{"id":185,"date":"2026-03-27T14:46:38","date_gmt":"2026-03-27T14:46:38","guid":{"rendered":"https:\/\/dev.www.purdue.edu\/ai\/?page_id=185"},"modified":"2026-04-08T19:28:41","modified_gmt":"2026-04-08T19:28:41","slug":"submitting-ideas-for-ai-use","status":"publish","type":"page","link":"https:\/\/www.purdue.edu\/ai\/ai-governance-and-review\/submitting-ideas-for-ai-use\/","title":{"rendered":"Submitting Ideas for AI Use\u00a0"},"content":{"rendered":"<div  class=\"purdue-home-cta-banner purdue-home-teritary-hero \">\n        <div class=\"section has-gold-background has-padding-top-large has-padding-bottom-large \">\n        <div class=\"container\">\n            <h1 class=\"second-level-page-heading purdue-home-teritary-hero__header\">Submitting\u00a0ideas for AI use<\/h1>\n            <div class=\"purdue-home-teritary-hero__content\">\n                \n\n<p class=\"wp-block-paragraph\">Using the best practices provided here will guide your decisions on possible AI use.<\/p>\n\n            <\/div>\n                    <\/div>\n    <\/div>\n<\/div>\n\n\n<div  class=\"section \">\n    <div class=\"container\">\n                \n\n<div class=\"wp-block-columns page-layout-columns columns is-multiline is-layout-flex wp-container-core-columns-is-layout-8f761849 wp-block-columns-is-layout-flex\">\n<div class=\"wp-block-column column is-full-tablet page-layout-main is-layout-flow wp-block-column-is-layout-flow\">\n<div  class=\"purdue-home-intro-text has-layout-one-centered \">\n        <div class=\"section has-none-background has-padding-top-none has-no-sidepadding\">\n        <div class=\"container\">\n            \n            <h2 class=\"purdue-home-intro-text__header header-font-acumin\">Best Practices Expedite AI\u00a0Decisions<\/h2>\n\n<p class=\"wp-block-paragraph\">Purdue uses the <a href=\"https:\/\/purdue0.sharepoint.com\/sites\/ExecutiveDataCouncil-EthicalandResponsibleDataUse\/_layouts\/15\/listforms.aspx?cid=Nzc2YTc4MzEtOTlhMy00ZGM4LWE3NGUtNTc2Zjc3NWNhZDJk&amp;nav=YjQ2MmQyMWYtNGFjOC00YThkLWE4N2QtNjBmYzY5MjVkZjll\" target=\"_blank\" rel=\"noreferrer noopener\">Data Ethics Committee Use Case Review form<\/a> to review your proposed use case for AI or data. The committee works to understand the scope, risks and potential impact of your project. This review process also serves as a tool to help submitters understand the risks associated with their use cases, so they can ensure that appropriate governance, oversight and safety protocols are in place. This process also supports conversations with vendors and guides additional information gathering.<\/p>\n\n\n                    <\/div>\n    <\/div>\n<\/div>\n\n\n\n<div class=\"purdue-accordion-wrap  purdue-accordion-wrap--small\">\n    <div class=\"section has-none-background has-padding-top-none has-padding-bottom-none has-no-sidepadding\">\n        <div class=\"container\">\n            <div class=\"purdue-accordion has-no-bottomborder\">\n                <h3 class=\"accordion__heading purdue-accordion__title  purdue-accordion__title--small\" id=\"title-641d9177-f4f0-46db-bf9b-5c94f95f2913\">\n                    <button aria-controls=\"content-641d9177-f4f0-46db-bf9b-5c94f95f2913\" aria-expanded=\"false\">General Best Practices for Completing This Form<strong>\u00a0<\/strong><\/button>\n                <\/h3>\n                <div class=\"accordion__content purdue-accordion__content\" id=\"content-641d9177-f4f0-46db-bf9b-5c94f95f2913\" hidden aria-hidden=\n                \"true\">\n                    \n\n<h4 class=\"wp-block-heading\">Be specific<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Avoid generic or unsupported claims like \u201cwe conduct fairness testing\u201d or \u201cno sensitive data is used.\u201d Instead, describe <em>how<\/em> you tested for fairness, e.g., \u201ctested against three demographic subgroups\u201d or <em>why<\/em> the data is not sensitive, e.g., \u201call PII fields are hashed before uploading.\u201d  <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Provide sufficient Information to enable evaluation<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Similarly, please include enough high-level information directly in the form or provide details on how to use any attached documentation to answer a question. Any answers lacking sufficient detail will receive follow-up questions. The committee meets monthly to review use cases, meaning unclear answers resulting in back-and-forth with the committee may delay approval by as much as a month. <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Articulate your strategy<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">If a formal procedure doesn\u2019t exist yet, describe your planned approach. It is acceptable to say, \u201cWe do not have a fixed metric yet, so we will conduct a manual review of 50 outputs every quarter to establish a baseline and share the results with the committee.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Explain to a broader audience<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Write for a generalist audience, and assume the readers do not have deep familiarity with your work in context, like the relevant job activities, workflow, existing organizational safeguards, etc. If you must use technical or job-specific concepts (e.g., RAG, Financial Reporting Statement, our office\u2019s admissions routing procedure), briefly explain them in context. <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">Do not rely on \u2018magical\u2019 or generic oversight: <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Simply stating \u201cthere is a human in the loop\u201d or \u201cthe vendor ensures robust testing\u201d is not enough. You must describe the human\u2019s specific role, or the specific vendor control you have verified. <\/p>\n\n                <\/div>\n            <\/div>\n        <\/div>\n    <\/div>\n<\/div>\n\n\n\n<div class=\"purdue-accordion-wrap  purdue-accordion-wrap--small\">\n    <div class=\"section has-none-background has-padding-top-none has-padding-bottom-none has-no-sidepadding\">\n        <div class=\"container\">\n            <div class=\"purdue-accordion has-no-bottomborder\">\n                <h3 class=\"accordion__heading purdue-accordion__title  purdue-accordion__title--small\" id=\"title-21cd1ad0-3fc4-48c4-83c3-b3a40e691df9\">\n                    <button aria-controls=\"content-21cd1ad0-3fc4-48c4-83c3-b3a40e691df9\" aria-expanded=\"false\">Section\u00a01: Purpose and Scope<strong>\u00a0<\/strong><\/button>\n                <\/h3>\n                <div class=\"accordion__content purdue-accordion__content\" id=\"content-21cd1ad0-3fc4-48c4-83c3-b3a40e691df9\" hidden aria-hidden=\n                \"true\">\n                    \n\n<h4 class=\"wp-block-heading\">1.1. What specific services or functionalities does your AI product, service, or tool provide? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Core use cases and the specific problem being solved <\/li>\n\n\n\n<li>Intended user groups (students, faculty, admin staff, public) <\/li>\n\n\n\n<li>Technical dependencies (APIs, external data sources, cloud platforms) <\/li>\n\n\n\n<li>Workflow descriptions for nonspecialists <\/li>\n\n\n\n<li>Operational diagrams <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response<\/strong> (relies on generic claims): \u201cThe tool automates routine communication to increase office productivity. It uses advanced AI to draft responses, but a human always reviews them to ensure quality.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response <\/strong>(specific and clear): \u201cWe are using Microsoft Copilot (Enterprise license) to summarize transcripts from internal administrative meetings and draft follow-up emails. It connects to our existing Office 365 tenant. No data leaves the institutional boundary during processing. Data is stored on our SharePoint and only available to members of the HR team.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">1.2. What are the explicit limitations and delimitations of your service\u2019s capabilities? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u201cNot for use in\u2026\u201d disclaimers or documented boundaries <\/li>\n\n\n\n<li>Specific contexts where use would be inappropriate or dangerous <\/li>\n\n\n\n<li>Known technical limitations (e.g., knowledge cutoffs, hallucination rates) <br> <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cThe AI is generally accurate, but users are reminded that it can make mistakes and they should use their best judgment.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cThis tool should not be used for grading student work, generating letters of recommendation or interpreting medical leave requests. It has a knowledge cutoff of 2023 and cannot access real-time student SIS data.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong respons<\/strong><strong>e:<\/strong> \u201cDuring testing, the AI regularly confused the specific ancestries of Asian students \u2014 for instance, confusing Japan and Korea. For this reason, we have explicitly shut off all recommendations from the tool tagged with \u2018nation of origin\u2019 as an important reason for the recommendation.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">1.3. What ethical, regulatory, or quality guidelines did you follow?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Alignment with frameworks (NIST AI RMF, OECD AI Principles, EU AI Act) <\/li>\n\n\n\n<li>Compliance with specific regulations (GDPR, HIPAA, FERPA) <\/li>\n\n\n\n<li>Results from internal IRB reviews or third-party audits <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cWe follow all standard ethical AI principles and the vendor\u2019s trust and safety guidelines to prevent misuse.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cWe have confirmed the tool is HIPAA-compliant via a BAA with the vendor, and we applied the NIST AI RMF; a report is attached. An associated pilot project was reviewed by the IRB (Protocol #12345) and deemed exempt.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">1.4. What foreseeable risks or harms could arise from using this service? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Anticipated risks (hallucinations, bias, data leakage) <\/li>\n\n\n\n<li>Risk categories (reputational, financial, individual or societal harm) <\/li>\n\n\n\n<li>Results from impact assessments (ethical, safety, human rights) <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cThe risk is minimal because staff are required to check the output. The vendor also has filters in place to stop harmful content.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cPrimary risks include \u2018hallucination\u2019 of policy details in draft emails, which could misinform students. There is also a risk of non-native English speakers receiving lower-quality summaries. We mitigate this by mandatory manual review (see Section 4.1), a disclaimer on all outputs, prohibitions on certain uses and an outreach number for harms that may occur.\u201d <\/p>\n\n                <\/div>\n            <\/div>\n        <\/div>\n    <\/div>\n<\/div>\n\n\n\n<div class=\"purdue-accordion-wrap  purdue-accordion-wrap--small\">\n    <div class=\"section has-none-background has-padding-top-none has-padding-bottom-none has-no-sidepadding\">\n        <div class=\"container\">\n            <div class=\"purdue-accordion has-no-bottomborder\">\n                <h3 class=\"accordion__heading purdue-accordion__title  purdue-accordion__title--small\" id=\"title-5c0ccd3c-ba7b-4138-b221-28043556da55\">\n                    <button aria-controls=\"content-5c0ccd3c-ba7b-4138-b221-28043556da55\" aria-expanded=\"false\">Section 2: Data and Privacy<strong>\u00a0<\/strong><\/button>\n                <\/h3>\n                <div class=\"accordion__content purdue-accordion__content\" id=\"content-5c0ccd3c-ba7b-4138-b221-28043556da55\" hidden aria-hidden=\n                \"true\">\n                    \n\n<h4 class=\"wp-block-heading\">2.1. What data sources were used to train and test your AI service?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Dataset size, quality and provenance (e.g., open source vs. scraped) <\/li>\n\n\n\n<li>Distinction between pre-training data and fine-tuning\/RAG data. <\/li>\n\n\n\n<li>Licenses, ownership status, or datasheets.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response: <\/strong>\u201cThe model is trained on a massive, diverse dataset curated by OpenAI to ensure high-quality responses across many topics.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cWe utilize the base GPT-4 model (pretrained by OpenAI on public data). We also use retrieval-augmented generation (RAG), giving the RAG access to our internal departmental policy PDFs (approximately 500 documents). No student data is used to train or fine-tune the model.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">2.2. How representative or diverse is the data, and how do you mitigate bias?<strong> <\/strong><\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Demographic coverage and subgroup analysis <\/li>\n\n\n\n<li>Steps taken to identify bias (fairness audits, red-teaming) <\/li>\n\n\n\n<li>Citations of academic studies if relevant<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cThe vendor has removed bias from the model using RLHF (reinforcement learning from human feedback), so it treats all users equally.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cThe underlying model has known biases regarding gender roles. For our specific use case (technical support chat), we audited 100 interactions and found no significant performance difference between queries phrased in standard vs. nonstandard English.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">2.3. What data do you collect or log? Does it include PII? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Logs of inputs\/outputs, metadata and retention periods <\/li>\n\n\n\n<li>Transparency reports or data protection impact assessments <\/li>\n\n\n\n<li>Whether PII or sensitive attributes are solicited or stored <br> <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cWe log interactions to help improve the user experience, but we protect user privacy according to our standard privacy policy.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cWe log all user prompts and system outputs for 30 days for debugging, then permanently delete them. IP addresses are anonymized. No student PII is solicited, but if a user accidentally enters it, it resides in the logs until the 30-day purge.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">2.4. How do you protect the privacy and security of sensitive information? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Access controls (SSO, MFA) and encryption methods (transit\/rest) <\/li>\n\n\n\n<li>Adherence to standards (ISO 27001, NIST Cybersecurity Framework, SOC 2) <\/li>\n\n\n\n<li>Anonymization or differential privacy techniques <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cWe use a secure cloud platform that is trusted by major enterprises and encrypts data.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cData is encrypted in transit (TLS 1.3) and at rest (AES-256). Access is restricted via single sign-on to authorized staff only. The vendor provides a SOC 2 Type II report verifying these controls. Data does not leave the U.S.-East region.\u201d <\/p>\n\n                <\/div>\n            <\/div>\n        <\/div>\n    <\/div>\n<\/div>\n\n\n\n<div class=\"purdue-accordion-wrap  purdue-accordion-wrap--small\">\n    <div class=\"section has-none-background has-padding-top-none has-padding-bottom-none has-no-sidepadding\">\n        <div class=\"container\">\n            <div class=\"purdue-accordion has-no-bottomborder\">\n                <h3 class=\"accordion__heading purdue-accordion__title  purdue-accordion__title--small\" id=\"title-b0336a7e-0a88-4eef-b5c1-d25d9718a0ff\">\n                    <button aria-controls=\"content-b0336a7e-0a88-4eef-b5c1-d25d9718a0ff\" aria-expanded=\"false\">Section 3: Performance and Safety<\/button>\n                <\/h3>\n                <div class=\"accordion__content purdue-accordion__content\" id=\"content-b0336a7e-0a88-4eef-b5c1-d25d9718a0ff\" hidden aria-hidden=\n                \"true\">\n                    \n\n<h4 class=\"wp-block-heading\">3.1. How do you evaluate the performance of your AI service? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Specific metrics (accuracy, recall, time saved, cost reduction) <\/li>\n\n\n\n<li>Benchmarking results against human performance <\/li>\n\n\n\n<li>Pilot deployment data or peer-reviewed studies <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cThe tool has been very effective in our tests and users are happy with the quality of the writing.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cWe conducted a pilot where the AI summarized 50 past meetings. Staff rated accuracy at 4.5\/5. The tool reduced draft time from 30 minutes to five minutes per meeting. We track \u2018accepted vs. edited\u2019 rates for generated text.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">3.2. How do you evaluate the safety of the system (risks, errors, harms)? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Error rates and adverse event tracking <\/li>\n\n\n\n<li>Scenario testing or \u201cred-teaming\u201d (trying to break the model) <\/li>\n\n\n\n<li>Continuous monitoring plans after deployment <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cWe rely on the underlying model\u2019s built-in safety features, which prevent it from generating harmful or illegal content.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cWe utilized \u2018red teaming,\u2019 where staff intentionally tried to trick the bot into revealing confidential salary data. It failed two out of 10 times; we adjusted the system prompt to fix this before release. We monitor logs weekly for \u2018jailbreak\u2019 attempts.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">3.3. How do you make sure the service works for new groups or contexts? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Testing in varied settings or demographics <\/li>\n\n\n\n<li>Procedures for local validation or calibration before expansion <\/li>\n\n\n\n<li>Portability assessments <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cBecause the model is trained on general knowledge, it should work well for any department.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cCurrently tested only in the engineering department. Before expanding to the humanities department, we will run a new validation set to ensure the model understands their specific terminology and citation styles.\u201d <\/p>\n\n                <\/div>\n            <\/div>\n        <\/div>\n    <\/div>\n<\/div>\n\n\n\n<div class=\"purdue-accordion-wrap  purdue-accordion-wrap--small\">\n    <div class=\"section has-none-background has-padding-top-none has-padding-bottom-none has-no-sidepadding\">\n        <div class=\"container\">\n            <div class=\"purdue-accordion has-no-bottomborder\">\n                <h3 class=\"accordion__heading purdue-accordion__title  purdue-accordion__title--small\" id=\"title-25334fd6-0aa2-4ff2-b4f4-4460b3bcecb2\">\n                    <button aria-controls=\"content-25334fd6-0aa2-4ff2-b4f4-4460b3bcecb2\" aria-expanded=\"false\">Section 4: Transparency, Human Oversight\u00a0and Accountability<strong>\u00a0<\/strong><\/button>\n                <\/h3>\n                <div class=\"accordion__content purdue-accordion__content\" id=\"content-25334fd6-0aa2-4ff2-b4f4-4460b3bcecb2\" hidden aria-hidden=\n                \"true\">\n                    \n\n<h4 class=\"wp-block-heading\">4.1. How are humans involved in the use of this service?<\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Specific roles (frontline users, reviewers, auditors) <\/li>\n\n\n\n<li>Workflow steps (e.g., \u201chuman-in-the-loop\u201d vs. \u201chuman-on-the-loop\u201d) <\/li>\n\n\n\n<li>Frequency or ratio of human interventions<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cWe maintain a human-in-the-loop approach where staff oversee the AI to prevent any issues.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cThe AI generates a draft email, but the \u2018Send\u2019 button is disabled until a staff member edits the text and manually checks a \u2018Reviewed\u2019 box. Supervisors audit 10% of sent emails monthly.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">4.2. How do you communicate capabilities and risks to users? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>User guides, FAQs, model cards or system cards <\/li>\n\n\n\n<li>Plain-language disclaimers and uncertainty indicators <\/li>\n\n\n\n<li>Transparency notices distinguishing AI from humans <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cUsers are informed during onboarding that they are using an AI tool.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cThe chat interface includes a permanent banner stating: \u2018AI can make mistakes. Verify important info.\u2019 All generated documents include a watermark or footer indicating AI origin.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">4.3. What training or qualifications are required for users? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Required certifications or professional qualifications <\/li>\n\n\n\n<li>AI ethics or safety training modules <\/li>\n\n\n\n<li>Onboarding curriculum content <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cOur staff are highly trained professionals who know how to evaluate information.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cAll users must complete the \u2018Data Privacy and Generative AI\u2019 30-minute learning module before being granted access. Supervisors receive additional training on spotting hallucinated citations.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">4.4. What processes exist for escalation or overriding outputs? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Escalation triggers and supervisory approval steps <\/li>\n\n\n\n<li>Emergency stop mechanisms <\/li>\n\n\n\n<li>Override procedures <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cIf the AI produces something wrong, the user can simply ignore it and write the text themselves.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cUsers can flag a response as \u2018harmful\u2019 or \u2018inaccurate\u2019 directly in the UI. This triggers a review by the IT team. If the error rate exceeds 5% in a week, the system is automatically paused for maintenance.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">4.5. Who is responsible if the AI makes a mistake? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Assignment of responsibility (project lead, department head, vendor) <\/li>\n\n\n\n<li>Complaint hotlines or feedback portals <\/li>\n\n\n\n<li>Incident response procedures <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cWe work closely with the vendor to ensure that any errors are corrected in future updates.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cThe director of admissions retains final responsibility for all decisions, regardless of AI input. Complaints regarding AI interactions can be routed through the standard student grievance portal, with a checkbox for \u2018Technical\/AI Issue.\u2019\u201d <\/p>\n\n                <\/div>\n            <\/div>\n        <\/div>\n    <\/div>\n<\/div>\n\n\n\n<div class=\"purdue-accordion-wrap  purdue-accordion-wrap--small\">\n    <div class=\"section has-none-background has-padding-top-none has-padding-bottom-none has-no-sidepadding\">\n        <div class=\"container\">\n            <div class=\"purdue-accordion has-no-bottomborder\">\n                <h3 class=\"accordion__heading purdue-accordion__title  purdue-accordion__title--small\" id=\"title-c2b238e5-0b6c-4182-85c2-216a395fee09\">\n                    <button aria-controls=\"content-c2b238e5-0b6c-4182-85c2-216a395fee09\" aria-expanded=\"false\">Section 5: Deployment, Maintenance\u00a0and Support<strong>\u00a0<\/strong><\/button>\n                <\/h3>\n                <div class=\"accordion__content purdue-accordion__content\" id=\"content-c2b238e5-0b6c-4182-85c2-216a395fee09\" hidden aria-hidden=\n                \"true\">\n                    \n\n<h4 class=\"wp-block-heading\">5.1. What technical support and documentation are available? <\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>User manuals, admin runbooks and troubleshooting guides <\/li>\n\n\n\n<li>Help desk contacts and support hours<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cSupport is available through our standard IT ticketing system.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cWe have created a dedicated Confluence page with FAQs and a \u2018Prompt Engineering Guide\u2019 for staff. Level 1 support is handled by the departmental IT lead; Level 2 is escalated to the vendor.\u201d <\/p>\n\n\n\n<h4 class=\"wp-block-heading\">5.2. How do you plan to update and improve the service?<strong> <\/strong><\/h4>\n\n\n\n<p class=\"wp-block-paragraph\">Examples of information to provide: <\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Versioning policies and road maps <\/li>\n\n\n\n<li>Monitoring for regulatory changes <\/li>\n\n\n\n<li>Update frequency and postrelease monitoring <\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Weak response:<\/strong> \u201cThe system is cloud-based so it is automatically updated with the latest features.\u201d <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Strong response:<\/strong> \u201cWe review the system performance quarterly. We subscribe to the vendor\u2019s changelog to monitor for model updates (e.g., GPT-4 to GPT-5) and will freeze our current version until the new model passes our internal validation test.\u201d <\/p>\n\n                <\/div>\n            <\/div>\n        <\/div>\n    <\/div>\n<\/div>\n\n\n\n<p class=\"wp-block-paragraph\"><\/p>\n<\/div>\n\n\n\n<div class=\"wp-block-column column is-one-quarter-desktop is-full-tablet is-full-mobile page-layout-sidebar is-layout-flow wp-block-column-is-layout-flow\">\n<p class=\"wp-block-paragraph\"><\/p>\n<\/div>\n<\/div>\n\n    <\/div>\n<\/div>\n\n\n<p class=\"wp-block-paragraph\"><\/p>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":2,"featured_media":0,"parent":182,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"_acf_changed":false,"footnotes":""},"class_list":["post-185","page","type-page","status-publish","hentry"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.purdue.edu\/ai\/wp-json\/wp\/v2\/pages\/185","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.purdue.edu\/ai\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/www.purdue.edu\/ai\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/www.purdue.edu\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.purdue.edu\/ai\/wp-json\/wp\/v2\/comments?post=185"}],"version-history":[{"count":30,"href":"https:\/\/www.purdue.edu\/ai\/wp-json\/wp\/v2\/pages\/185\/revisions"}],"predecessor-version":[{"id":528,"href":"https:\/\/www.purdue.edu\/ai\/wp-json\/wp\/v2\/pages\/185\/revisions\/528"}],"up":[{"embeddable":true,"href":"https:\/\/www.purdue.edu\/ai\/wp-json\/wp\/v2\/pages\/182"}],"wp:attachment":[{"href":"https:\/\/www.purdue.edu\/ai\/wp-json\/wp\/v2\/media?parent=185"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}