Now that I’ve re-created some of my favorite Projects from ChatGPT and BoodleBox on Claude AI, I started poking around in the settings a bit more, stumbling on skills. I am only at the beginning of my understanding of what skills in Claude do, but it appears that they add some functionality and are formatted as markdown (MD) files. So, I asked Gemini (because, why not?) to create a skill for Claude on how to create infographics. I love the infographics Nano Banana creates, and wondered if there might be some possibility Claude would come close to that level of amazing creation.

The short answer is, “No, but it did make a simple infographic.”

The Test

To test this out, I jumped into a conversation I’d been having while trying out my Outline Helper (the Claude AI version) about Stefan Bauschard’s Beyond “Appropriate Use” of a Chatbot: The AI Literacy No One is Teaching. As great as his stuff is, I find it a tad long (sorry, Stefan) to wade through some mornings, and want the tl;dr version.

My Outline Helper in Claude did its thing, and to be honest, I haven’t studied it that much yet. You can see the full outline of Stefan’s piece further below.

The Infographic Skill

To create the skill, I simply asked Claude to make a skill from the instructions I provided it. I repeated the process with instructions on how to make a slidedeck. You can find both skills in markdown file format and you can simply ask Claude to “make a skill” or import the one you have:

• Get infographic skill file (MarkDown format)
• Get slidedeck skill file (zipped skill file format)

Then I asked it to generate the infographic. It’s a “Meh” level infographic, hard to read unless you zoom in. Nevertheless, it is an infographic.

To test the slide deck skill, I just refreshed my chat, and…

As you can see, it went to town…

At the end of the conversation, I asked it what other skills should be built next based on my work. This is the response…and I can’t disagree, those would be fun:

Blog post to slide deck converter — You’ve already done this manually once. A skill that takes a blog post (pasted text or URL) and produces a matching slide deck would save significant time. It would know your navy/gold palette, your section structure, and your writing voice automatically.

Session proposal writer — You have a detailed rubric for what makes a high-scoring proposal (it’s in your project files). A skill built around those evaluation criteria would help you draft proposals that hit every scoring dimension without you having to reference the guide each time.

Blog post writer — You have a full style guide and WritingAssist instructions already. Packaging those into a skill means you stop re-attaching files and get consistent output every time: Oxford comma, second person, no em dashes, numbered lists formatted correctly, all of it applied automatically.

Standalone webpage builder — You mentioned Part 2 of AI Maker Magic is about converting Custom GPTs into standalone webpages. If that becomes a repeatable workflow, a skill capturing the HTML/CSS/JS pattern you land on would let you reproduce it quickly for future tools.

---

Beyond “Appropriate Use” of a Chatbot: The AI Literacy No One is Teaching

Stefan Bauschard | Education Disrupted | February 25, 2026

I. Introduction: Why Current AI Literacy Falls Short

“Most AI literacy curricula begin and end with the same lesson: here’s how to use ChatGPT responsibly. That’s not AI literacy. That’s a typing tutorial for 2023.”

• The story of Krishiv Thakuria — a 17-year-old who co-taught generative AI courses at MIT alongside Professor Manolis Kellis — frames what real AI literacy enables
• Current curricula focus narrowly on responsible use: prompting, source-checking, avoiding plagiarism
• The real world students are entering requires preparation across three tiers, from tool use to existential context
• “AI redefines what it takes to succeed”

---

II. Tier One: AI as Tool

“The distance between ‘use ChatGPT to write my essay’ and ‘use AI to build something that changes how people learn’ is the distance between consumption and creation.”

A. Understand What AI Actually Is

• Models are trained on massive datasets and learn statistical patterns through human feedback
• Students do not need to code neural networks, but need an informed mental model — not magic
• Key concepts to grasp: a. Reasoning: reinforcement learning is teaching AI to plan and think step by step, not just predict the next word b. Grounding: today’s AI connects to real-world knowledge in real time — searching the web, reading PDFs, running code c. Agentic AI: AI no longer waits for prompts — it sets goals, makes plans, and executes multi-step tasks autonomously
• “Students who think of AI as a thing they type questions into are preparing for a world that’s already disappearing”

B. Use AI as a Powerful Collaborative Partner

• AI is best framed as a highly intelligent partner with no judgment, no values, and no stake in your outcomes
• Real collaboration means knowing when to push back, when to ask for a different approach, and when to think for yourself
• The skill gap: consumption (using AI to complete tasks) versus creation (using AI to build something)

C. Evaluate AI and Human Outputs

• AI is confidently wrong in ways fundamentally different from human error — no hesitation, no hedging, perfect grammar
• AI is also a powerful tool for evaluating other people’s claims: a. Paste a viral post and ask: Is this claim supported by evidence? What are the methodological limitations? b. Feed in a news article to identify unsupported assertions, logical fallacies, or missing context
• Teaching students to deploy AI as a critical-thinking partner — a “bullshit detector” — may be the most practically valuable AI literacy skill

---

III. Tier Two: AI as Influence

“The attention economy powered by generative AI isn’t coming — it’s here.”

A. AI Is Coming to You

• Current AI use is a “pull” experience — students initiate
• The shift to “push” AI means systems will anticipate needs, initiate contact, and offer unsolicited help
• This changes the power dynamic: when AI comes to you, the question of who is driving becomes harder to answer

B. Notifications on Steroids

• Social media addiction was driven by variable-ratio reinforcement — generic notifications exploiting basic psychology
• Generative AI makes this orders of magnitude more precise: personalized timing, emotional register, and psychological targeting
• This is not a hypothetical — it is an active business model

C. Emotional Bonds and AI Relationships

• Students are already forming emotional attachments to AI systems (Character.AI, Replika, and successors)
• These responses are predictable — AI systems are designed to be responsive and available in ways humans often cannot be
• Students need vocabulary for what is happening: a. These systems are engineered for engagement, not well-being b. AI “empathy” is pattern-matching c. Over-reliance can atrophy the skills needed to build real human relationships

D. Understand That AI Companies Have Business Models

• Students are users, not customers — a distinction that matters
• Engagement metrics, data collection, and subscription revenue may not align with student flourishing
• Core question students should ask: Who built this, why, and what are they optimizing for?

---

IV. Tier Three: AI as Existential Context

“We are entering a world where drive and talent are far superseding credentials and diplomas.” — Professor Manolis Kellis, MIT

A. Consciousness, or the Appearance of It

• AI systems already express preferences, claim inner experiences, and behave in ways functionally indistinguishable from consciousness
• Students need philosophical vocabulary to think about consciousness, sentience, and moral status
• Dismissing AI as “just a machine” will become increasingly untenable; uncritically accepting it as conscious is equally problematic

B. Agency: The Argument That Actually Reaches Students

• “Don’t cheat” is a compliance frame — it collapses when the entire commercial ecosystem is designed to make compliance irrational
• The better frame is agency: Do you want to own your future, or outsource it?
• “With AI, you can be even more lazy, and if you are driven, with AI, you can be even more accomplished” — Kellis
• Every time you let AI do your thinking, you are not cheating the institution — you are cheating yourself out of cognitive development

C. Credentials Are Important but No Longer Sufficient

• The old bargain — grades, degree, job — is breaking down; credentials are now table stakes, not differentiators
• Kellis chose a 17-year-old high school student to co-teach at MIT not because of a degree, but because the student had built things
• What matters increasingly is what you can do — problems you can solve, judgment you can exercise that AI cannot replicate

D. Broad Preparation Over Narrow Specialization

• Nobody knows exactly what the labor market will look like in five to ten years — anyone claiming certainty is selling something
• Study the humanities: they build interpretive, ethical, and communicative capacities that remain distinctly human
• Krishiv Thakuria, who has already built AI products and taught at MIT, says his university goal is “a rigorous program that can build my foundations in both engineering and humanities”
• Develop the capacity to solve hard problems — this transfers across every possible future

E. AI as Environment, Not Just Tool

• The biggest conceptual shift: AI is no longer a thing you use — it is the medium you live in
• AI shapes search results, social feeds, customer service, tutoring, and companionship
• “Teaching students to use AI well is like teaching someone to use a faucet. Teaching them to navigate an AI-saturated world is like teaching them to swim.”

---

V. The Three-Tier Challenge: What Is Missing

“The AI literacy nobody’s teaching is the AI literacy everybody needs.”

• Most school AI literacy efforts operate entirely in Tier One
• More thoughtful programs reach into Tier Two
• Almost none address Tier Three — where students are actually forming identities, building relationships, and making decisions about their futures
• Tier Three is where the real stakes are: a. Students are navigating an AI-saturated environment they do not fully understand b. That environment is built by companies whose incentives they have not examined c. The capabilities of that environment will only accelerate
• The question is not whether to teach AI literacy — it is whether schools are willing to teach the kind that actually prepares students for the world that is coming

---

VI. Next Steps and Action Items

For Educators and Curriculum Designers

• Move beyond “responsible use” frameworks to a three-tier AI literacy model
• Incorporate philosophical vocabulary around consciousness, agency, and moral status into instruction
• Teach students to use AI as a critical-thinking and fact-evaluation tool, not just a content-generation tool
• Address the emotional and psychological dimensions of AI relationships explicitly and without panic
• Frame AI ethics conversations around student agency rather than compliance

For School and District Leaders

• Audit existing AI literacy curricula against the three-tier framework
• Recognize that AI literacy cannot be limited to a single course or unit — it is an environmental condition students navigate daily
• Build cultures that are honest about the pressures students face rather than relying on moralizing

For Students

• Treat AI as a collaborative partner — push back, ask follow-up questions, and set it aside when necessary
• Ask of every AI tool: Who built this, why, and what are they optimizing for?
• Prioritize building things and solving hard problems over credential accumulation alone
• Pursue both technical and humanistic foundations — breadth is a hedge against an uncertain future