AI, Prompt-

Writing, &
Instructional
Design
AN ETHICS-FOCUSED OVERVIEW

JENNY ROACH, MBA

MLIS CANDIDATE
Our Focus Today
Contextualizing AI’s Development Ideas for Q&A:
The AI Link between Learners, • Your struggles and successes with AI
Educators and Designers
• How it’s manifesting in your professional life
AI Myths and Concerns
• The ideas you have for using it
Prompt-Writing 101 • Implementation issues
▪ Generative AI: Creates new content by learning
existing patterns.
▪ Ex: ChatGPT, DALL-E, Copilot

Ethics and a Framework for Responsible

Implementation
Contextualizing
AI’s
Development
Introduction
• Ai is here.

• The debate is no longer about its feasibility as a tool, but rather its ethical and successful
implementation in our society.
• AI is not new, and it is not a fad.

• AI is not an over-glorified version of predictive texting.

• AI is more likely to enhance human work than replace human workers.

• AI-assisted creativity is a skill in itself.

• It’s evolving faster than it probably should.

• AI will not cheapen education—it will elevate it into something better.

• … but only if valued and if implemented correctly and ethically.

 The Omniscient Chef
Imagine you have a chef who has an incredible memory—they can recall
every recipe you've ever shared with them. However, the quality of the
food they prepare depends entirely on what the quality of what you
provide them to remember and work with.

What would you give this chef to read and remember?

❖ Diverse recipes from global cuisines to expand their repertoire

❖ Are the recipes detailed and accurate, covering ingredients,

techniques, and presentation?

❖ Update their recipe collection to include modern trends and dietary

needs

❖ Maybe even tiny little nuances to cooking, such as only using the
Crisco brand for lard.
 How do I personally use AI?
• Creative writing
• Brainstorming/soundboard
• Daily tasks / life-planning / goal-setting
• Introspection and personal reflection
• Real-time feedback (on anything)
• Simplifying data / too much information
• Learning new languages, learning about new cultures
• One-on-one learning in general
• Checking understanding of complex topics
• Mental health and self-growth
• Artwork for “work”
• Video creation (Sora)
• Tasks that would otherwise be unnecessarily labor-intensive or slow
 Internet Adoption in
Phase AI Adoption in Education
Education

1990s: The Internet faced

2010s: Early AI applications
hesitation in education due to
faced skepticism due to ethical
Early Skepticism and concerns about accessibility
concerns, biases, and fears of
Limited Adoption and the digital divide. Critics
diminishing the human role in
worried it would exacerbate
teaching. (PMC)
inequalities. (PMC)

Internet vs. AI
Adoption
Late 1990s – Early 2000s: The
Late 2010s – Early 2020s: AI
Internet became more
technologies, such as
accessible and was increasingly

Timeline
Gradual Integration and personalized learning platforms
recognized as a valuable
Increased Acceptance and intelligent tutoring systems,
educational tool, expanding
were integrated into education.
learning opportunities. (Pew
(Springer)
Research)

Mid 2020s – Present: AI is now

Mid 2000s – Present: The
widely used for automated
Internet became essential for
Mainstream Adoption and grading, adaptive learning, and
education, but issues like the
Ongoing Challenges administrative tasks, but
digital divide persisted, affecting
concerns over data privacy and
equitable access. (PMC)
ethics remain. (Ed.gov)
 Realm of the Purely Theoretical
… but it may be possible with mushroom neural
networks to fill in any computing skill gaps. So
keep an eye out for that.

We are here (mainstream use)

Most AI today is narrow AI, excelling

in specific tasks but unable to
generalize across domains. For
example:

ChatGPT is excellent at natural

language understanding but cannot
directly operate robots or make
scientific discoveries independently.

Chart made in collaboration with ChatGPT

The AI Link:
Learners, Educators
and Designers
 Wisdom
(higher-order thinking and contextualization)

The Hierarchical

Path of the Learner

Knowledge
Layers of Knowledge (applying patterns and understanding)

(DIKW Pyramid)
Information
(processed and organized data)

Data
(raw facts and figures)

From Marcia J. Bates’ “The Invisible Substrate of Information Science”

 How AI Supports Learning at Every Stage of Knowledge
Applications by
Layer of Knowledge How AI Works Learner Impact Real-World Example
Educators/IDs

Collects raw data from Students have a record of A learning management Analyzes AI-generated data
student interactions (e.g., their interactions, but the system logs quiz attempts, reports on student
Data quiz scores, time spent on data itself is not yet tracking how long students performance trends to adjust
lessons, click patterns). meaningful. spend on each question. instructional strategies.

AI processes and organizes AI generates a report

Students receive structured Uses AI-powered analytics
data into structured insights, showing that a student
feedback on their progress, tools to identify patterns in
Information such as identifying patterns in
allowing them to understand
consistently struggles with
student engagement and
performance and geometry problems involving
their performance. adapt content accordingly.
engagement. spatial reasoning.

Employs AI-driven adaptive

AI applies past learning data Students engage in AI adapts the student's
learning systems to
to make recommendations, personalized learning learning path, offering
personalize coursework,
Knowledge adapting lesson plans based experiences, receiving interactive 3D visualization
ensuring each learner
on identified strengths and targeted content to reinforce exercises to strengthen
receives content suited to
weaknesses. their understanding. spatial reasoning skills.
their needs.

A student writing a research

Uses AI to synthesize
AI facilitates critical thinking Students develop the ability paper uses AI to summarize
research from multiple fields,
by enabling students to apply to transfer knowledge into multiple peer-reviewed
integrating findings into new
Wisdom concepts in real-world real-world applications, studies, comparing
curriculum models or
scenarios, encouraging making learning more perspectives to develop a
innovative teaching
metacognitive learning. meaningful and lasting. nuanced argument based on
strategies.
qualitative analysis.
 Emerging Trend How AI Enhances It Key Benefit

AI-powered analytics, AI-

More efficient teaching &
AI in Education generated content, reduced
data-driven decisions.
administrative workload.

Emerging AI-driven adaptive learning

Trends in
platforms, real-time Customized education for
Personalized Learning
feedback, individualized better student success.
pathways.

Education Mobile Learning

AI-powered apps, real-time
tutoring, personalized
Flexible learning anytime,

+ AI recommendations.
anywhere.

AI-driven adaptive
difficulty, immediate Higher engagement &
Gamification
feedback, interactive motivation.
learning.

AI-optimized online & in-

Seamless learning
person learning, tracking
Blended Learning experience with improved
student progress,
efficiency.
customized instruction.
Adaptive
Learning
Systems

Source: Data Science Dojo (2024). "Adaptive AI 101." Link

Multi-Modal AI
Multi-Modal AI
How It Relates to
Study Description
Instructional Design
AI-driven adaptive learning systems use Multimodal AI enhances personalized
Joshi, R. (2023). Adaptive
multimodal inputs to dynamically adjust learning by adapting content to each
learning through artificial
instruction based on individual needs. This helps student’s progress, ensuring tailored
intelligence. ResearchGate.
bridge knowledge gaps and improve learning instruction that optimizes engagement
Link
outcomes. and comprehension.

Gibson, J. (2023). 10 ways Multimodal AI streamlines instructional

AI tools assist instructional designers by
artificial intelligence is design, allowing educators to create
automating content creation, quiz generation,
transforming instructional adaptive, interactive, and scalable
and real-time feedback, enhancing efficiency and
design. EDUCAUSE Review. learning experiences without manual
engagement.
Link intervention.

Chen, C.-M., & Chiu, P.-S. AI-driven multimodal systems enhance

Research suggests multimodal AI improves
(2015). The learning style- learning outcomes by supporting
comprehension and retention by aligning
based adaptive learning multiple engagement formats (text,
instructional materials with cognitive learning
system architecture. images, and audio), making education
models.
ResearchGate. Link more interactive and effective.
 AI-Assisted Content Generation
Aspect AI-Assisted Content Generation AI-Generated Content
Human creators utilize AI tools to enhance and streamline Content is autonomously created by AI
Definition the content creation process, maintaining primary control systems with minimal to no human
over the output. intervention.

Low: AI systems independently generate

High: Humans guide the AI, providing input and making
Human Involvement content based on their training data and
critical decisions throughout the creation process.
algorithms.

Ensured by human oversight, allowing for contextual May lack nuanced understanding,
Quality Control understanding, ethical considerations, and alignment with potentially leading to inaccuracies or
specific objectives. misinterpretations without human review.

Relies on patterns in existing data, which

Combines human creativity with AI efficiency, leading to
Creativity and Originality may result in less original or creative
innovative and contextually relevant content.
outputs.
AI may inadvertently produce biased or
Humans can apply ethical standards and cultural
Ethical Considerations culturally insensitive content due to
sensitivities, ensuring content appropriateness.
training data limitations.
Enhancing writing clarity, generating data-driven insights, Producing large volumes of content, such
Use Cases and automating repetitive tasks while retaining human as news articles or reports, with speed but
judgment. requiring human oversight for accuracy.

Source: Johnson, A. (2024). What is the difference between AI-assisted and AI-generated content? Clearscope.
AI Myths & Concerns
 Ai Myths vs. Reality
Myth Reality Explanation

While AI automates repetitive tasks, it also

creates new opportunities and enhances
AI Will Replace All Human Jobs AI Augments Human Work human roles, leading to job transformation
rather than elimination. Microsoft News

AI processes data through algorithms and

AI Operates Differently from Human lacks consciousness or understanding,
AI Functions Like the Human Brain operating based on patterns rather than
Cognition
human-like thought. postindustria.com

AI systems function optimally with human

AI Systems Are Fully Autonomous guidance to ensure accuracy, ethical
AI Requires Human Supervision considerations, and contextual relevance.
and Require No Human Oversight
Carlson School of Management
Hallucinations
Mitigating AI Hallucinations
• Improve Training Data: Use diverse and well-curated datasets.
• Enhance Fine-Tuning: Continuously refine models for specific tasks
• Integrate Fact-Checking: Pair AI with real-time verification tools.
• Write Clear Prompts: Provide precise and unambiguous instructions.
• Encourage Iterative Refinement: Refine prompts and verify outputs step-by-
step.
• Compartmentalize your task into connected steps: How would you instruct a
student to do [x]?
• Vet your chosen AI for the task – some are better for certain tasks than
others!

Factset Insight, 2024 | Wired, 2024 | Eliot, 2024

Does it Really “Lie”?
Answer: Not purposefully.
Internet metadata refers to structured data about online content, including:

Source URLs (where information originates)

Publication dates (when it was published or last updated)
Authorship and credibility (who created it and their authority)
Engagement metrics (how often it's referenced, linked, or shared)
This metadata helps search engines and AI models categorize, rank, and retrieve information when
generating responses.

AI models, like ChatGPT, rely on vast amounts of text data scraped from the internet. However, AI doesn’t
inherently "understand" metadata the way humans do, which leads to hallucinations—plausible but false or
misleading information.
 How AI Hallucinates Misattributed Quotes
Type of AI Quote Error AI-Generated Quote Reality (Correct Attribution or Issue) How AI Hallucinates This Error

“Insanity is doing the same thing over and Incorrect – No historical record of Einstein AI associates Einstein with intelligence & science, so it
Famous People Saying
over and expecting different results.” — saying this. The quote originates from a 1981 predicts he said something about logical reasoning—
Things They Never Said Albert Einstein Narcotics Anonymous pamphlet. even though he didn’t.

“Not everything that is faced can be Incorrect – This quote actually belongs to AI links themes, not sources. Since Baldwin & MLK
Wrong Author, Right
changed, but nothing can be changed until James Baldwin, but AI associates it with MLK both spoke on justice and change, AI misattributes the
Sentiment it is faced.” — Martin Luther King Jr. because of its similar theme. quote based on conceptual similarity.

AI recognizes speech patterns and merges phrases

Merging Two Quotes “We shall fight on the beaches, and never, Incorrect – AI combined two separate
from different sources, assuming they belong
Together never, never give up.” — Winston Churchill Churchill quotes into one that never existed.
together.

AI predicts text patterns based on how historical

“The strength of a nation is measured by Incorrect – AI generated this quote from
Completely Fabricated figures spoke. Since Washington spoke about
the wisdom of its leaders.” — George scratch based on Washington’s style, but it has
Quotes leadership, AI invents plausible-sounding quotes in his
Washington no historical record.
style.

“The only thing necessary for evil to Incorrect – AI slightly changed the real AI attempts to simplify complex ideas into digestible
Paraphrasing Errors triumph is for good people to do nothing.” sentiment. Burke never wrote this exact phrase, statements. If a real quote is long or obscure, AI may
— Edmund Burke but similar variations exist in his work. paraphrase it incorrectly.
Prompt-Writing
101
Key Factors Influencing AI Responses
Prompt Quality:
• Specific, clear prompts yield better responses than vague or ambiguous ones.

Training Data:
• The knowledge and perspective embedded in the training data shape the AI’s capabilities and
limitations.

Model Size and Scope:

• Larger models with more parameters typically produce more nuanced and contextually appropriate
responses.

System Constraints:
• Filters or community guidelines influence what the AI can and cannot generate (e.g., avoiding
sensitive or harmful content).

Internet Metadata
• It’s like trying to sift through a garbage dump for pieces to a puzzle it’s assembling.
What AI Needs to Work Well

AI is not a mind reader—it relies on patterns in language and structured input to generate
relevant responses. Here are the key core ingredients to a strong prompt:
• Specificity – Vague prompts lead to vague answers. AI works best with clear, precise
details.
Clarity – Remove ambiguity. AI needs well-defined tasks and expectations.
Structure – Organizing prompts logically improves accuracy and response quality.
 The Five Key Components of a Good Prompt
Context – Provide background information to frame the task.
Example: “Use this writing sample to analyze my voice, then help me revise my essay (prompt provided) for style.”

Task – Clearly state what the AI should do.

Example: "Explain photosynthesis in simple terms."

Output Type – Define the expected format of the response.

Example: "Provide a bulleted summary."

Tone & Style – Specify the desired tone or complexity level.

Example: "Explain it in a way a 5th grader would understand."

Constraints & Rules – Set specific boundaries to refine the response.

Example: "Limit the explanation to 100 words.“

Lo, L. S. (2023). The art and science of prompt engineering: A new literacy in the information age. Internet
Reference Services Quarterly, 27(4), 203–210. https://doi.org/10.1080/10875301.2023.2227621
AI "Thinking Styles" – How Wording Changes AI’s Response
Here’s how even small changes impact AI’s approach:
❖ Directive Prompt: “Summarize this article in three bullet points.” → AI gives a concise summary.

❖ Exploratory Prompt: “Explain three different perspectives on this topic.” → AI analyzes multiple
viewpoints.

❖ Persona Prompt: “You are an instructional designer. How would you develop an e-learning course on
this topic?” → AI adapts its response to the ID role.

The more intentional you are with wording, the better AI performs.
 Effective AI Prompt Writing for Instructional Designers
Use Case Bad Prompt ( ) Good Prompt ( ) Why It Works

Brainstorming “Give me ideas for a “Generate five creative ways to introduce leadership skills in a Provides specificity (5 ideas), context (corporate
Ideas leadership course.” corporate training program.” training), and focus (leadership skills).

Summarizing “Summarize this research “Summarize this 5-page research paper in 100 words, keeping Defines length (100 words) and tailors the summary to
Content paper.” key takeaways for instructional designers.” the instructional design audience.

Assessment “Write a quiz on conflict “Create a 5-question multiple-choice quiz on conflict resolution Specifies quiz type (MCQ), number of questions (5),
Generation resolution.” in the workplace, with correct answers and explanations.” and adds explanations for learning reinforcement.

“Write a case study for a retail training program about a

Adds context (retail training), structure (conflict
Case Study “Write a case study on customer service conflict between an employee and a
scenario), and expected response types (three
Development customer service.” frustrated customer. Show three possible resolutions and their
resolutions).
outcomes.”

Interactive AI “You are a virtual tutor for an online cybersecurity course. Ask Uses persona prompting and interactive learning to
“Teach me cybersecurity.”
Prompting me questions and provide hints instead of direct answers.” engage learners.
 Framework Description Example Prompt Why It Works

“Explain instructional scaffolding.”

Guides AI to break complex
Chain-of- Forces AI to reason “Explain instructional scaffolding in
topics into logical steps,
Thought (CoT) step-by-step before a step-by-step manner. Start with a
leading to deeper, structured
Prompting responding. definition, provide a real-world
Advanced AI example, then describe its benefits in
instructional design.”
explanations.

Prompting
Frameworks for
Instructional Makes AI generate
“How can I use gamification in
corporate training?”
Encourages exploring
Designers Tree-of-Thought
multiple ideas or
approaches before “Provide three different approaches
multiple solutions, improving
(ToT) Prompting decision-making and depth
deciding on the best to gamification in corporate training.
of responses.
one. Analyze the pros and cons of each
before selecting the most effective.”

“Define microlearning.”
Uses questions instead Makes AI engage the learner
Socratic “Instead of defining ‘microlearning,’
of direct instructions to in critical thinking, improving
Prompting ask me three thought-provoking
guide AI responses. retention and active learning.
questions to help me understand it
myself.”
The “CLEAR” Framework
• Introduced by Leo S. Lo in his article The Framework Components:
Art and Science of Prompt Engineering:
A New Literacy in the Information Age ❖ Conciseness
• Exploration of prompt engineering as a
❖ Logic
critical skill in human-AI interaction
• Blends principles from artificial ❖ Explicitness
intelligence, linguistics, and user
❖ Adaptability
experience design.
❖ Reflectiveness

Lo, 2023
Component of CLEAR Framework Guiding Principles

Craft prompts that are brief and to the point, eliminating

CONCISENESS unnecessary words to prevent ambiguity.

Organize the prompt logically, ensuring the AI follows a

LOGIC clear sequence or structure.

Clearly specify the task, desired format, and scope to

EXPLICITNESS ensure precise AI responses.

Tailor prompts to the AI's capabilities and the specific needs

ADAPTABILITY of the task or audience.

Encourage the AI to assess its responses and consider

REFLECTIVENESS improvements or alternative perspectives. Also – give it
regular feedback.

Lo, 2023
 CLEAR Framework Examples

Conciseness Logic Explicitness Adaptability Reflectiveness

"Can you write a

story about a knight
in medieval times,
"Tell me about "Tell me about the
including details "Describe a good "Give me a summary
Bad Example about their daily life,
photosynthesis and
morning routine."
history of space
of this article."
why it’s important." travel."
their adventures, and
how they dealt with
challenges?"

"Explain the process "List a morning "Provide a summary

of photosynthesis routine for increased "Explain the history of this article and
"Write a story about
step-by-step, then productivity, of space travel in include any areas
Revised with a knight's adventure,
describe its including three simple terms for a where additional
CLEAR principles focusing on a single importance in specific tasks and 10-year-old information might
challenge they face."
sustaining estimated times for audience." improve
ecosystems." each." understanding."
Practical Techniques
FUNDAMENTAL: SITUATIONAL: ENRICHES OUTPUTS:

❖ Be Clear and Specific ❖ Use Role-Based Prompts ❖ Encourage Creativity or

❖ Include Constraints Exploration
❖ Provide Context

❖ Be Concise ❖ Experiment with Few-Shot ❖ Test and Iterate

Examples
❖ Structure Prompt Logically ❖ Reflect and Review

❖ Specify Desired Format ❖ Experiment with Prompt

Variations
❖ Tailor For Audience
❖ Be nice

Giray, 2023 | Lin, 2023

 FUNDAMENTAL Prompt-Writing Techniques
Technique Explanation

Be Clear and Specific Use precise language to avoid ambiguity.

Provide Context Give background or role details for clarity.

Structure the Prompt Logically Organize tasks in a logical sequence.

Specify the Format Indicate desired response structure.

Tailor for the Audience Adjust tone and complexity to the audience.

Be Concise Keep prompts short and to the point.

Giray, 2023 | Lin, 2023

 Examples: Fundamental Techniques
Technique Bad Example Corrected Example
Explain photosynthesis, focusing on its
Be Clear and Specific Tell me about photosynthesis.
role in ecosystems.

Explain surplus spending as it relates to

Provide Context Explain surpluses.
ethical business practices.

Structure the Prompt Write a step-by-step guide on AI First, describe AI ethics; then give
Logically ethics. examples of its applications.

List three benefits of renewable energy

Specify the Format Write about renewable energy.
in bullet points.

Describe blockchain technology for a

Tailor for the Audience Describe blockchain technology.
10-year-old audience.

What’s quantum computing? (too Briefly define quantum computing in

Be Concise ambiguous) simple terms.

Giray, 2023 | Lin, 2023

 SITUATIONAL Prompt-Writing Techniques

Technique Explanation

Use Role-Based Prompts Frame the response with a role or perspective.

Include Constraints Set limits on length, style, or content.

Experiment with Few-Shot

Prime AI with example tasks and outputs.
Examples

Giray, 2023 | Lin, 2023

 Examples: SITUATIONAL Prompt-Writing Techniques
Technique Bad Example Corrected Example

Act as a high school science teacher.

Use Role-Based
Explain Newton’s laws of motion. Explain Newton's laws of motion to 10th
Prompts graders.

Give me a detailed answer about AI in Write a 100-word answer summarizing

Include Constraints AI in medicine.
medicine.

Here are two examples of how I want the

instructional content to look:
Experiment with Make a quiz on cellular
• A brief definition of photosynthesis
Providing Examples and respiration and make a similar followed by a real-world example.
Modeling one for photosynthesis. • A short quiz with three questions.

Now create similar content for cellular

respiration.
Giray, 2023 | Lin, 2023
 Techniques to Enrich AI Outputs

Technique Explanation

Encourage Creativity or Exploration Inspire novel ideas or brainstorming.

Test and Iterate Refine prompts iteratively based on output.

Reflect and Review Prompt AI to self-evaluate or expand responses.

Experiment with Prompt Variations Rephrase prompts to compare outputs.

Combine Techniques (Chaining

Use follow-ups to refine or expand results.
Prompts)

Giray, 2023 | Lin, 2023

 Techniques to Enrich AI Outputs: Examples
Technique Explanation Practical Use for IDs
Evaluate instructional materials:
Prompt AI to self-evaluate or - Does this align with Bloom's
Reflect and Review Taxonomy?
expand responses.
- How can this content be more
accessible?
Test content approaches:

Experiment with Prompt Rephrase prompts to compare - “Compare content styles” (e.g.,
engaging vs. humorous)
Variations outputs.
- Test different formats: “Use bullet
points vs. prose”
Develop interconnected materials:

- Step-by-step course development

Combine Techniques (Chaining Use follow-ups to refine or expand
FUNDAMENTAL
Prompts) TECHNIQUES - EXAMPLES
results. - Content creation pipeline

- Scaffolded learning paths

Giray, 2023 | Lin, 2023
 Techniques to Enrich AI Outputs: Examples
Technique Explanation Practical Use for IDs

Brainstorming creative lesson

ideas or designing engaging
Encourage Creativity or Inspire novel ideas or
activities that inspire novel
Exploration brainstorming. teaching methods or student
participation.

Iteratively refine content:

Refine prompts iteratively - “Rewrite learning objective to

Test and Iterate make it SMART.”
based on output.
- “Make these quiz questions more
challenging.”

Giray, 2023 | Lin, 2023

It’s not weird to be polite to the robots.
Waseda University and Riken Center for Wall Street Journal:
Advanced Intelligence Project:
❖ Strategic questioning enhances AI
❖ Courteous prompts lead to more accurate responses.
and detailed responses.
❖ Role-assuming prompts yield more detailed
❖ Prompts ranked 7 out o 8 in politeness and engaging explanations.
elicited higher-quality outputs.
❖ Framing prompts politely also yields more
❖ Overly deferential prompts (8 out of 8) were detailed and engaging explanations.
less effective.
❖ Guiding AI to break down problems step-by-
❖ The AI is guided to mimic high-quality step is proven effective for methodical
discourse. responses.

Yin, Wang, Horio, Kawahara, & Sekine, 2024

Ethics and a
Framework for
Responsible
Implementation
Challenges and Opportunities Specific to Instructional Design

Challenges:
 Algorithmic bias in AI outputs.
 Skill gaps in prompt-writing for IDs.
 Uneven access to AI tools.

Opportunities:
 Train IDs in ethical AI and inclusive prompt-writing.
 Collaborate with small, often overlooked communities for authentic materials.
 Scale justice-oriented learning environments.
 Never a better time to invest in your team: Professional development
opportunities.

Constanza-Chock, 2020 | Olla et al., 2024

 THE SPARRO FRAMEWORK (Olla et al., 2024)
Component What It Does Application for IDs

Encourages planning how AI will be used in

Strategy assignments or projects, including defining its role,
limitations, and alignment with objectives.
Determine when and how AI tools should support
course design or training programs.

Prompt Leverages the CRAFT model (Clarity, Rationale,

Audience, Format, Tasks) to craft effective prompts
Helps generate precise prompts that result in AI outputs
closely aligned with course objectives.
Design tailored to specific needs.

Focuses on seamlessly integrating AI-generated Refine and contextualize AI-generated materials to

Adopting content into deliverables, ensuring it aligns with
project objectives and the user’s unique voice.
align with institutional branding, learning outcomes, and
the intended audience.
 THE SPARRO FRAMEWORK (Olla et al., 2024)
Component What It Does Application for IDs

Involves critically assessing AI outputs for accuracy, Ensure quality and accuracy in AI-generated materials
Reviewing relevance, and coherence, comparing them against
reliable sources.
before they are delivered to learners, e.g. fact-checking
content, consistency with educational standards.

Encourages iterative improvements to AI outputs, Polish AI-generated lesson plans, scripts, or activities,
Refining adding insights, restructuring content, and enhancing
language.
ensuring they meet instructional goals and resonate
with the target audience.

Focuses on maintaining academic integrity by Use plagiarism checkers and reference tools to confirm
Optimizing ensuring originality and proper citation of AI-
generated content.
the credibility of AI-generated materials, ensuring
adherence to institutional policies.
Goals for Everyone
• Address and mitigate bias in AI-generated content.

• Always review AI-generated content with human

oversight.
• AI Policy Updates: Stay informed on evolving
regulations.
• Stay informed on developments – they happen faster
than you think!
• Inspire confidence and prepare IDs for the next wave
of AI tools.
REFERENCES
• Bates, M. J. (1999). The invisible substrate of information science. Journal of the American Society for Information Science, 50(12), 1043–1050.
https://pages.gseis.ucla.edu/faculty/bates/substrate.html
• Carlson School of Management. (n.d.). Debunking 5 artificial intelligence myths. University of Minnesota.
https://carlsonschool.umn.edu/graduate/resources/debunking-5-artificial-intelligence-myths
• Chen, C.-M., & Chiu, P.-S. (2015). The learning style-based adaptive learning system architecture. ResearchGate.
https://www.researchgate.net/publication/344169047
• Ch’ng, L. K. (2023). How AI makes its mark on instructional design. Asian Journal of Distance Education, 18(2), 32-45.
https://files.eric.ed.gov/fulltext/EJ1410083.pdf
• Costanza-Chock, S. (2020). Design justice: Community-led practices to build the worlds we need. The MIT Press.
• Eliot, L. (2024, December 3). Breakthrough in preemptive detection of AI hallucinations reveals vital clues to writing prompts that keep
generative AI from freaking out. Forbes. Retrieved December 5, 2024. https://www.forbes.com/sites/lanceeliot/2024/12/03/breakthrough-in-
preemptive-detection-of-ai-hallucinations
• FactSet Insight. (2024, June 17). AI strategies series: 7 ways to overcome hallucinations. Retrieved December 5, 2024.
https://insight.factset.com/ai-strategies-series-7-ways-to-overcome-hallucinations
• Gibson, J. (2023). 10 ways artificial intelligence is transforming instructional design. EDUCAUSE Review.
https://er.educause.edu/articles/2023/8/10-ways-artificial-intelligence-is-transforming-instructional-design
• Giray, L. (2023). Prompt engineering with ChatGPT: A guide for academic writers. Annals of Biomedical Engineering, 51(12), 2629–2633.
https://doi.org/10.1007/s10439-023-03272-4
• Hariri, W. (2023). Unlocking the potential of ChatGPT: A comprehensive exploration of its applications, advantages, limitations, and future
directions in natural language processing. arXiv. https://doi.org/10.48550/arXiv.2304.02017
• Joshi, R. (2023). Adaptive learning through artificial intelligence. ResearchGate. https://www.researchgate.net/publication/372701884
• Lin, Z. (2023). Ten simple rules for crafting effective prompts for large language models. ResearchGate.
https://www.researchgate.net/publication/373809003
• Lo, L. S. (2023). The art and science of prompt engineering: A new literacy in the information age. Internet Reference Services Quarterly, 27(4),
203–210. https://doi.org/10.1080/10875301.2023.2227621
• Luo, T., Muljana, P. S., Ren, X., & Young, D. (2024). Exploring instructional designers’ utilization and perspectives on generative AI tools: A mixed
methods study. Education Tech Research Development.
• Microsoft. (2023, November 13). 4 misconceptions about AI. Microsoft Source. Retrieved from https://news.microsoft.com/source/features/ai/4-
misconceptions-about-ai/
REFERENCES
• National Institutes of Health (NIH) - PMC. (2022). The digital divide and the role of the internet in education: A systematic review. Journal of
Digital Literacy and Education Research, 10(2), 56-72. https://pmc.ncbi.nlm.nih.gov/articles/PMC9008096/
• National Institutes of Health (NIH) - PMC. (2023). Ethical concerns in AI adoption for education: A review of emerging risks. Journal of AI Ethics
and Learning, 12(3), 45-60. https://pmc.ncbi.nlm.nih.gov/articles/PMC11087970/
• National Institutes of Health (NIH) - PMC. (2024). The impact of AI-driven education on digital accessibility and equity. Journal of Educational
Technology Research and Innovation, 15(1), 89-102. https://pmc.ncbi.nlm.nih.gov/articles/PMC10990165/
• Noble, S. U. (2018). Algorithms of oppression: How search engines reinforce racism. NYU Press.
• Olla, P., Elliott, L., Abumeeiz, M., Mihelich, K., & Olson, J. (2024). Promptology: Enhancing human–AI interaction in large language models.
Information, 15(634). https://doi.org/10.3390/info15100634
• Park, J., & Choo, S. (2024). Generative AI prompt engineering for educators: Practical strategies. Journal of Special Education Technology.
• Pew Research Center. (2014, December 8). More information yields more learning and sharing. Pew Research.
https://www.pewresearch.org/internet/2014/12/08/more-information-yields-more-learning-and-sharing/
• Postindustria. (n.d.). What are common misconceptions about AI & machine learning? Retrieved from https://postindustria.com/what-are-
common-misconceptions-about-ai-machine-learning/
• SpringerOpen Educational Technology Journal. (2021). Artificial intelligence and education: A critical review of the literature. International
Journal of Educational Technology in Higher Education, 18(1), 1-21.
https://educationaltechnologyjournal.springeropen.com/articles/10.1186/s41239-021-00292-9
 Helpful Websites/Applications
❖ LLM Leaderboard - Verified AI Rankings
• Helps with vetting models for projects and tasks
• Larger models = more diverse data = best option for
general instruction

Q&A ❖ HuggingFace - An AI Marketplace

❖ AIOne
• A free ChatGPT, Gemini, and Claude “All-in-one”
roach.jenny.e@gmail.com
interface.
Evanston, IL
❖ Creative Writing Applications
• NovelAI, SudoWrite