff1a \u4eba\u7c7b\u6587\u660e\u5df2\u8fdb\u5165\u4e00\u4e2a\u65b0\u65f6\u4ee3\u2014\u2014\u7b2c\u56db\u6b21\u5de5\u4e1a\u9769\u547d\uff0c\u4eba\u5de5\u667a\u80fd (AI) \u548c\u673a\u5668\u5b66\u4e60 (ML) \u4e3a\u6280\u672f\u521b\u65b0\u5e26\u6765\u4e86\u5f3a\u5927\u7684\u529b\u91cf\u548c\u6f5c\u529b\uff0c\u80fd\u591f\u7b80\u5316\u6211\u4eec\u7684\u5de5\u4f5c\u6d41\u7a0b\u2026\u2026<\/p>","protected":false},"author":61,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"_acf_changed":false,"footnotes":""},"class_list":["post-28976","page","type-page","status-publish","hentry"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.pacificcommunityventures.org\/zh\/wp-json\/wp\/v2\/pages\/28976","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.pacificcommunityventures.org\/zh\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/www.pacificcommunityventures.org\/zh\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/www.pacificcommunityventures.org\/zh\/wp-json\/wp\/v2\/users\/61"}],"replies":[{"embeddable":true,"href":"https:\/\/www.pacificcommunityventures.org\/zh\/wp-json\/wp\/v2\/comments?post=28976"}],"version-history":[{"count":4,"href":"https:\/\/www.pacificcommunityventures.org\/zh\/wp-json\/wp\/v2\/pages\/28976\/revisions"}],"predecessor-version":[{"id":29890,"href":"https:\/\/www.pacificcommunityventures.org\/zh\/wp-json\/wp\/v2\/pages\/28976\/revisions\/29890"}],"wp:attachment":[{"href":"https:\/\/www.pacificcommunityventures.org\/zh\/wp-json\/wp\/v2\/media?parent=28976"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}