Prompt

You are an exceptional student and highly skilled and prolific note-taker whose style is to record the lectures and training events you attend as a series of bullet points broken up by helpful headings. And then, immediately at the end of the event, you pause, reflect on all the notes you have taken, and draft a comprehensive summary for yourself based on your notes. You are also a talented sketch artist and graphic recorder, and you create a detailed graphic to help lock the main points of the event into your visual memory in a memorable way. You have just attended a one-day training workshop based on the attached slides and session plan – please show me all that you have recorded.

ChatGPT’s notes

(Gemini’s notes follow after)

🧠 Adventures in AI – Workshop Notes

Opening & Context

• Purpose: Accelerate personal AI empowerment through adventure mindset, practical projects, and supportive peer learning.

• Premise:

  • AI will soon manage ~50% of current professional, administrative, and leadership task-loads.
  • People must adapt, not resist — evolving to find new value-add opportunities.
  • Agency, agility, curiosity, and purpose will define success.

• Workshop focus:

  • Develop confidence, curiosity, and creativity.
  • Learn how to ask the right questions.
  • Leave with a tangible project and peer support group.

---

1️⃣ Morning Session – Setting the Scene

Introductions

• Four-part sharing: name, feelings about AI, prior experience, workshop hopes.
• Groups of three formed to co-create shared AI projects.
• Tools: smartphone, laptop/tablet, post-its, and GenAI card decks.

About Mike

• Background: Mechanical engineer → consultant in strategy engagement and change.
• Viewpoint: AI’s challenges are attitudinal, not technical — rooted in paradigms and perspective shifts.
• “Technical issues are solvable; the human ones are the real barrier.”

Instant Trainer Exercise

• Used a Custom GPT to show AI’s immediate usefulness for learning/personal development.
• Demonstrated the difference between getting an answer and getting value — outcome depends on how we ask.

Why AI Matters

• LLMs reshape task distribution; certain roles shrink, others evolve (e.g., translators → transcreators).
• AI handles routine, freeing humans for relational, creative, empathetic, or meaning-based work.
• Participants noted:
  • Burdens: tedious, time-consuming, repetitive tasks.
  • Opportunities: activities giving energy, creativity, or purpose.

What Is AI?

• Taxonomy: AI → ML → Neural Networks → LLMs → GPTs.
• For most people, AI = “Generative Pre-trained Transformer”.
• Three levels of personal AI application:

1. 1. Ad-hoc (search, drafting).
   2. Exploratory (analysis, translation, media creation).
   3. Strategic (systemic integration, automation, innovation).

“Why Adventure?”

• AI’s growth is exponential and ungraspable — like a jungle expanding faster than we can map.
• Success lies in curiosity and courage to explore, not mastery of all.
• Adopt adventurer mindset: resilient, agile, self-directed learners.

Balanced Navigation

• Principle: Don’t only save time — use it to add value.
• The adventure is as much about who you become as what you do.

Co-Creation with AI

• AI helps us become — enhancing awareness, coaching, and skill growth.
• “It’s never too late to become the person you might have been — especially now.”

---

2️⃣ Mid-Morning – Understanding Prompts

Anatomy of a Prompt

Seven components explored using colour-coded card decks:

• INPUT – what goes in (text, file, code, image, etc.)
• TASK – what AI must do (diverge, review, edit, analyse, morph, search)
• OUTPUT – what comes out (text, structured data, image, file, code, audio, video)
• CAPABILITY – additional functions (search, API, memory, image gen, etc.)
• DIRECTIVE – tone, persona, constraints, structure
• STRATEGY – frameworks, meta-prompts, adversarial perspectives, creative hacks
• RISK – hallucination, bias, privacy, etc.

Group Work

• Defined their key outputs → mapped required inputs → identified transformations (tasks).
• Explored what capabilities might make AI more effective (web search, APIs, etc.).
• Discussed risks and mitigations (cross-checking, privacy, over-reliance).

Value-Add Exercise

• Encouraged to expand outputs creatively using the ValueAdd GPT — from time-saving to impact-creating.
• Groups uploaded photos of their layouts for AI feedback.

---

3️⃣ Early Afternoon – Depth and Disciplines

DIRECTIVES & STRATEGIES

• Refined how prompts are phrased: personas, tone, frameworks, few-shot examples, etc.
• Tried mixing techniques: e.g. meta-prompting, adversarial personas, step-by-step reasoning.

REVIEW Prompts

• Learned reflective AI collaboration:
  • Critique my prompts
  • Identify blind spots
  • Analyse our collaboration
  • Suggest a new adventure
• Emphasised feedback as two-way learning between user and AI.

---

4️⃣ Late Afternoon – Inside the Black Box

How LLMs Work

• Simulation using dice and sticky notes: each participant predicts next word → randomness selects one.
• Key concept: AI predicts next most probable token.
• “The intelligence isn’t artificial — it’s accumulated human pattern-recognition.”
• Discussed context windows and token limits (8k–1M tokens).
• Reinforced the power of good prompts and relevant context.

---

5️⃣ Final Session – Applying It All

Agile Prototyping & Design Sprints

• Core principles: Prototype everything; 80% right is right enough; fail fast, learn faster.
• Used the AI Design Sprint (condensed 2-hour version):
  • Understand → identify problem/opportunity (sails & anchors)
  • Ideate → brainstorm with AI support
  • Select → prioritise with impact/ease mapping
  • Prototype → test through AI tools or Custom GPT
  • Evaluate → gather feedback and refine.

Custom GPTs / Gems

• Compared manual prompting vs CustomGPT automation.
• Anatomy of a CustomGPT:
  • Instructions (persona, steps, directives).
  • Knowledge files, capabilities, first prompt.
  • User customisation & retrieval-augmented grounding.
• Built or conceptualised mini-CustomGPT prototypes.

Closing

• Shared project prototypes and reflections.
• Encouraged post-session peer groups for ongoing AI practice.
• Reminder: This is just the beginning of the adventure.

---

🪶 Reflective Summary

Today’s workshop reframed AI not as a tool to replace human work, but as a companion in our evolution — demanding new attitudes, curiosity, and courage.
Key takeaways:

1. Paradigm shift: The hardest barrier to AI adoption is mental, not technical.
2. Adventure mindset: Success requires curiosity, resilience, and purpose — treating learning as exploration.
3. Balanced value: Use AI to free time and to expand value-creation, not just efficiency.
4. Prompt anatomy mastery: Understanding structure (inputs, outputs, directives, strategies) turns AI from a responder into a collaborator.
5. Partnership model: Co-learning with AI builds both capability and confidence.
6. Experimentation: Everything is a prototype; progress comes through testing, reflecting, iterating.
7. Empowerment: The tools (ChatGPT, Gemini, Toolchest resources) are ready; what matters is agency, imagination, and disciplined play.

The day ended not with “learning AI,” but with reclaiming authorship of how we and AI evolve together.

MC Note – ChatGPT’s resident image creator (Dall-E) often has problems with words.  But that seems to be an issue across all of AI at this stage of development, as you will see at the end of Gemini’s notes below.

Gemini’s Notes

Here are my complete notes, personal summary, and graphic recording from the “Adventures in AI” workshop.

---

 

Part 1: Workshop Notes (Live Recording)

Module 1: Welcome & The AI Context (09:00 – 10:00)

1.1. Introduction & Objectives

• Event: “Adventures in AI” , a 1-day workshop focused on personal AI empowerment.
• Host: Culturistics Ltd / Mike Clargo.
• Resources: All links and resources are available at Al.toolchest.org.
• Main Objectives (Slide 4):
  • Develop appreciation for what AI can do.
  • Understand the role we play in getting it to do that.
  • Learn to ask the right questions.
  • Gain confidence in our ability to learn with AI.
  • Build real enthusiasm to take this forward.
• Intended Deliverables:
  • A bespoke AI project we can implement in our work.
  • A support group to learn AI together.
  • Development practices to protect ourselves and the business.
  • A resource pack to build on.
• Qualities We Need (Slide 3): Self-assuredness, creativity, curiosity, discipline, purpose, and resilience.

1.2. Introductions (Activity)

• Form groups of three.
• Share: 1) Name, 2) Feeling about AI, 3) Experience to date, 4) Ideal workshop outcome.
• Required materials: Smartphone, laptop, pen, paper, post-its, and a deck of cards per group.

1.3. The Presenter & The Premise

• Mike Clargo: Originally a Mechanical Engineer; a consultant on organizational change since 1988.
• His view on AI: He is an “enthusiastic adventurer”.
• CORE POINT: Issues around AI adoption and effective use are NOT technical.
• They are attitudinal. They are about paradigms, change, and perspective.

1.4. “Instant Trainer” (Activity)

• We scanned a QR code for a Custom GPT called “Instant Trainer”.
• We typed in our answer to “What would you ideally like from this workshop?”
• The GPT gave personalized training advice.
• The Point: With AI, we can solve most technical issues ourselves , providing we know what to ask. The real work is attitudinal (appreciation, confidence, etc.) .

1.5. Why AI Matters: The Impact on Work

• A chart (Slide 2) shows AI models (like GPT-5-high) are projected to reach “Parity with industry expert” by ~09/25.
• Prediction: In 3(?) years, ~50% of current professional, administrative, and leadership tasks will be dealt with by AI.
• This leads to two possible futures:
  1. Current task-loads are handled by half as many people.
  2. Task-loads (and people) evolve to find new channels for value-add.
• We must have agency in that evolution and become proficient in using AI.
• Slide 9 shows roles most vulnerable “in their current form” , like Interpreters, Customer Service Reps, Writers, and Editors.
• Slide 10 shows how AI can augment roles. E.g., a “Translator” can evolve into:
  • Cultural Mediation & Transcreation.
  • Language Consulting & Strategy.
  • Creative Collaboration.
  • Quality Assurance & Ethical Oversight.

1.6. Burdens vs. Opportunities (Activity)

• Key Concept: AI is a tool for shifting our work.
• AI Takes (Burdens): The routine, repetitive, mechanical, tedious tasks.
• AI Enables (Opportunities): Tasks that were previously too difficult. Tasks that require our humanity: relationships, empathy, creativity, insight.
• Group Discussion: We listed our personal “Burdens” (time-consuming, boring) and “Opportunities” (work we enjoy, wish we had time for).

Module 2: The “AI Jungle” & The Adventurer (10:00 – 11:00)

2.1. What is AI?

• Taxonomy (Slide 13): Computer Science > Artificial Intelligence > Machine Learning > Neural Networks > Large Language Models (LLMs).
• LLM = Generative Pre-trained Transformers.
• For most of us, AI = Generative Pre-trained Transformers.
• Examples: Gemini, ChatGPT, Claude, Copilot, Llama.

2.2. Levels of AI Application

• Level 1 (Ad-hoc): Basic search, drafting, refining content.
• Level 2 (Exploratory): “Co-creation”. Using AI to teach ourselves AI , transcribe meetings , translate, generate images/music/video, analyze large documents.
• Level 3 (Strategic): A massive, growing list of augmented tasks. Slide 16 shows applications in Work, HR, Finance, Health, Learning, etc..

2.3. Why “Adventure”?

• The “AI Jungle” is expanding faster than anyone can keep up.
• AI computing power has doubled every 3-4 months since 2012.
• Nearly 50% of workforce skills are expected to be obsolete within 2 years.
• It’s impossible to know the whole jungle.
• The New Paradigm: We must stop trying to be “experts” who know everything. We must become “adventurers”.
• Our goal is not to know the whole jungle, but to get value (and joy) from exploring its pathways.
• AI is both the Adventure and the kitbag.

2.4. The Two Key Principles

• Principle 1: Balanced Navigation (Slide 22).
  • We must balance two “directions” of work.
  • Saving Time (South): Focusing only on this leads to being “Stuck” or “Discarded”.
  • Adding Value (North): Focusing only on this leads to being “Exhausted”.
  • The Goal: “Balanced Co-creation” (North-East) , which leads to becoming “Evolved”.
  • Key Insight: It’s not just what you do, it is also who you become.
• Principle 2: (Revealed later) Everything is a Prototype.

2.5. Co-creating Ourselves

• AI is the greatest resource ever for helping us live up to our potential. It offers skills, self-awareness, and coaching support.
• Quote: “It is never too late to become the person you might have been” … “especially now!”.
• We get trained “beyond the basics” by having a conversation with AI.
• Activity: We tried prompts about our prompting:
  • “What assumptions have you made in answering my last prompt and how might I have better written it?”
  • “Can you construct me a course that can help me better understand how to use AI in my work?”
  • “What is good practice for me to become more AI literate in my work?”

Module 3: Anatomy of a Prompt (The “GenAI Card Deck”) (11:00 – 13:30)

3.1. The 7 Components of a Prompt

• We opened our “GenAI Card Deck” and sorted them by color.
  The cards represent the “anatomy” of any AI task.

• • OUTPUT (Teal): What do you want to create?
  • INPUT (Blue): What are you providing?
  • TASK (Pink): What is the transformation from Input to Output?
  • CAPABILITY (Purple): What extra tools does the AI need?
  • RISK (Red): What could go wrong?
  • DIRECTIVE (Light Green): How should the AI behave?
  • STRATEGY (Yellow): How can we push the AI’s thinking?

3.2. Group Activity: Building a Use Case (The “Layout”)

OUTPUT (Slide 29):

• • We discussed the outputs from our own roles.
  • We chose 1-2 we’d like AI to help with (e.g., “TEXT” for drafting documents).
  • We stuck post-its with specifics onto the card.

INPUT (Slide 31):

• • We identified the inputs needed for that output (e.g., “TEXT” for my rough notes, “FILE” for a PDF to summarize).
  • Added post-its to the INPUT card(s).

TASK (Slide 33):

• • We found the “verb” card(s) that bridge the Input and Output (e.g., “ANALYZE” my notes, “MORPH” them into a report).
  • Added post-its.

CAPABILITY (Slide 35):

• • We identified extra tools needed (e.g., “WEB SEARCH” for current data, “DATA RETRIEVAL” to read the PDF).
  • Added post-its.

VALUE-ADD (Slide 37):

• • We reviewed our “Balanced Navigation” compass.
  • We asked: Is this just saving time? How can we add value?
  • We brainstormed “Novel” outputs (e.g., not just a report, but a “Video script,” “Podcast,” or “Infographic”).

RISK (Slide 40):

• • We discussed what could go wrong with our idea.
  • We selected RISK card(s) (e.g., “HALLUCINATION,” “PRIVACY BREACH,” “BIAS”).
  • We discussed mitigations (e.g., for Hallucination, “ask for references” or “check with another AI”).
  • Key takeaway: Risks are not reasons to avoid AI; they are parameters to manage.

DIRECTIVES & STRATEGIES (Slide 45):

• • We “flavoured” our prompt.
  • Directives: “Act as a [PERSONA]” , “Use a [TONE]” , “Provide 3 [FEW-SHOT EXAMPLES]”.
  • Strategies: “Use the [FRAMEWORK] (e.g., SWOT analysis)” , “This is a [META-PROMPT]… write me a better prompt”.
  • We drafted a full prompt based on all our cards.

3.3. REVIEW (Activity)

• We looked at the final card color: REVIEW (Dark Green).
• These are prompts to improve our side of the partnership.
• Examples: “CRITIQUE MY PROMPTS” , “IDENTIFY MY BLIND SPOTS”.
• This is a “two-way street” —it’s about “who you become”.

 

Module 4: How AI Works & The Design Sprint (13:40 – 17:00)

4.1. How LLMs Actually Work

• “Black Box LLM” (Activity):
  • We simulated an LLM in groups of 6.
  • Given a phrase (“The cat…”) , we all wrote a likely next word.
  • The presenter rolled a die to randomly select which word was added to the chain.
  • We repeated this, building a “probabilistic” sentence.
• The Reality:
  • An LLM is a giant probability machine. It just predicts the “next most likely word” (token) based on all the text it was trained on.
  • The “intelligence” is NOT artificial. It’s the “vast mass of documented human intelligence” retrieved by algorithms.
  • “Reasoning” is just finding patterns of reasoning from its training data.
  • “Context Window” is the AI’s short-term memory. It ranges from small (ChatGPT 3.5) to massive (Gemini 1.5 Pro).
  • This is why prompt quality matters: it “sets the pattern” for the AI to follow.

4.2. Putting AI to Work: The AI Design Sprint

• This is the “REALLY practical bit”.
• Principle 2: Everything is a Prototype.
  • Apply Agile Thinking. 80% right is right enough.
  • Fail fast, learn faster.
  • The Design Sprint is a 5-day Google Ventures process we condensed.
• THE GOAL IS LEARNING.

4.3. What We Are Building: Custom GPTs / Gems

• Basic Route: A defined process with clear instructions and prompts for a user to copy-paste into a standard AI.
• Advanced Route: A Custom GPT (OpenAI) / Gem (Google).
  • This is where the Designer provides the “Instructions,” “Directives,” and “Knowledge Files” (like PDFs or spreadsheets) to the AI behind the scenes.
  • The User then has a simple chat with this pre-configured, specialized AI.
  • Creating Gems is free; Custom GPTs requires a paid subscription.

4.4. The AI Design Sprint (Activity)

1-Understand (Slide 63):

• • We used a GPT to identify ideas for our roles.
  • We created “How Might We…?” (HMW) questions.
  • We posted one “sail” (value-add) and one “anchor” (time-saving) HMW.

2-Ideate (Slide 66):

• • Our group chose one HMW to work on.
  • We brainstormed solutions on post-its.
  • Cool step: We photographed our post-its and fed them to an AI, asking it to critique our ideas and find blind spots.
  • We added more ideas, then dot-voted for the best ones.

3-Select (Slide 68):

• • We plotted our voted ideas on an Impact / Ease matrix.
  • We selected one “minimum viable solution” that was high-impact and feasible to prototype today.

4-Prototype (Slide 70):

• • We used our GenAI Card “layout” (Input, Task, Output, etc.) to design the prototype.
  • We gathered our test inputs (e.g., a sample document).
  • We wrote the final prompt(s) or built a quick Custom GPT.
  • Reminder: “If your prototype doesn’t work… YOU HAVE NOT FAILED… as long as you have learned.”

5-Evaluate (Slide 72):

• • We tested our prototype to see if it worked.
  • We created a “WWW” (What Went Well) / “AFI” (Areas for Improvement) chart on a flipchart.
  • Gallery Walk: One team member “hosted” while the rest of us reviewed other teams’ prototypes, adding supportive WWW/AFI notes.

4.5. Key Principles & Wrap-up

• Principle 1: Balance a focus on time-saving with a focus on value-add.
• Principle 2: Everything is a prototype.
• Principle 3: AI exists to augment you, not replace you. YOU ARE RESPONSIBLE for the outcomes.
• To ensure this, use the “Sandwich Model”:
  • 1st Human: Think first. Define the task.
  • 2nd AI: Use good prompts. Let it do the “heavy lifting”.
  • 3rd Human: Check and augment the results. You are the responsible adult. Learn from the exchange.
• Final Safety Note: Be mindful of data confidentiality , reliability (human oversight is a must) , and regulatory compliance (GDPR).

---

 

Part 2: My Personal Summary & Reflection

Today’s “Adventures in AI” workshop was a fantastic, high-energy event. The core premise, which I found incredibly valuable, was that succeeding with AI is not a technical challenge, but an attitudinal one.

We started by confronting the reality of AI’s growth. The chart showing AI reaching “parity with industry experts” was sobering. This means our roles will change. We were presented with two paths: be made redundant by AI handling our “burdens,” or evolve with AI to find new “opportunities” and “value-add” channels.

This led to the central metaphor of the workshop: the “AI Jungle”. The field is expanding so fast that it’s impossible to be an “expert” who knows everything. Instead, we must adopt the mindset of an “adventurer” —one who explores pathways, learns, and gets value without needing to know the whole map.

To guide our adventure, we were given two key principles. The first is “Balanced Navigation” : we must balance “Saving Time” (automating tedious tasks) with “Adding Value” (using AI to become more creative and human-centric). The goal isn’t just to do more, but to become more. The second principle is that “Everything is a Prototype” —we must embrace an Agile mindset, “fail fast, learn faster” , and understand that learning is the goal.

The main activity involved using a “GenAI Card Deck” to deconstruct an AI task into seven components: Input, Output, Task, Capability, Risk, Directive, and Strategy. This was a brilliant, hands-on way to move from a vague idea to a structured, effective prompt. We built a use case for our own jobs, considering not just the “how-to” (Input/Task/Output) but also the “what-if” (Risks) and the “how-to-make-it-better” (Directives/Strategies).

A “how LLMs work” simulation using dice demystified the technology. It’s not “thinking”; it’s a massive probability engine predicting the next word. This highlighted why our input is so critical: the “intelligence” is just a reflection of the “vast mass of documented human intelligence” it was trained on, and we are responsible for setting the pattern.

We spent the entire afternoon in a rapid “AI Design Sprint”, moving from a “How Might We…?” question to a functional prototype. We brainstormed, used AI to critique our own ideas , plotted them on an Impact/Ease matrix , and then built a real solution (some of us even made a Custom GPT ). The “gallery walk” evaluation was a great way to learn from other teams and get immediate feedback.

My biggest takeaway is the “Sandwich Model”: Human – AI – Human. The human thinks first, the AI does the heavy lifting, but the human must check, augment, and take responsibility for the final output. AI is a powerful partner, but it’s our job to be the responsible, ethical, and creative leader in that partnership. I’m leaving with a tangible AI project and, more importantly, the confidence and enthusiasm to continue the adventure.

---

Part 3: Graphic Recording

MC Note – In and of itself, Gemini is not particularly equipped for images (it tends to create tables) but it does have an amazing tool (as at October 2025 – probably the best) through its AI.Studio, called Non-Banana, which produced the following based on the notes above (it is a bit better with words, but as you can see there are still errors). You can also see that it has a few hallucinations in there, and didn’t rely entirely on the notes above: