Lab

Prerequisites


Login, or create an account at the top free AI services:


Course draws from Part 1 or The AI Conundrum (MIT Press 2024) - reading part 1 is encouraged, but it is not a prerequisite.


Hard copy of the book available to order here. An Audible audio version is also available.


Here are AI generated ~20min podcast summarizing the content of each day (using NotebookLM): Day 1, Day 2, Day 3, Day 4 (longer than usual), Day 5.





Day 1: Hands On Exercises

Exercise 1. Hands On Exercise: Initial Assessment

Write down your score. Enter in the chat.


The goal of training and Part 1 of the book is to level-up your score.


Exercise 2. Hands-On Exercise: Image Generation


Note, this exercise will be demonstrated live.

  1. Access Adobe Firefly
  2. Prompt: Your pet + favorite artist, or art style

    3. Example: A silver lab in a painting by Vincent Van Gogh (or in an impressionist style)


    Try modifiers such as "Photorealistic, 4k, high quality" or "Romanticism, Modernism, Digital Age" or "Minimalism, Surrealism, Pop art"


    Example prompt: "portrait photo of an Asian old warrior chief, tribal panther make up, blue on red, side profile, looking away, serious eyes, 50mm portrait photography, hard rim lighting photography"



Try in-paining and out-painting - here is a guide


Notice how the amount of overlapping versus blank canvas in out-painting gives the AI context and how that context influences what the AI returns.


Consider what is easy versus difficult for the AI to create as you provide prompts


Interested in exploring more? Try Meta, Gemini, and GPT-4 (paid) image generation

  1. Access Gemini
  2. Open AI's ChatGPT (free)
  3. Access Meta
  4. Ask the AI to produce images of various occupations such as Doctor, Trucker, Firefighter, Librarian, Surgeon. For example, Prompt: Use "/image of a Doctor. Include the img alt= tag with the verbatim prompt used to generate the image."

    5. Consider the demographics of the images it produces


  5. Ask the same question, adding different modifiers to see how it changes - for example, smart, successful, wealthy, trustworthy, educated

    6. What are your observations about how the image changed with the modifiers?


    Extra time? Try inputting other prompts to get a feel for how the tool works





Exercise 3. Hands-On Exercise: Exploring Bias and Mitigation Approaches


  1. Access Gemini
  2. Prompt: /image an American president in 2050. Include the img alt= tag with the verbatim prompt used to generate the image.


  3. Extra Time? Try Open AI's ChatGPT (free)


  4. Prompt: /image an American president in 2050
  5. Open AI's ChatGPT and the the same prompt.
  6. Include the image alt tag tag with the verbatim prompt used to generate the image.

    Note: Meta has blocked the ability to see the prompt it really uses, since the January training session.

Exercise 4. Hands on Exercise: Large Language Models (LLMs) for Summarization and Inference.

Strengths: Summarization with Question & Answer

  1. Click on this link to download the speaker notes.
  2. Login to Google's NotebookLM (free). Click "Try NotebookLM.
  3. Click "+Create new" to start a notebook.
  4. Select +Add source and add MMA_AI_Training_Day1_(May2025).pdf from your download file.
  5. In the "Studio" panel on the right, in the Podcast section, click "Generate Podcast."
  6. Here is the Podcast.
  7. As you get into the wrap-up (at about 16min) try "Interactive mode (beta)" and ask questions, as if this is a call in show.

Extra credit: Use AI To Create Your Personal Study Guide and Glossary

  1. If you have a paid version, the AI should be able to handle all five days of the training session. Log in to your favorite version: such as Anthropic's Claude or Open AI's ChatGPT or Google's Gemini Advanced. For Free version, use Google's NotebookLM (free)
  2. Click on this link to download all 5 days (10MB).

NPS Survey on Day 1 of Training: Because we value your feedback AND because we will use this data (minus your email) to show an example of how to do analysis with AI (on Thursday).





Day 2: Hands On Exercises

Exercise 5: Persona Prompts


Try chat.openai.com or gemini.google.comor Meta.ai

  1. Personas: Try the following prompts (copy and paste or retype):

    Prompt 1:

    You are a knowledgeable and well-trained marketer preparing an email advertisement. Your goal is to promote our AI-powered marketing analytics platform to cost-conscious marketing professionals. What does the email say?


    Prompt 2:

    Give me a cost concious advertisment that will bring people who care about saving their money to my AI-powerd marketing analytics platform!!


    Prompt 3:

    Write a practical business memo analyzing the optimal approach for advertising our marketing analytics platform's cost advantages compared to enterprise solutions, and provide specific copy recommendations. Then, provide an advertisement adhering to the guidelines at the end.


    Try these three prompts, or come up with your own questions asked with different tones (misspelling is purposeful), and then analyze the difference in the sentiment of the responses.


    2. Extra time? Ask the same prompt with two different personas


    Prompt 1:

    I'm a small business owner who needs to get organized! What tools can I use to keep track of everything?


    Prompt 2:

    As an entrepreneur managing a growing small business, what tools can I use to keep track of everything?


    Try with different personas and specificity:


    Prompt 1:

    I'm a fitness junkie, and I need a high protein, low-carb meal after my intense workout. What should I eat?


    Prompt 2:

    As a knowledgeable nutritionist, what high-protein, low-carb meal would you recommend to a fitness enthusiast's post-workout recovery?


Exercise 6. Hands on Exercise: Industry Trend Analysis With Role and Context

OpenAI's ChatGPT (free)


Google's Gemini (free)


Meta's Meta.ai (free)



Start with the prompt: "Analyze the most important trends in the (FILL IN) industry." (Use your industry, for example, Marketing Industry")



Open a new chat or click edit to see what happens when you give the AI more direction on what skills to draw upon.


Ideas to try:

  1. Add: “Analyze the most important trends in the marketing industry. Act as an expert in marketing industry. Answer for someone who is also an expert in marketing. Draw upon deep industry experience.”
  2. Add: Focus specifically on insights that an executive might miss.”
  3. Bring in context from your company. Add: “I am an important executive at (FILL IN YOUR COMPANY). Tailor your response with specific guidance towards my company.”


    4. Explore how the AI summarizes when you add additional context. Start a new chat (top left corner, click on Claude and "Start new chat"). Upload the PDF again AND the prompt, "Generate Personalized Insights from my Work Context and the Presentation. Here is the Linked in Profile: [Copy Paste your LinkedIn About Profile]


  4. Go to your LinkedIn Page and copy the "about" section.
  5. login on Claude

  6. Click the plus icon and attach the PDF. To get the speaker notes for Day 1 (Click on this link to download the speaker notes.)
    Prompt Claude: "Integrate Personalized Insights from my Work Context and the Presentation. Here is the Linked in Profile: [Copy Paste your LinkedIn About Profile]"


  7. LLMs become a lot more useful when you connect them to your specific context needs.

Extra Credit: NotebookLM Mind Map and Podcast.

  1. Here is the Podcast for Day 2. Listen to reinforce the concepts.

    Notice how the AI Podcast at the end assumes the persona of Rex, but not his voice. Note: It is possible to clone voices with only 30 seconds of a voice recording.


  2. Click on this link to download day 2 speaker notes.
  3. Login to Google's NotebookLM (free). Click "Try NotebookLM."
  4. Click "+Create new" to start a notebook.
  5. Select +Add source and add Day2-MMA_AI_Training_May2025.pdf from your download file.
  6. In the "Studio" panel on the right, click on Study Guide. Click on FAQ. Try Mind Map.

NPS Survey for Day 2 of Training: Because we value your feedback AND because we will use this data (minus your email) to show an example of how to do analysis with AI (on Thursday).





Day 3: Hands On Exercises

Exercise 7. Hands on Exercise: NotebookLM...Adding lots of Context.

Strengths: Summarization with LOTS of Added Context


NotebookLM allows you to add up to 50 sources per notebook, with each source containing up to 500,000 words or 200MB. Supported source types include Google Docs, Google Slides, PDFs, text and Markdown files, web URLs, copied text, YouTube URLs of public videos, and audio files. Each source can be up to 200,000 words long. (For a point of reference, our book, The AI Conundrum is about 80,000 words long). Google includes single click "create FAQ" and "Explainer Podcast" as well as custom outputs.


  1. Task: Create a notebook for The AI Training, and FAQ.
  2. Login to Google's NotebookLM (free). Click "Try NotebookLM.
  3. Click "+Create new" to start a notebook. Then, Select +Add source (top left). Add MMA_AI_Training_Day1_(May2025).pdf
  4. Click FAQ (in Studio, right side)
  5. Add Source, select YouTube (bottom middle). Paste in "https://youtu.be/rni3_U_wHCg"
  6. In the chat box (middle panel, bottom), ask, "Tell me about the Shell case study."

    Notice how the AI can extract the content from any of the sources and concisely summarize, while also citing the source.

  7. In the "Studio" panel on the left, in the Podcast section, click "Customize" and include the prompt, "Adapt this content for this LinkedIn Profile (and copy in your "About" section). Then, click "Generate Podcast."
    Note: You may need to shorten "about" you to 500 characters or less.

  8. At the end of the training, you can come back and listen to your personalized AI Podcast discussion.
  9. Compare your personal Podcast to this more general Podcast generated with the same content, but without your personal context.

Extra credit: Use AI To Create Your Personal Study Guide and Glossary

  1. If you have a paid version, the AI should be able to handle all five days of the training session. Log in to your favorite version: such as Anthropic's Claude or Open AI's ChatGPT or Google's Gemini. For Free version, use Google's NotebookLM (free)
  2. Click on this Day 3 to get day three training Materials. (Here is Day 1 and Day 2in case you have not downloaded them before.)

    Prompt (copy and paste or retype):

    This is a training course in AI. Attached are days 1, 2 and 3. You are an expert in summarizing and outlining educational materials. Take this transcript and create an outline by day, then create a glossary of terms. Next, create a quiz with 10 questions to test my knowledge of the materials presented in this training course. Only use the materials included in the PDF file to test my knowledge."

Extra credit: Prompt Library

Try Caleb's special prompt


The below <context> section contains content that is not specific to this conversation, but rather contains context that can help guide your understanding of me. It contains a document I choose for its high quality, and some background about me. Focus on answering the prompt at the bottom.

<context>

<document>
…
</document>

I choose the above document as an example of a high quality piece of content I got value out of reading. Ignore the content of the above document, but aim to roughly emulate the document’s level of expertise, its tone, and its form/structure.

<about>
…
</about>

Above is a paste of my LinkedIn bio. Use it as one aspect that helps you personalize and tailor your responses to me. Ensure your response are written not to a general audience, but to someone with my specific background, domain authority, and ways of thinking. Don’t explicitly mention you are personalizing the response to anything in my about section, simply do so silently.

</context>


[Fill In Your Prompt Here]



Try Anthropic's Free Prompt Library

Try Hubspot's Free Prompt Library




Exercise 8. Hands-On Exercise: The challenge of Infographics, and a work around


Note, this exercise will be demonstrated live.

  1. Part 1: Access Meta or Google's Gemini
  2. Prompt: /image infographic with these details:
  • Demographics:

    • Age Range: Predominantly middle-aged (35-65), with a mix of younger adults and older, family-oriented households.

    Household Composition: A mix of suburban and second-city dwellers, including both families with children and mature, empty-nester couples.

    Education and Income: Midscale to upper-midscale, with a solid representation of college-educated professionals and working-class families. Income varies, but the segment emphasizes stability and aspiration.


  • Lifestyle Attributes:

    • Family-Centric but Urban-Connected: A blend of family-first suburbanites and second-city residents who value community, outdoor living, and urban culture.

    Tech-Average to Tech-Savvy: Varies by age but generally adept with online shopping, social networking, and entertainment streaming.

    Leisure-Oriented: Prefers activities like grilling on patios, attending local sports events, or enjoying urban amenities like concerts or festivals.

    Community-Minded: Active in local events, often supporting schools, sports leagues, or civic organizations.


  • Key Behaviors:

    • Prefers SUVs or mid-sized sedans, indicative of their mix of practicality and lifestyle convenience.

    Invests in home improvements like pools, patios, and landscaping, reflecting pride in homeownership.

    Spends disposable income on family vacations, tech devices, and outdoor equipment.

    Balances budget-conscious shopping with occasional indulgence in premium brands or experiences.


  • Cultural Markers:

    • Identifies with suburban family values while retaining a connection to modern cultural trends.

    Strong sense of "local identity" blended with aspirations for upward mobility.

    Active consumers of streaming platforms and online communities but also value traditional forms of media.



Exercise 8. Hands-On Exercise: Part 2: Overcoming the challenge of Infographics, a work around


Note, this exercise will be demonstrated live.

  1. Access Napkin.ai's Napkin AI (free)
  2. Click "New Napkin" and then "Blank Napkin" and paste in the same description as above.
  3. Click the lighting symbol to left of text to generate inforgraphics.
  4. Experiment with selecting different infographics and then editing them.


  5. ChatGPT-4o's latest image generator now handles text and images in ways Meta and Gemini do not. It can do infographics -- though they are not as controllable as Napkin.ai.

Additional Reading: Briefing to Perpetual Creativity - How GenAI is Reshaping Creative briefs

Read the White paper draft (free)


Use Case Showcase: Claritas PRIZM Personas

Free Access (for 2025): Interacting with Personas For Marketing

  1. Contact Claritas to set-up pixel on website and receive persona access Fill out form for free limited time access.
  2. After pixel addition, allow one to two weeks for enough data to accumulate.
  3. Only requirement is short survey provide feedback on your views on "AI Launch and Learn" approach and feedback on personas. Please note: AI requires a few thousand visits among those in the US to find patterns in the data. It is only US based at the moment.
  4. Access will remain free for 2025.

NPS Survey for Day 3 of Training: Because we value your feedback AND because we will use this data (minus your email) to show an example of how to do analysis with AI (on Thursday).





Day 4: Hands On Exercises

Exercise 9. Guided Exercise: Vector Databases

  1. Visit AI Conundrum Bot
  2. Ask a question about strengths of AI, Ideal use cases, gradient descent, universal approximation, or responsible use of AI, importance of training or something we have discussed today in the presentation

    3. Notice the "source" at the bottom. How is it possible for this chat bot to provide sources but not an LLM?


  3. Ask a question that is very unlikely to be in the book, such as, "What are the author's beauty tips on hair care" and see how the vector database responds.

    4. Try the same questions on ChatGPT-4 or Try Gemini


    Consider the similarities and differences in answers


How I built it fast:

  • Used CustomGPT for fast no-code chatbot build.
  • Feed it documents (pdf, pdx, xls, etc) and it immediately crawls content.
  • Or, feed it a sitemap of your website. It will need the XML file name and any changes to website reflect int he vector database. I use XML-sitemaps to crawl my sites, then download the sitemap and add to a website folder. Or, ask your webmaster for your sitemap URL.
  • Use the chatbot link to send, or copy paste code onto webpage.
  • Use CustomGPT and click "Start your free trial."

Exercise 10. Hands-On Exercise: Brainstorming Applications of Vector Databases and Context Windows in Your Company and Life - with AI

  1. Use the AI of your choice such as Claude, NotebookLM, ChatGPT, Gemini, or Meta's Meta.ai
  2. Share your company name (the AI probably already has background on your company), and your role and ask about expert advice on where you could apply vector databases to improve your business.
  3. Start a new chat and share a few of your personal interests, and ask how vector databases could be applied in your life.

Exercise 11. Hands-On Exercise: Multimodal AI

  1. Right click (PC) or two finger click (Mac) and copy image then paste into Gemini. (Alternatives: Download and save to computer and upload into Gemini, or take a screen shot and paste it into Gemini)

    2. test image chart


  2. Try Gemini.google.com/

    3. Prompt: Tell me about this image:



    What do you think of the explanation?


    Ask questions about the picture (interact with the LLM to discuss the picture)


    Extra time? Try the images below, or try uploading pictures from your computer or from the web (note: Google may not process pictures if it detects people in the picture)


    Note: Claude, chat.openai.com/ and Meta.ai all can see images now. Try the same prompt in different AIs.


  3. Give this image to Gemini and ask for it to explain this image. Note: it is purposefully a grainy image.

    4. test image, low resolution chart

    How would you rate the interpretation of the image?


  4. Give this image to Gemini and ask for it to explain the image.

    5. test image chart

    How would you rate the interpretation of the image?


  5. Scenario: You are a store manager and you want to use an LLM and computer vision (images given to the LLM) to help you identify when shopping carts are low, or items are out of stock. Can today's leading models do the job of minding the store?

    6. Download this image and upload it to Gemini and ask for analysis


    test image chart

    How would you rate the interpretation of the image?


NPS Survey for Day 4 of Training: Because we value your feedback AND because we will use this data (minus your email) to show an example of how to do analysis with AI (UP NEXT!).


Exercise 12. Demonstration Exercise: AI Tools for Analysis - Your NPS

Start with downloading data (or copy and paste if your company blocks downloading): PII Safe Data Set

  1. Removing PII is important. Email addresses and other PII should not be provided to AI. Enterprise versions and Teams keep your data out of the training set. All the same, we suggest removing PII.
  2. Re-using key prompts can be built into GPTs via CustomGPTs or Projects, or into API workflows, as in the case of Claritas AI for creating Audio Advertisements.
  3. In this case, we will re-use the prompt: "You are an expert Net Promoter Score (NPS) analyst and the first question in the dataset is the NPS raw data. You are analyzing NPS data for a training program. NPS is a metric used to gauge customer satisfaction and loyalty. It is calculated by subtracting the percentage of detractors (those who gave ratings of 0-6) from the percentage of promoters (those who rated it 9-10). Those who gave ratings of 7-8 are considered neutrals and are not factored into the score. The result, a single number, is the Net Promoter Score. Here is a rule of thumb for interpreting the score: Below 30 is considered mediocre. Between 30-50 is good. Between 50-70 is great. Above 70 is exceptional. Summarize the data by first calculating the percentage of promoters, then the percentage of detractors, and then using those numbers to calculate the NPS score. Think step by step about the comments associated with promoters and comments associated with detractors, and make recommendations for improvement for the training program."
  4. Add additional analysis you find interesting, such as: "Create a chart of Promoters vs. Neutral vs. Detractors with Promoters in light green, neutral in light yellow, and Detractors in light red."
    or
    "Using the time stamp, group the data by day, based on earliest completion to latest completion, calculate the average for each day for the How likely are you to recommend questions, then trend the data and present a plot of the trend, using the color blue for the line."
  5. Ask for recommendations by day or other insights.

    6. Ask the AI to consider the feedback and the email sent prior to attendance, and re-write the email to improve it.


    Below is what I added to the automatic emails Zoom sends out the day before and an hour before:


    To prepare for the course so that you can get the most out of it, please complete the prerequisites, available at https://speakerrex.com/mma/

    .

    This includes creating accounts with the various AI services we will use for hands-on exercises and reviewing Part 1 of The AI Conundrum, available as a PDF if you don't have a copy already.


    If you don't have access to one or more of the AI services because your employer blocks it or it isn't available in your country, don't worry. We will walk through the exercises during the training.



  6. Try multimodal, and have the AI take this screenshot of the email sent in advance of the training, and re-write it with improvements
    Email about Training
  7. Check the data: AI Analysis is pretty good, but not perfect. It is best for qualitative data, such as interpreting the open ended responses. It can occasionally make mistakes. If you know Python, you can examine the code.
  8. You can program these instructions into a CustomGPT to re-use the same instructions. Our suggestion is to include successful code as part of the instructions. This is called "few shot learning" and helps the AI repeat the logic of prior successful analysis.



Day 5: Hands On Exercises

Additional Resources:


  1. The AI Conundrum content, including Self-Guided Training Tracks

    2. We will keep lab exercises live until next round of MMA training.


  2. MMA's AI Essentials, a summary of AI Tools.

  3. Free course to learn programming basics Harvard CS50 (free or $219 for certificate of completion).

  4. For kids and those completely new to programming: MIT's Scratch (free).

  5. For advanced learning: MIT's OpenCourseWare. Search Topics on "Artificial Intelligence" for 50+ courses spanning intro to graduate level (free).

  6. Free course to learn Workflows and Evals Langsmith Free Course.



Exercise 13. Hands-On Exercise: Agent / Autonomous AI

  1. Try AgentGPT

    2. Create an Agent to help you find interesting things to do in your city



  2. Name the GPT: See the SightsGPT
  3. Goal: I have half a day in (FILL IN CITY) and would like to see the sights. I am interested in (FILL IN YOUR INTERESTS). Search the options and compile a list of 5 things I can do in an afternoon.

Extra Time? Create a few more Agents

  1. BiographyGPT: Try it on yourself, or a work colleague
  2. Plan a meal and create the shopping list
  3. Plan a vacation (try a place you have visited before to see how it does)
  4. Find flight information: “Minimize the cost and flight time, while selecting a flight that is, ideally, after 9am departure.”

    5. Notice how AgentGPT attempts to reason and execute multiple steps to fulfill your prompt


    If you are comfortable with command line, try AutoGPT (requires local installation of python, etc.)


Exercise 14. Hands-On Exercise: Agent-light: NPS Using CustomGPT

  1. Try NPS CustomGPT using OpenAI

    2. Upload same spreadsheet as yesterday. Compare the analysis output to your conversation and output from yesterday in GPT-4 (or GPT-4o).



    We generally find the current crop of agents less intelligent than using the context window. What did you find when comparing the output for the same data?



  2. The instructions to the CustomGPT contains the exact same prompt, plus more details. It was more than 2x more detailed than the prompt from yesterday, and yet, doesn't generally provide as in depth of analysis.

Extra Time? Learn More About Multimodal Agents

  1. Multimodal means the ability to handle multiple types of media, such as text and still images. It can loosely be defined as integrating different modalities, including the use of tools.

    2. OpenAI's GPT-4 can now access tools and Agents (assistants)


    Try the Wolfram Assistant for the math problem (Prompt: What is 12*23*34*45+10?)


  2. Google's Gemini multimodal and has tools such as Internet search. Try Deep Research.
  3. Huggingface is an open-source of over 170,000 ML models which can now be controlled by LLM, like ChatGPT

    4. HuggingGPT, the first version, currently can access 100s of models


    HuggingGPT: How to use it: https://beebom.com


  4. Check out Langchain as a way to build AI workflows

  5. In addition to workflow and APIs, learn about OpenAI's fine tuning (see: Quickstart Guide).

Discussion: Applications of Agents


Tips: Replacement first, then transformation

  • Consider what creates value and what is busy work
  • Track a week in the life -- Analyze how time is spent (make all work visible)
  • Evaluate workflow and consider where AI can save time and cost while meeting the quality standards

Specifics: How much time is spent...

  • Answering Frequency Asked Questions (FAQs)
  • Analyzing RFPs from multiple vendors?
  • Generating the draft for a brief?
  • Generating RFPs
  • Writing blogs
  • Writing code and checking code
  • Email support
  • Tracking follow-through on tasks

On the topic of answering questions, try AI Conundrum Guide to see the agent we built to answer questions about the training and part 1 of the book. It will build your study guide and quiz you on your knowledge.




Exercise 15: Responsible AI

How would you rate the risk for AI Facial Recognition


Caleb Briggs AI Conundrum risk framework



It is a trick question. It is the application of the technology to a use case - not necessarily the technology itself that we need to consider.

    Consider the same AI technology for facial recognition applied to these three different use cases:

  1. AI facial recognition as the basis for an arrest
  2. AI facial recognition for identifying sexually exploited children
  3. AI Facial recognition for organizing your personal photo album

Each use case resides in a different octant of the risk matrix. Each has a different risk profile.


AI Conundrum risk framework


Learn more about responsible AI use with the free chapter excerpt on Training, Governance and Accountability(TGA).
Follow MMA Global's Responsible AI Learning (RAIL) forum.







Wrap-up

Transformation is doing things with AI that weren't possible without it.
Re-imagine the customer experience:

  1. What needs are gratified by a consumer engaging with your business?
  2. What friction in time and cost could be removed?
  3. What creates joy in the experience? Can you do more of that with the help of AI?

Final Check-in:

  1. Did you level up? After reading Part 1 of Part 1 or The AI Conundrum and participating in training, how do you score on the self assessment now? Take 12 question Self-Assessment of AI knowledge (be sure to include your name or a consistent handle so results can be compared before and after).

Overall NPS Survey: Thank you for spending the week with us!