Complete dissertation appendix
This complete applied case preserves the project scenario, governance, assessment, solution analysis, implementation guidance, safeguards, measures, risks, sustainability provisions, and decision checklist developed in the dissertation.
Applying the Dissertation Conceptual Model to an Emerging AI-Driven Digital Divide
Project Context: AI-enabled tools are increasingly embedded in education, employment, public services, communication, and everyday decision-making. Many people may already possess broadband and capable devices, yet differ substantially in their ability to use AI effectively, evaluate outputs critically, recognize misinformation or bias, protect sensitive information, and apply human judgment responsibly.
This report applies the Closing of the Digital Divide conceptual model to an emerging divide driven by the rapid diffusion of artificial intelligence. The community may already have broadband access, connected devices, learning platforms, workplace applications, and digital services, yet people differ sharply in their ability to understand and use AI-enabled systems. The visible technology is widely available, but meaningful participation increasingly depends on a new layer of digital skills and literacy.
The scenario includes generative, conversational, predictive, assistive, and algorithmic tools used for writing, research, tutoring, translation, coding, job search, productivity, customer service, public information, decision support, and content creation. The divide appears when some users can formulate productive questions, iterate with the system, verify claims, recognize limitations, protect data, detect manipulation, and incorporate AI into real tasks, while others can only accept or reject outputs at face value.
The dissertation codebook explicitly identifies AI as a new layer of the digital divide, including the need for prompt literacy, new digital-literacy frameworks, safeguards against bias, and responsible use. The conceptual model, therefore, treats AI not only as a technology opportunity but also as a potential amplifier of existing differences in education, language, disability, income, age, occupational opportunity, institutional capacity, and digital self-efficacy.
The initiative is not designed to make every participant an AI engineer. Its purpose is to establish a practical level of AI literacy appropriate to a person’s learning, work, civic, and daily life contexts. Participants should know what AI can and cannot do, how to use it productively, when to verify it, what information should not be shared, how bias and misinformation can enter outputs, and when a human decision-maker must remain accountable.
Program success statement: The initiative succeeds when people who were previously disadvantaged by limited AI knowledge can use AI-enabled tools productively and safely, critically evaluate generated information, protect themselves and others from foreseeable harms, and participate more fully in education, employment, public services, and community life without becoming dependent on unverified automated output.
This case treats AI literacy as an emerging dimension of digital inclusion. Traditional digital-skills programs often focus on device use, internet navigation, productivity applications, cybersecurity basics, and access to online services. Those remain necessary, but they are no longer sufficient as AI systems become embedded in search, writing, learning, hiring, customer service, workplace software, public information, and decision support.
The conceptual model requires practitioners to diagnose the divide before prescribing training. A participant who cannot use AI effectively may be experiencing a foundational digital-skills gap, a lack of access to suitable tools, weak language or accessibility support, low confidence, unclear organizational policy, privacy concerns, poor critical information literacy, or a lack of opportunity to practice on authentic tasks. A training course alone will not solve every cause.
The recommended approach is a coordinated AI-literacy and digital-skills initiative built around six principles:
Preserve foundational digital literacy as the base layer rather than replacing it with prompt training.
Teach AI use together with verification, critical thinking, source evaluation, privacy, security, bias awareness, accessibility, and human accountability.
Adapt instruction to developmental, educational, occupational, language, disability, and community contexts.
Provide hands-on access to appropriate tools, support, and realistic tasks rather than relying on lecture-only awareness training.
Measure demonstrated capability and safe judgment, not merely attendance or tool usage.
Use a continuous feedback loop because AI systems, institutional policies, workforce expectations, and risks will continue to change.
The preferred program is therefore not a single AI workshop. It is an adaptive skills-and-literacy system linking access, education, workforce preparation, responsible-use governance, trusted support, and evidence-based improvement.
The purpose of the initiative is to reduce emerging inequities in the ability to understand, use, evaluate, and govern AI-enabled tools. The project focuses on practical participation rather than technical specialization and should complement existing digital literacy, education, workforce, library, and community support programs.
The initiative will address seven connected needs:
AI access. Can participants reach an appropriate AI-enabled tool through a suitable device, account, language, and accessibility pathway?
Foundational digital skills. Can participants manage files, accounts, browsers, documents, security basics, and online information well enough to use AI productively?
Interaction literacy. Can participants define a task, provide useful context, ask follow-up questions, compare outputs, and refine an interaction?
Critical literacy. Can participants recognize uncertainty, hallucinated information, bias, persuasive manipulation, deepfakes, and unsupported claims?
Responsible use. Do participants understand privacy, sensitive data handling, attribution, academic or workplace rules, and when human review is required?
Applied capability. Can participants use AI to improve an authentic learning, work, civic, or daily-life task rather than simply generate content?
Sustainability. Can institutions keep curricula, policies, staff skills, access to tools, and support current as AI systems evolve?
| The program can help address | The program cannot solve itself |
|---|---|
| Unequal AI knowledge, confidence, critical evaluation, and task-use skills | All structural inequities in education, employment, language access, disability access, or economic opportunity |
| Unsafe or ineffective use caused by limited literacy and support | Bias or harmful design choices embedded in external AI systems that the community does not control |
| Gaps in institutional guidance, educator readiness, and responsible-use practices | Every legal, ethical, labor, copyright, or regulatory question created by AI |
| Unequal access to guided practice, trusted coaching, and appropriate tools | The underlying quality of data, models, vendors, or automated decisions without separate governance and procurement controls |
| Weak verification habits and difficulty distinguishing generated content from reliable evidence | The need for human expertise, subject knowledge, professional judgment, and accountable decision-making |
Eligibility and outreach should not assume that age, education, income, or occupation alone determine AI literacy. People with advanced academic or professional credentials may have limited experience with AI, while some younger or less formally educated users may be highly capable. The program should segment participants by demonstrated need, context, and risk rather than by stereotype.
| Population or context | Likely AI-literacy need | Program implication |
|---|---|---|
| K-12 and college students | Use AI without weakening learning, source evaluation, authorship, or academic integrity | Teach age-appropriate AI literacy, verification, citation, task boundaries, and reflection on when AI should or should not be used |
| Educators and instructional staff | Design learning and assessment in an environment where AI is available | Provide professional development on responsible use, assignment design, verification, accessibility, and student guidance |
| Adult learners and job seekers | Use AI for resumes, applications, learning, research, and skill development without accepting unreliable output | Use authentic workforce tasks, coaching, privacy guidance, and evidence-checking routines |
| Current workers and supervisors | Integrate AI into workflows while protecting quality, confidentiality, and accountability | Provide role-specific training, human-review standards, data rules, and change-management support |
| Older adults and late digital adopters | Build confidence while avoiding scams, deepfakes, misinformation, and overreliance | Use paced instruction, trusted facilitators, repetition, fraud awareness, and practical daily-life examples |
| People with disabilities | Use assistive AI while ensuring accessibility, accuracy, autonomy, and privacy | Test accessible interfaces and provide compatible assistive technology, alternate formats, and individualized support |
| Multilingual and limited-English-proficiency users | Use translation and language tools critically without assuming perfect accuracy | Provide multilingual instruction and verification practices for high-consequence translation |
| Small businesses and entrepreneurs | Apply AI to marketing, administration, analysis, and customer service while protecting data and brand trust | Teach practical productivity use, disclosure, verification, privacy, and customer-facing safeguards |
| Community and public-service users | Understand AI-mediated information, forms, eligibility tools, and public communication | Provide civic-information literacy, trusted assistance, appeal pathways, and human alternatives when automated processes fail |
Power-weighting reminder: People with the least experience using AI may be most affected by institutional decisions to adopt it. Their limited familiarity should not be interpreted as consent, lack of interest, or inability. Program design should create meaningful opportunities for those users to influence training, policies, support, and safeguards.
A cross-domain governance group is needed to define the problem, conduct a baseline assessment, design the curriculum, select tools, establish safeguards, plan training, evaluate, and support continuous improvement. Because AI affects education, work, information, privacy, accessibility, and institutional policy simultaneously, ownership should not rest solely with an IT department or a single training unit.
| Role | Primary responsibility |
|---|---|
| Executive sponsor or institutional leader | Defines mission, authorizes resources, resolves policy conflicts, and keeps equity and public-interest goals visible. |
| Program manager | Coordinates assessment, curriculum, partnerships, scheduling, staff, reporting, support, and improvement work. |
| AI literacy/curriculum lead | Defines competencies, learning objectives, instructional sequence, practice activities, and assessment methods. |
| Technology and AI platform lead | Evaluates tool access, licensing, configurations, interoperability, account management, and technical support. |
| Privacy, cybersecurity, and data governance lead | Defines approved data practices, incident response, security expectations, and vendor/data safeguards. |
| Accessibility and inclusion lead | Tests accessibility, language access, assistive-technology compatibility, and accommodations. |
| Evaluation and data lead | Builds the baseline, measures skills and outcomes, protects participant data, and reports disparities and progress. |
| Community/learner/worker representatives | Provide experience-based feedback on usefulness, barriers, trust, risk, language, support, and real-world application. |
The conceptual model requires that stakeholders be considered in terms of role, influence, power, community voice, and external conditions. In this case, stakeholders include both people affected by AI systems and institutions that purchase, teach, regulate, design, deploy, or evaluate them.
| Stakeholder | Project-specific role |
|---|---|
| Learners and students | Participate in baseline assessment and training; test whether AI supports learning without replacing reasoning, authorship, or subject mastery. |
| Parents and guardians | Support younger learners, understand school expectations, and provide feedback on privacy, safety, homework, and family impact. |
| Adult learners and job seekers | Test workforce and life-use modules and identify practical barriers to applying AI in employment and skill development. |
| Workers and frontline employees | Identify real workflow opportunities, quality risks, data concerns, and training requirements. |
| Supervisors and managers | Define acceptable workplace use, human-review expectations, job redesign, and accountability. |
| Teachers and faculty | Teach subject-specific AI literacy, redesign assignments and assessments, model verification, and identify learning impacts. |
| Instructional designers and curriculum developers | Embed AI literacy, critical thinking, accessibility, and responsible use across curricula rather than isolating them in one course. |
| Librarians and media/information-literacy staff | Teach source evaluation, verification, misinformation detection, citation practices, and public access to trusted support. |
| Digital navigators and community trainers | Provide low-barrier, individualized coaching for people who need foundational or confidence-building support. |
| School and college administrators | Align policy, professional development, student support, procurement, assessment, and institutional accountability. |
| Workforce agencies and career centers | Connect AI literacy to employability, occupational transitions, job-search practices, and employer needs. |
| Employers and industry partners | Identify emerging task requirements and provide realistic practice contexts without converting training into vendor-specific promotion. |
| Community organizations and nonprofits | Reach populations that may distrust institutions or be excluded from formal training and provide culturally relevant support. |
| Senior centers and aging-services organizations | Support age-inclusive instruction, scam/deepfake awareness, and confidence-building. |
| Disability and accessibility advocates | Evaluate whether AI tools and training expand or restrict access and whether accommodations preserve autonomy. |
| Multilingual / language-access specialists | Review translated materials, interpretation, and the reliability of AI-assisted language use in high-consequence contexts. |
| IT and identity/access-management staff | Provide secure accounts, device compatibility, access controls, technical support, logging, and lifecycle management. |
| AI platform and software vendors | Provide product capabilities, accessibility documentation, security/data terms, support, and change notices; respond to defects and risks. |
| AI system designers / data-science specialists | Explain system limits, model behavior, evaluation, bias, and technical risk where deeper expertise is needed. |
| Cybersecurity and privacy staff | Set data-sharing rules, threat controls, safe authentication practices, incident procedures, and user guidance. |
| Legal, compliance, and policy staff | Review institutional rules, contracts, disclosure, records, intellectual property, employment, student, and sector-specific obligations. |
| Procurement and vendor-management staff | Evaluate contracts, data use, accessibility, portability, pricing, support, lock-in, and exit terms. |
| Public agencies and policymakers | Establish public-service expectations, workforce/education policy, consumer protections, and accountability where applicable. |
| Consumer and civil-rights advocates | Represent concerns about discrimination, manipulation, surveillance, automated decision-making, and unequal access to recourse. |
| Media and journalists | Support public understanding of AI claims, misinformation, provenance, and the distinction between generated content and verified reporting. |
| Researchers and evaluators | Measure skill changes, equity impacts, unintended effects, and whether the intervention improves real participation. |
| Funders and philanthropic partners | Support curriculum, access, staffing, evaluation, and scale while avoiding short-term pilots with no sustainability plan. |
| Community advisory group | Reviews priorities, safeguards, outcomes, complaints, and changes in AI use from the perspective of affected users. |
The AI initiative needs to create a baseline before selecting a curriculum or purchasing large numbers of AI licenses. The assessment should identify both what participants can currently do and the conditions that may prevent improvement.
| Readiness question | Evidence to collect | Proceed when |
|---|---|---|
| Who is being excluded or disadvantaged? | Demographics, role, education/work context, language, disability, prior digital-skills evidence, self-reported barriers | Priority groups and use contexts are defined without relying only on demographic assumptions |
| Do participants have usable access? | Devices, browsers/apps, accounts, licensing, connectivity, accessibility, language support | Participants can reach an appropriate tool or a supported alternative |
| What foundational skills are missing? | File handling, accounts, search, documents, security basics, source evaluation | The program includes prerequisite support rather than assuming AI training can bypass foundational gaps |
| Can participants interact productively with AI? | Task framing, context setting, iterative prompting, comparison, refinement | Baseline performance can be observed using realistic tasks |
| Can participants evaluate outputs? | Fact checking, source tracing, uncertainty recognition, bias detection, deepfake/misinformation awareness | Assessment distinguishes fluent use from critical use |
| Are institutional policies clear? | Approved tools, sensitive-data rules, academic/workplace use, disclosure, human-review requirements | Participants and staff can explain permitted, prohibited, and high-risk uses |
| Can the institution sustain the program? | Staff skills, training time, support model, tool budget, update process, evaluation capacity | Recurring training, support, curriculum refresh, and governance responsibilities are funded and assigned |
| Competency level | Demonstrated capability | Typical support need |
|---|---|---|
| 1. Foundational digital readiness | Can use a device, browser, account, files, search, and basic security practices | Foundational digital-skills instruction and confidence building |
| 2. AI awareness | Understands that AI generates or predicts outputs, may be wrong, and should not automatically be trusted | Plain-language orientation and guided examples |
| 3. Functional AI use | Can describe a task, provide context, iterate, and apply outputs to a low-risk activity | Practice, templates, coaching, and task-specific examples |
| 4. Critical AI literacy | Can verify claims, compare sources, recognize uncertainty, bias, manipulation, and weak evidence | Information-literacy practice and domain expertise |
| 5. Responsible applied use | Can protect data, follow policy, document use, preserve human accountability, and select when AI is or is not appropriate | Role-specific standards, supervision, and governance |
| 6. Adaptive capability | Can evaluate new tools, transfer principles across systems, and continue learning as models and policies change | Ongoing communities of practice and periodic reassessment |
The visible problem with end users and AI is often described as “people do not know how to use AI,” but that label can conceal very different causes. The conceptual model requires the program to diagnose those causes before selecting an intervention.
| Barrier area | Possible condition | Required response |
|---|---|---|
| Foundational literacy | Weak file, search, account, document, or information skills | Provide prerequisite digital-skills support rather than jumping directly to prompts |
| Access and cost | No suitable device, account, subscription, or accessible interface | Provide tool access, shared labs, approved free/low-cost alternatives, and accessibility support |
| Prompt / interaction literacy | The user cannot translate a goal into useful instructions or iterative dialogue | Teach task framing, context, constraints, examples, iteration, and comparison |
| Critical evaluation | User treats fluent output as accurate evidence | Teach verification, source checking, uncertainty, domain review, and evidence comparison |
| Misinformation/ deepfakes | Generated or manipulated content is mistaken for authentic information | Teach provenance checks, cross-source verification, skepticism, and escalation for high-consequence claims |
| Bias and representation | Outputs reflect stereotypes, missing perspectives, or unequal data histories | Teach bias recognition, inclusive prompting, comparison, and human review; escalate systemic issues |
| Privacy and security | Sensitive personal, student, client, health, or organizational data is entered into unsuitable tools | Define data classifications, approved tools, safe alternatives, and incident reporting |
| Language and accessibility | Interfaces, training, or output quality do not work equally for all users | Provide multilingual and accessible instruction, assistive-technology testing, and human support |
| Institutional readiness | Policies conflict, staff are unprepared, or guidance changes by department | Establish common governance, role-specific rules, professional development, and escalation pathways |
| Low self-efficacy / fear | Participants avoid AI because of anxiety, prior negative experiences, or belief that it is “not for them” | Use trusted facilitators, low-stakes practice, peer support, and visible progress |
| Overconfidence / overreliance | Participants automate judgment before developing expertise or verification habits | Teach human accountability, reflection, domain limits, and appropriate non-AI alternatives |
| Rapid technology change | Training becomes obsolete as tools, features, and policies change | Teach transferable principles and maintain a scheduled curriculum-refresh process |
| Workforce displacement concern | Participants experience AI mainly as a threat to employment | Connect literacy to task redesign, career navigation, worker voice, reskilling, and transparent organizational planning |
The learning-environment logic applies across schools, homes, libraries, community sites, campuses, workplaces, and other settings. AI literacy should therefore be designed around the location and purpose of use rather than treated as a universal, standalone computer skill.
| Context | AI opportunity | Digital-divide risk | Program response |
|---|---|---|---|
| K-12/higher education | Tutoring, brainstorming, feedback, translation, research support, coding | Unequal guidance, weak verification, overreliance, assessment confusion, and privacy concerns | Age-appropriate literacy, educator development, assignment redesign, source verification, and transparent rules |
| Workforce/job search | Writing, analysis, workflow automation, learning, applications, career exploration | Unequal productivity gains, job displacement, confidential data misuse, and shallow skill development | Role-based training, human review, data rules, reskilling, worker participation |
| Libraries/community centers | Public access, information help, digital navigation, learning support | Users may have access, but no trusted coach or verification support | Drop-in coaching, guided labs, media literacy, multilingual and accessible assistance |
| Small business/entrepreneurship | Marketing, planning, customer communication, analysis, and administration | Poor-quality outputs, legal/reputation risk, sensitive data exposure, vendor dependence | Practical use cases, verification, privacy, disclosure, customer-trust safeguards |
| Civic / public-service participation | Forms, translation, information navigation, benefits and service access | Generated misinformation, automation errors, weak appeal paths, and unequal ability to question systems | Human assistance, source verification, clear official channels, accessible alternatives, and recourse |
| Home/family life | Learning support, planning, translation, health information questions, creativity | Deepfakes, scams, wrong advice, privacy loss, family conflict over use | Family-facing literacy, safety guidance, high-consequence escalation, and age-appropriate expectations |
Each proposed intervention should pass the conceptual model’s strategic decision gate. AI enthusiasm should not be allowed to substitute for causal fit, and risk concerns should not be used to justify blanket exclusion where guided access could improve opportunity.
| Fit test | Question | Application to this case |
|---|---|---|
| Equity fit | Does the strategy reach people with the greatest skill, confidence, access, language, disability, or opportunity gaps? | Prioritize demonstrated need and ensure training does not reward only users who already possess advanced digital skills. |
| Causal fit | Does the intervention address the verified barrier? | Do not prescribe prompt training when the actual barrier is foundational literacy, inaccessible tools, unclear policy, or lack of trusted support. |
| Feasibility fit | Can the institution staff, fund, update, and support the program? | Use realistic training time, qualified facilitators, support capacity, licensing, and curriculum-refresh plans. |
| Technology fit | Are selected tools appropriate, accessible, secure, and stable enough for the intended use? | Use tool-neutral competencies and approved systems suited to the risk level and participant context. |
| Learning/opportunity fit | Does the intervention improve participation and capability rather than simply increase tool use? | Require authentic tasks, reflection, verification, and evidence of transferable skills. |
| Safeguard fit | Are privacy, accessibility, bias, cybersecurity, cost, family impact, well-being, and sustainability addressed? | Treat safeguards as design requirements and decision gates, not as a separate compliance handout after launch. |
The appropriate solution may be a portfolio rather than a single course. Different participants may need different entry points and levels of support.
| Option | Best fit | Advantages | Limitations |
|---|---|---|---|
| Short AI-awareness orientation | Broad populations need a basic understanding | Fast, low-cost introduction to capabilities, limitations, privacy, and verification | Insufficient for demonstrated applied skill |
| Foundational digital-skills bridge | Participants lacking core device, search, file, account, or security skills | Addresses prerequisite barriers directly | Requires time before AI-specific outcomes appear |
| Structured AI-literacy course | Students, workers, adult learners, community cohorts | Provides sequenced competencies and assessment | May become outdated if tied too tightly to one tool |
| Embedded role-based modules | Courses, occupations, departments, or service programs with specific tasks | Makes learning relevant and transferable to real work | Requires subject matter and training collaboration |
| Library/community AI lab | Public users needing access and trusted assistance | Reduces access and confidence barriers; supports drop-in learning | Requires staffing, devices, accounts, privacy controls, and ongoing support |
| Peer mentor / digital navigator model | Users needing individualized confidence and problem-solving support | Trusted, flexible, culturally responsive support | Quality depends on mentor training and supervision |
| Educator/supervisor professional development | People responsible for guiding others | Creates consistent expectations, coaching, assessment, and human review practices | Requires protected time and leadership support |
| Responsible-AI policy and job aids | Institutions already using AI but lacking clear boundaries | Reduces confusion and unsafe use; supports consistent escalation | Policy alone does not build capability |
| Microcredential/competency badge | Participants who benefit from visible evidence of skill | Encourages progression and workforce signaling | Must assess demonstrated capability rather than attendance |
The recommended model-guided program combines the following components:
Baseline and segmentation
Assess foundational digital skills, AI access, confidence, critical evaluation, privacy awareness, and applied-use needs.
Segment participants by demonstrated capability and context rather than by age or job title alone.
Identify groups who are affected by AI decisions, even if they are not active AI users.
Foundational and applied AI literacy
Provide prerequisite digital-skills support where needed.
Teach task framing, context, iteration, comparison, and appropriate use of AI assistance.
Use authentic education, workforce, civic, and daily-life tasks instead of generic prompt demonstrations.
Critical information literacy
Teach verification, source tracing, corroboration, uncertainty, hallucination awareness, deepfake recognition, and bias detection.
Require participants to explain why an output should or should not be trusted before using it in a consequential task.
Responsible use and safeguards
Define privacy, sensitive data, accessibility, cybersecurity, authorship, disclosure, and human review expectations.
Create clear pathways for questions, incidents, appeals, and situations where AI should not be used.
Institutional capacity
Train educators, librarians, navigators, supervisors, support staff, and administrators.
Maintain approved tool guidance, accessible materials, vendor review, policy updates, and communities of practice.
Evaluation and adaptation
Measure demonstrated skills, safe judgment, real participation, and equity outcomes separately from attendance and tool usage.
Refresh the program as technology, risks, policies, workforce expectations, and participant needs change.
The equity and safeguard review explicitly includes privacy, accessibility, AI bias, cybersecurity, cost burden, family impact, well-being, and sustainability. This initiative should operationalize those safeguards in training, tool selection, policy, and evaluation.
| Safeguard area | Minimum practice |
|---|---|
| Privacy and data protection | Teach what information must not be entered into public or unapproved AI systems; use approved tools and data classifications for institutional work. |
| Cybersecurity | Use secure accounts, multifactor authentication where appropriate, phishing/scam awareness, safe file handling, and incident reporting. |
| AI bias and fairness | Teach participants to look for missing perspectives, stereotypes, unequal treatment, and data-history effects; require human review in consequential contexts. |
| Misinformation and deepfakes | Use verification routines, trusted-source comparison, provenance checks where available, and escalation for high-consequence claims. |
| Accessibility | Select accessible tools, provide alternative formats and assistive technology support, and include people with disabilities in testing. |
| Language access | Provide multilingual instruction and human interpretation pathways; do not treat AI translation as infallible for legal, health, education, or safety-critical content. |
| Human accountability | Identify decisions that require a responsible human reviewer and preserve appeal or correction mechanisms. |
| Academic / professional integrity | Define when AI assistance is permitted, how it should be disclosed or cited, and what work must demonstrate independent human competence. |
| Well-being and dependency | Teach appropriate boundaries, reflection, and non-AI alternatives; monitor whether AI use reduces rather than expands independent judgment. |
| Cost and sustainability | Avoid creating a curriculum that requires unaffordable premium subscriptions; maintain viable tool alternatives and budget for continuing support. |
| Vendor and lifecycle risk | Review data practices, accessibility, portability, feature changes, support, and exit options rather than assuming a current tool will remain suitable. |
Trust standard: Participants should understand not only how to use an AI system, but also what the system may retain, what it may get wrong, what risks apply to the task, who remains accountable, and where to obtain a human alternative.
| Phase | Primary work | Decision gate |
|---|---|---|
| Phase 1: Initiation and problem definition | Appoint sponsor and program team; define populations, use contexts, outcomes, and boundaries; inventory current AI use and policies. | Proceed when the project is framed as a skills, literacy, equity, and governance initiative rather than a tool rollout. |
| Phase 2: Baseline and readiness assessment | Assess foundational digital skills, AI access, current practices, confidence, verification habits, staff readiness, accessibility, and policy clarity. | Proceed when priority gaps and participant segments are evidence-based. |
| Phase 3: Program and safeguard design | Create competencies, learning modules, tool-access model, policies, facilitator training, accessibility review, privacy/security guidance, and evaluation instruments. | Proceed when the program passes equity, feasibility, technology, learning, and safeguard-fit tests. |
| Phase 4: Pilot and validation | Pilot with diverse users across education, workforce, age, language, and accessibility contexts; observe actual task performance and support needs. | Scale only when participants can demonstrate useful and safe behavior, not merely favorable reactions. |
| Phase 5: Scaled delivery | Launch cohort training, embedded modules, labs, coaching, professional development, and communications; maintain support and escalation channels. | Expansion is complete only when priority groups can access training and support without structural barriers. |
| Phase 6: Operational review and improvement | Review skill outcomes, policy questions, incidents, technology changes, inequities, support patterns, and stakeholder feedback each term or cycle. | Revise curriculum, tools, policies, and support whenever evidence shows a mismatch or new risk. |
| Operating area | Minimum procedure | Suggested review |
|---|---|---|
| Participant intake | Assess goals, foundational skills, accessibility/language needs, and appropriate learning pathway | At enrollment and when context changes |
| Tool access | Maintain approved tools, account instructions, accessible alternatives, and clear data-use guidance | Monthly and after major vendor changes |
| Facilitator preparation | Train facilitators on competencies, common misconceptions, safeguards, escalation, and accessible instruction | Before delivery, refresh each term/cycle |
| Hands-on practice | Use low-risk, realistic tasks, iterative prompting, verification, and reflection rather than passive demonstrations | Every learning module |
| Verification routine | Require source comparison, evidence checks, and human review for consequential claims | Built into assessments and applied tasks |
| Help and coaching | Provide office hours, library/community support, peer mentoring, digital navigation, or help-desk channels | Weekly during launch; adjust to demand |
| Policy questions | Maintain a published escalation path for uncertain academic, workplace, privacy, legal, or ethical uses | Continuous; summarize recurring issues monthly |
| Accessibility and language | Provide alternate formats, interpretation, assistive-technology support, and accessible interfaces | At intake and after tool/material changes |
| Incident handling | Document privacy, security, harmful-output, discrimination, or misuse concerns and route them to responsible staff | Immediately; trend review monthly |
| Curriculum refresh | Review tool changes, new risks, participant errors, workforce/education changes, and outdated examples | Quarterly or each academic/workforce cycle |
Support triage should answer four questions in order:
Is the barrier basic digital access or a foundational skill rather than AI-specific literacy?
Is the user able to define the task and interact with the system effectively?
Can the user evaluate the output, verify evidence, and recognize uncertainty or bias?
Is the remaining issue one of privacy, policy, accessibility, language, domain expertise, or human accountability?
Outputs are the countable products that confirm implementation. Recommended outputs include:
Completed AI-literacy and digital-skills baseline assessment.
Approved competency framework and learning pathways.
Number of facilitators, educators, navigators, supervisors, and support staff trained.
Number of participants assessed, enrolled, completing modules, and receiving coaching.
Number of accessible and multilingual learning resources produced.
Number of AI labs, workshops, embedded course modules, workforce sessions, or community support events delivered.
Number of approved tools or access pathways available to participants.
Published responsible-use, privacy, security, accessibility, and escalation guidance.
Number and type of support requests, policy questions, incidents, and referrals resolved.
Completed participant performance assessments and follow-up evaluations.
Curriculum and policy refreshes completed after tool or institutional changes.
A program can train hundreds of people and still widen the AI divide if advanced users benefit most, priority populations cannot access suitable tools, participants cannot verify the information generated, or institutional policy remains too unclear for people to apply what they have learned.
Outcomes describe whether the initiative changed meaningful capability and participation. The program should distinguish attendance and tool usage from demonstrated literacy, responsible judgment, and opportunity outcomes.
| Domain | Recommended measures |
|---|---|
| Access and reach | Participation by priority group; access to approved tools; completion by language/accessibility need; waitlist or exclusion patterns |
| Foundational skills | Improvement in prerequisite digital tasks required for AI use |
| AI interaction | Ability to define tasks, provide context, iterate, compare outputs, and select useful results |
| Critical literacy | Accuracy of fact-checking, source verification, uncertainty recognition, bias identification, and deepfake/misinformation response |
| Responsible use | Appropriate handling of sensitive data, disclosure, attribution, policy compliance, and human-review decisions |
| Applied capability | Successful completion of authentic education, workforce, civic, or daily-life tasks with appropriate AI assistance |
| Confidence and agency | Improved self-efficacy without increased blind trust or dependence |
| Education outcomes | Improved learning participation, research practice, feedback use, assignment design, or ability to demonstrate independent understanding |
| Workforce outcomes | Improved task efficiency, job-search capability, reskilling progress, role readiness, and ability to adapt to AI-enabled workflows |
| Equity outcomes | Narrowing of skill and participation gaps across age, income, language, disability, education, and occupational groups where measurable and privacy-protective |
| Institutional maturity | Policy clarity, staff readiness, support resolution, accessibility, incident management, curriculum refresh, and stakeholder satisfaction |
Minimum Dashboard
| Monthly / operational | Each term or program cycle | Annually |
|---|---|---|
| Participation, access issues, support volume, incidents, tool changes, facilitator capacity | Competency gains, participant feedback, equity gaps, applied-use performance, unresolved barriers | Program reach, workforce/education outcomes, policy maturity, tool strategy, cost, sustainability, redesign decisions |
| Risk | Likely effect | Mitigation |
|---|---|---|
| Training rewards already-advanced users | Existing skill gaps widen | Segment by need, provide foundational pathways, and monitor outcomes by priority group |
| Prompt fluency is mistaken for literacy | Participants generate polished but unreliable output | Assess verification, reasoning, source use, and human judgment, not prompt style alone |
| AI hallucinations or misinformation are accepted | Bad decisions, weak learning, or reputational harm | Teach verification routines and require human/domain review for consequential tasks |
| Deepfakes and synthetic media increase confusion | Fraud, manipulation, and declining trust | Media-literacy practice, provenance awareness, trusted-source comparison, and escalation |
| Bias affects outputs or decisions | Discrimination or exclusion | Bias testing, inclusive review, human accountability, complaint/appeal mechanisms, and vendor escalation |
| Sensitive data is entered into unsuitable tools | Privacy, legal, or security harm | Approved-tool rules, data classification, practical examples, access controls, and incident response |
| Tool costs create a new access divide | Only higher-income users obtain advanced functionality | Provide institutional access, free/low-cost alternatives, and avoid premium-only curriculum dependencies |
| Rapid vendor changes make training obsolete | Confusion and wasted curriculum effort | Teach transferable competencies and refresh content on a defined schedule |
| Educators or supervisors are unprepared | Inconsistent expectations and weak support | Professional development, communities of practice, clear policies, and escalation pathways |
| Users become overdependent on AI | Reduced independent reasoning, writing, problem solving, or domain learning | Require reflection, human-authored work, non-AI practice, and appropriate use boundaries |
| Fear of AI suppresses participation | Priority users avoid training or opportunity | Trusted facilitators, low-stakes practice, transparent communication, and worker/community voice |
| AI is used where accessibility is weak | People with disabilities are excluded or forced into inferior workflows | Accessibility testing, alternate tools, accommodations, and advocate participation |
| Language output is trusted without verification | High-consequence translation errors | Human review and specialist support for legal, health, safety, and formal institutional communication |
| Workforce training ignores displacement concerns | Low trust and resistance | Include career navigation, reskilling, worker voice, job redesign, and transparent organizational planning |
| Institution collects excessive evaluation data | A program designed for inclusion creates a privacy risk | Use minimum necessary data, aggregation, retention limits, and transparent evaluation practices |
| Vendor lock-in narrows future choices | Higher cost and constrained curriculum | Use tool-neutral competencies, portable content, contract review, and exit planning |
Communication should explain why AI literacy is becoming part of digital inclusion without presenting AI as either universally beneficial or inherently harmful. Messages should emphasize capability, critical judgment, safety, opportunity, and choice.
| Audience | Key message | Recommended channel |
|---|---|---|
| Students and learners | AI can assist learning, but useful participation requires verification, independent understanding, privacy, and clear rules | Classroom orientation, LMS, library, advising, and short demonstrations |
| Parents and families | The program teaches both opportunities and safeguards, including misinformation, privacy, age-appropriate use, and learning expectations | Family nights, translated guides, school/community communication, trusted liaisons |
| Educators and trainers | AI literacy should be embedded into learning design, source evaluation, assessment, accessibility, and student support | Professional development, faculty/staff meetings, instructional coaching, communities of practice |
| Workers and job seekers | AI skills can support productivity and employability, but safe use requires data protection, verification, and human accountability | Workforce centers, employer training, career services, union/worker channels, where applicable |
| Older adults/community users | Training is practical, paced, nonjudgmental, and includes scams, deepfakes, privacy, and everyday tasks | Libraries, senior centers, community groups, printed guides, and peer mentors |
| Institutional leaders | Success depends on capability, policy, support, accessibility, and outcomes not tool adoption alone | Dashboards, governance meetings, risk reports, and annual strategic review |
| Community partners and funders | The initiative links AI access with foundational skills, critical literacy, responsible use, and measurable inclusion outcomes | Partner briefings, grant reports, public summaries, advisory meetings |
The full cost of the initiative extends beyond software licenses. Sustainable AI literacy requires recurring investment in staff development, facilitation, accessible materials, support, curriculum maintenance, evaluation, privacy/security review, and replacement or diversification of tools as conditions change.
| Cost category | Examples | Planning action |
|---|---|---|
| Program design | Competency framework, curriculum, assessment, accessible/multilingual materials | Fund initial design and scheduled refresh cycles |
| Tool access | Licenses, accounts, lab devices, assistive technology, approved alternatives | Model recurring cost and maintain non-premium pathways |
| Staffing | Program management, trainers, librarians, navigators, IT, accessibility, evaluation, policy support | Fund delivery and support, not only initial content creation |
| Professional development | Educator, supervisor, navigator, and support-staff training | Build recurring release time and onboarding into operating budgets |
| Evaluation | Baseline, competency assessment, dashboards, qualitative feedback, privacy-protective analysis | Treat measurement as a core program cost |
| Safeguards and governance | Privacy/security review, accessibility testing, legal/policy work, incident response | Assign recurring ownership and update triggers |
| Community access | Library/community labs, outreach, interpretation, transportation, or scheduling support where needed | Budget for inclusion barriers outside formal institutions |
| Transition and lifecycle | Vendor changes, data export, curriculum updates, feature changes, replacement tools | Maintain exit options and a reserve for rapid changes |
Sustainability principle: A program that teaches one current AI product but cannot adapt when the product, pricing, policy, or capabilities change is not a sustainable literacy intervention.
The program should use the conceptual model’s feedback loop to determine whether participants are gaining meaningful capability and whether new AI practices are creating additional barriers or risks.
| Observed issue | Likely question | Corrective action |
|---|---|---|
| Participants complete training but still trust unsupported outputs | Was verification taught and assessed strongly enough? | Increase source-checking practice, domain review, and performance assessment |
| Advanced users improve faster than beginners | Are prerequisites and support pathways sufficient? | Expand foundational digital-skills bridge, coaching, and differentiated instruction |
| Low participation among older adults or community users | Is the barrier trust, scheduling, language, location, confidence, or relevance? | Use trusted sites, peer mentors, practical use cases, and accessible scheduling |
| Educators or supervisors give conflicting guidance | Are policies and role expectations clear and up to date? | Issue common standards, role-specific examples, professional development, and escalation |
| Frequent privacy/security questions | Are tool boundaries and data classifications understandable? | Simplify guidance, add realistic examples, review vendor settings, and reinforce approved alternatives |
| Participants misuse AI in high-consequence tasks | Were human-review boundaries and domain limits explicit? | Strengthen safeguards, supervision, scenario practice, and escalation requirements |
| Tool changes disrupt training | Is the curriculum too vendor-specific? | Rebuild around transferable competencies and update demonstrations separately |
| Workforce participants fear displacement | Is training addressing the opportunity without acknowledging the job change? | Add career navigation, reskilling, worker voice, and transparent change-management content |
| Accessibility complaints increase after platform update | Did the vendor or configuration introduce new barriers? | Retest accessibility, provide alternate access, escalate defects, and revise tool choice if necessary |
Diagnostic response: When outcomes do not improve, return to the barrier map. Poor AI use may reflect foundational skills. Low adoption may reflect cost or fear. Weak verification may reflect information-literacy gaps. Unsafe behavior may reflect unclear policy. A technically powerful AI tool should not be retained merely because the original plan selected it.
Before declaring the initiative operationally complete, the program manager should be able to answer yes to the following questions.
Problem and population
Has the project defined the AI-driven divide in terms of capability and participation rather than tool access alone?
Have foundational digital skills, AI access, confidence, critical literacy, and support needs been assessed?
Have priority populations been identified by evidence rather than stereotype?
Stakeholders and governance
Are learners, workers, educators, community members, accessibility advocates, technology staff, policy staff, and support partners represented?
Are stakeholder power, influence, community voice, and accountability clear?
Is there a process for complaints, incidents, policy questions, and corrective decisions?
Skills and learning design
Does the program include foundational readiness, functional AI use, critical evaluation, responsible use, and authentic application?
Are participants assessed on demonstrated capability rather than attendance?
Do educators and supervisors understand how to guide and evaluate AI-supported work?
Safeguards
Are privacy, cybersecurity, accessibility, language, bias, misinformation, human accountability, cost, well-being, and sustainability addressed?
Are high-consequence uses subject to appropriate human and domain review?
Can participants access a human alternative when an AI process is unsuitable or fails?
Implementation and sustainability
Are tools, staffing, professional development, support, evaluation, and curriculum refresh funded beyond launch?
Is the program tool-neutral enough to survive vendor and technology change?
Are accessible and low-cost participation pathways available?
Measurement and improvement
Are outputs and outcomes tracked separately?
Are skill and participation gaps reviewed across priority populations where privacy-protective?
Is stakeholder feedback used to revise curriculum, tools, safeguards, and support?
Can the program return to problem definition when AI creates a new barrier that the original design did not anticipate?
The community or institution should proceed with an AI literacy and digital-skills inclusion initiative as a new application of the dissertation’s conceptual model. The project should not define the emerging divide as a simple difference between people who use AI and people who do not. The more meaningful divide concerns who can use AI with sufficient knowledge, confidence, critical judgment, access, support, and safeguards to convert the technology into real educational, workforce, civic, and personal benefits.
The program should preserve foundational digital literacy, provide practical access to appropriate tools, teach task framing and iterative use, make verification and information literacy central, protect privacy and security, address accessibility and language, explain bias and misinformation, and maintain clear human accountability. Training should be adapted to developmental, institutional, occupational, and community contexts rather than imposing one universal level of AI use.
The strongest measure of success will not be the number of accounts created, prompts submitted, workshops delivered, or AI tools adopted. Success will be demonstrated when participants can decide when AI is useful, interact with it effectively, recognize when it is wrong or unsafe, verify important claims, protect sensitive information, apply human judgment, and continue learning as technology changes. Institutions should also be able to demonstrate that the benefits of AI adoption are not accruing only to people who already possess the strongest digital skills.
Applied in this way, the conceptual model makes visible a new form of the digital divide: people may be technically connected yet increasingly separated by their capacity to understand and act within AI-mediated environments. The response is therefore not simply more technology. It is a coordinated system of skills, critical literacy, responsible use, accessible support, institutional readiness, stakeholder voice, safeguards, measurement, and continuous improvement.
Conceptual foundation note: This case is derived from the dissertation model’s problem-framing, stakeholder/context, diagnosis, technology, solution-design, results, and continuous-improvement stages and from its explicit treatment of digital skills, literacy, AI applications, algorithmic bias, responsible AI, accessibility, privacy, cybersecurity, education/training, and stakeholder power.
Figure AH86 AI Literacy and Digital Skills Initiative Overview