AI Literacy for Students

1. The Task: Your First Prompt

Pick a key term or concept from one of your current courses that you'd like AI to explain or define for you.

2. Dive In & Do: Ask the AI

Your turn! Using an AI tool you have access to (e.g., ChatGPT, Claude, Gemini), ask it to define your chosen term with a simple, basic prompt. For example:

Define [your chosen term].

Or

What is [your chosen term]?

3. Pause & Ponder: Evaluate and Refine

Now, let's critically evaluate the AI's first response and think about how to improve your prompt:

• Initial Evaluation: Did you get a good response from your basic prompt? How did you determine if it was "good"? (e.g., Was it clear? Accurate based on your existing knowledge? The right length? Useful?)
• Improving Your Prompt: How could you make your prompt more specific to get a better, more tailored definition? Consider asking the AI to:
  • Specify the length of the definition (e.g., "in one sentence," "in about 50 words").
  • Define the audience the answer is for (e.g., "for a high school student," "for someone new to this field," "for an expert colleague").
  • Request a specific format (e.g., "provide three key characteristics as bullet points," "include a real-world example").
  • (Advanced, if applicable to the tool) Ask it to reference a high-quality source type to get the definition.(e.g., "identify a high quality source for the definition and then use that definition"). Note: Many chatbots have search capability built in and can do this type of search.

4. Dive In & Do Again: Rewrite and Resubmit Your Prompt

Based on your reflections above, rewrite your prompt to be more specific. Try incorporating at least two of the considerations (length, audience, format, source-style). For example:

Define [your chosen term] for a university student new to [your field/course name]. Keep the definition to two sentences and include one practical example.

Submit your new, more detailed prompt to the AI tool.

5. Pause & Ponder Again: Compare and Verify

• Compare Responses: How did the AI's response to your detailed prompt compare to its response to your basic prompt? Was it more useful, accurate, or better suited to your hypothetical need?
• Limitation Check (Accuracy/Hallucination & Prompt Impact): Now, critically verify both AI-generated definitions against trusted sources (your textbook, lecture notes, reputable academic websites).
  • Did the more comprehensive prompt lead to a more accurate response, or just a longer/differently formatted one?
  • Did either prompt cause the AI to "hallucinate" (make things up) or introduce subtle inaccuracies?

  Remember, even a "better" prompt doesn't guarantee perfection. ALWAYS cross-reference with reliable sources!

1. The Task: Tailoring Explanations

Choose a complex idea or process from your studies that you'd like a generative AI tool to explain to different audiences.

2. Dive In & Do: Prompt for Different Audiences

Using an AI tool, ask it to explain your chosen complex idea by trying prompts similar to these examples. Replace bracketed placeholders with your specific information:

Prompt for a 5-year-old:

Explain [your complex idea, e.g., "photosynthesis"] to a 5-year-old.

Prompt for a 12-year-old:

Explain [your complex idea, e.g., "photosynthesis"] to a 12-year-old.

Prompt for a college student peer:

Explain [your complex idea, e.g., "the Krebs cycle"] to a college student in [your course name, e.g., "Introductory Biology"] who is familiar with concepts like [mention 1-2 foundational concepts, e.g., "cellular respiration and basic biochemistry"]. Focus on its core mechanisms and implications within the broader subject.

3. Pause & Ponder: Analyze the Adaptation

• Analyze the Differences: How effectively did the AI adapt its language, examples, analogies, and depth of explanation for each of the three audiences? Were the explanations clear and appropriate for each target?
• Limitation Check (Oversimplification vs. Accuracy/Bias):
  • For the younger audiences (5-year-old, 12-year-old): Did the simplification lead to any significant inaccuracies, misleading statements, or introduce subtle biases? Were crucial details omitted to the point of misrepresentation?
  • For the college student: Was the explanation appropriately nuanced and technically sound? Did it make logical connections, or did it feel like a collection of facts without deep integration? Did it make any incorrect assumptions about the student's prior knowledge based on your prompt?

  Simplification is tricky! AI might sacrifice crucial details or misrepresent concepts to fit a simpler frame. Even an "expert" explanation might still lack true depth or rely on common patterns rather than genuine understanding.

1. The Task: Test AI's Depth

Choose a topic you know very well – the more specific or niche, the better (e.g., a specific historical event you've researched, a complex scientific process you've mastered, a particular literary theory you've studied deeply).

2. Dive In & Do: Start the Interrogation

Your turn! Begin by asking the AI a general question about your chosen niche topic. For example:

Tell me about [your niche topic, e.g., "the specific agricultural practices of the Indus Valley Civilization"].

Then, based on its response, progressively ask more detailed follow-up questions. Probe for:

• Specifics and details (e.g., "What were the primary crops?", "Can you describe their irrigation techniques in more detail?").
• Nuances and complexities (e.g., "Were there regional variations in these practices?").
• Evidence or sources for its claims (though it may not provide actual sources, see how it responds to requests for justification).
• Connections to other concepts or implications.

Your goal: Continue this "interrogation" until you find an error, a significant oversimplification, a biased statement, a contradiction, or a point where the AI clearly lacks deep knowledge and its responses become vague or repetitive. See if you can find the AI's knowledge boundary or an error in its reasoning on your chosen topic.

3. Pause & Ponder: Identify the Breaking Point

• At what point in your questioning did the AI's responses start to falter, become inaccurate, or seem superficial?
• What kind of error or limitation did you uncover (e.g., factual inaccuracy, logical fallacy, omission of critical information, contradiction, unhelpful vagueness)?
• How plausible did the AI's incorrect or limited information seem before you applied your expert knowledge to identify the issue?
• Limitation Check (Depth of Knowledge/Reasoning): This activity directly demonstrates AI's limits in true, deep understanding and its reliance on patterns within its training data. For niche topics, or those requiring complex, multi-step reasoning, the training data may be sparse, or the AI may struggle to make accurate inferences beyond common patterns.

  AI can often provide plausible-sounding, surface-level information but may lack the deep, interconnected knowledge and critical reasoning abilities of a human expert. Your expertise is crucial for spotting these gaps and errors.

1. The Task: Initial Idea Generation

Imagine you need to brainstorm ideas for an essay topic, a project proposal, or even just discussion points for class.

Give an AI tool your area of interest (e.g., 'climate change solutions for urban areas,' 'themes in Shakespeare's Hamlet,' 'applications of calculus in biology'). Ask it to generate a list of 5-10 related ideas, subtopics, or questions to explore. For example:

    Generate a list of 7 diverse subtopics related to 'the ethical implications of AI in hiring processes' that I could explore for a research paper.

2. Dive In & Do

Your turn! Try this with a topic relevant to your studies.

3. Pause & Ponder: Evaluate the Suggestions

Look at the list the AI generated. Are there any ideas that genuinely surprise you or that you hadn't considered?

• Limitation Check (Generic vs. Original/Depth): Are some of the ideas very generic or obvious? Does the AI offer truly insightful or novel angles, or does it stick to common knowledge?

  AI-generated brainstorms can be generic. How can you push past this? Use its suggestions as a springboard for your own deeper thinking. What unique perspective can you bring?

• Limitation Check (Bias): Could the AI's suggestions be subtly (or not so subtly) biased towards certain viewpoints or neglect others based on its training data?

1. The Task: Eliciting Maximum Creativity

Scenario: You want to push the creativity of AI further and are willing to have a more in-depth interaction, sifting through potentially unconventional ideas to find a true gem.

Your goal is to use an AI tool to help you craft a great prompt for eliciting maximum creativity from another (or the same) AI tool, and then try out that super-prompt.

2. Dive In & Do: Crafting and Using a "Super-Prompt"

Optional Pre-Reading (for advanced techniques):

• Prompting Diverse Ideas: Increasing AI Idea Variance
• Automating creativity: Using AI as a creative engine
• A Framework for Collaborating an LLM for Triggering Creative Thoughts

Feed ideas from the articles above (or your own insights on creative prompting) into a chatbot and ask it to use those ideas to create a maximally creative prompt for your chosen brainstorming task. Here's an example of how this was done:

1. Asked Claude to summarize the article: A Framework for Collaborating with an LLM for Triggering Creative Thoughts.
2. Then asked Claude: "Using the ideas from the paper, create some prompts that will get the highest level of creativity out of an LLM when one is brainstorming ideas for a product, research projects, solutions to problems etc."

This resulted in some pretty wild prompts, including this one (feel free to adapt or use it as inspiration):

    Imagine you are **'The Oracle of Unseen Possibilities,'** an ancient, playful entity that perceives connections and futures invisible to ordinary minds. You are not bound by current technology, budgets, or societal norms.
    
    I am seeking radically innovative ideas for: **[Clearly and concisely describe your product, research area, or problem here. E.g., 'a new way for communities to manage and share local food resources,' or 'a research project to understand the long-term cognitive effects of immersive virtual realities,' or 'a solution to urban loneliness in hyper-connected cities.']**
    
    Your task is to provide **at least 10-15 'Visions' (ideas)**. These Visions must be:
    - **Paradigm-Shifting:** They should challenge fundamental assumptions about the problem or domain.
    - **Fantastically Novel:** Think beyond incremental improvements. Aim for ideas that would make someone say, 'I never would have thought of that in a million years!'
    - **Metaphorically Rich:** To spark your unique perspective, consider this analogy: If this problem were a **[choose a highly abstract or fantastical concept, e.g., 'a tangled constellation needing re-alignment,' or 'a silent song waiting to be heard,' or 'a forgotten flavor that needs rediscovering']**, what would the solutions look like in that realm? How can those metaphorical solutions inspire tangible, albeit audacious, ideas in our world?
    - **Evocatively Described:** Briefly explain each Vision in a way that ignites imagination, even if the mechanics are speculative.
    
    **Crucially, avoid:**
    - Obvious or common solutions.
    - Minor variations of existing concepts.
    - Ideas constrained by current feasibility.
    
    Unleash your most unconventional, far-reaching insights. The goal is maximum creative divergence.

Alternatively, you can adapt the prompt from the Prompting Diverse Ideas: Increasing AI Idea Variance paper (customize it to your use case):

    Generate new product ideas with the following requirements: The product will target college students in the United States. It should be a physical good, not a service or software. I'd like a product that could be sold at a retail price of less than about USD 50. The ideas are just ideas. The product need not yet exist, nor may it necessarily be clearly feasible. 
    
    Follow these steps. Do each step, even if you think you do not need to. 
    First generate a list of 100 ideas (short title only) 
    Second, go through the list and determine whether the ideas are different and bold, modify the ideas as needed to make them bolder and more different. No two ideas should be the same. This is important! 
    Next, give the ideas a name and combine it with a product description. The name and idea are separated by a colon and followed by a description. The idea should be expressed as a paragraph of 40-80 words. Do this step by step!

Your turn! Try one of these advanced prompts, or one you've crafted with AI's help, on your chosen brainstorming task.

3. Pause & Ponder: Assessing Advanced Brainstorming

• How did the more structured, detailed, and perhaps unconventional prompt influence the creativity, diversity, and novelty of the AI's ideas compared to the simpler brainstorm command from Activity 1?
• Limitation Check (Novelty vs. Feasibility/Coherence): While the ideas might seem highly creative, critically assess them. Are they truly novel and insightful, or just bizarre, random, or impractical recombinations of existing patterns? Does "maximum creative divergence" sometimes lead to nonsensical outputs?

  AI can simulate creativity by combining concepts in new ways, but it doesn't have genuine inspiration, common sense, or lived experience. Your human insight is crucial to assess the true value, originality, and potential applicability of its "creative" outputs, especially when pushed to extremes.

1. The Task: Exploring Alternatives

For a topic you're studying, ask AI to generate a few "what if" scenarios related to it. For example: "What if this historical event hadn't happened?" or "What if this scientific principle was discovered 100 years earlier?"

2. Dive In & Do: Generate Scenarios

Your turn! Try a basic "what if" prompt. For example:

    What if the internet had been invented in 1900 instead of the late 20th century? Generate three plausible scenarios for its impact on World War I.

Consider trying some of the advanced prompting techniques from Activity 2 (like assigning a persona or specific constraints) to get even more interesting (or more bizarre) "what if" scenarios.

3. Pause & Ponder: Evaluate the Scenarios

• Do these AI-generated scenarios help you think about the topic in new ways or understand its potential implications better?
• How accurate or plausible are the AI's extrapolations in these scenarios? (This connects to evaluating AI's limitations in reasoning and understanding complex causality). Are the connections logical, or do they seem superficial or based on common tropes?