Lunch & learn Virtual TRAINING

1. A five day, 1 hour Lunch and Learn training series that gets marketers hands-on with Generative AI systems while explaining how the systems work and their strengths and weaknesses. The training is based on the MIT book, The AI Conundrum, authored by Caleb Briggs & Rex Briggs.
2. The training has been piloted with Kroger to rave reviews. An outline of the topics and exercise is available below. A sample of the 15min introduction content is available here. The Preface, Introduction and first two chapters of the book are available here.

MIT Press commissioned the book, The AI Conundrum, for publication in 2023 (for distribution in 2024) citing strong academic peer review, including: 

“The work is quite original -- it's the best thing I've read on AI, and I read a lot about AI. The scholarship is quite sound.  I appreciate very much that they use real world examples in addition to examples from their own experience.”

“I think this book will really help business leaders (who are in a state of panic about AI at the moment) understand better what AI can and can't do, how to assess the feasibility of a proposal, and how to mitigate some risks of AI.  The book does a great job of explaining AI to a reader whose stat knowledge is at the level of a solid understanding of regression. It demystifies several opaque AI concepts in a very effective way. The three dimensional model they propose works beautifully. I love that "whether or not there is data" is NOT part of that model. That's so smart.”

“IMHO, Tom Davenport's books on AI have been the best books of this type (for business leaders or instructors) and this one is far better than those. The authors have more hands-on experience with AI and probably more experience in explaining AI to others. This book is far more approachable and understandable than some more technical books I've read that are ostensibly written for the business audience.”

MIT Press has designated The AI Conundrum with the “SPL” designation based on their projection that this will be one of the top 15% of books they publish in 2024. MIT also signed Caleb Briggs with first right of negotiation for any future books he produces.

Caleb Briggs began coding when he was 10. By 14, he was instructing dozens of adult teachers to code in MIT’s Scratch program, so they could carry what they learned from him into their classrooms in the wider community. In high school, his math skills were beyond the curricula, so he attended Harvey Mudd College and later, Stanford University to advance his skills. At the same time, Caleb attended Sage Ridge, a small private high school in Reno-Tahoe area of Nevada, where he wrote a variety of AI programs using genetic algorithms, computer vision and natural language models. It was through this work, going deep into the mathematics of AI, that he recognized the weaknesses of AI. His senior thesis was entitled the Fundamental Limitations of AI, which MIT commissioned for publication. Caleb now attends Reed College, where he studies pure math and computer science.

Rex Briggs is an award-winning marketing researcher and data scientist. Rex has been working with machine learning and neural networks for over 30 years. Rex was the first director of research for WIRED, and a pioneer in digital measurement and AI with five patents. The Market Research Council named Rex as the 2022 Change Maker for his work with the Ad Council on applying AI. He is co-author of What Sticks (2006), a book that has been required reading at Wharton, Harvard, and other leading universities. He is also author of SIRFs Up, How Software and Algorithms Are Changing Marketing (2012). Rex is a board member at Cal Poly, SLO, College of Business Analytics’ program, and a business founder that made it to the top of Inc. 500 fastest growing companies multiple times. He successfully exited his AI business, Marketing Evolution, in 2019. He now enjoys mentoring growth stage companies on how to scale their business and training corporations to build an AI-first with responsibility.

“Really educational. Walked away feeling more confident about my knowledge on how AI works and the watchouts. It made a complex, overwhelming topic much more accessible.”

“Interesting subject, information it was great!!!!”

“I didn't know what to expect going into the training except that I figured most of the information would go over my head. The opposite was true! Rex and Caleb really broke it down into understandable real-life examples and how we might use AI in our day-to-day marketing work.” 

“I appreciated the simple language they used to describe the concepts!”

“I learned a lot and it gave me ideas for how I might be able to apply it to my work.”

“Didn't know much about AI before the training - liked that both the benefits and challenges were shared.”

Norm de Greve, Global Chief Marketing Officer at General Motors, praised the book for its compelling insights into the nature of AI. He emphasized the importance of understanding AI’s hidden weaknesses for business people, government decision makers, students, and citizens.

Beth Egan, Associate Professor and Graduate Program Director, Advertising, Syracuse University, has made the book required reading in her courses. She reported that the book is informative, hugely helpful, and easy to read. She appreciated its digestible content for students and business leaders alike.

Students from Syracuse University’s inaugural course have given the book five stars, praising it for being entertaining, witty, and extremely informative. They appreciated how it made the complex subject of AI digestible for anyone, regardless of their level of familiarity with AI, machine learning, or any related subject.

The book has been incorporated into corporate training at Kroger, used as the basis for the Inc5000 AI Bootcamp to the founders of the fastest growing private businesses in America, and added to the reading list for business analytics courses due to its comprehensive evaluation of AI’s strengths and weaknesses for the non-computer science audience.

Registrants receive a hard copy of book The AI Coundrum (MIT Press) when it ships in 2024 and an electronic advance copy (downloadable upon registration in 2023, prior to training). 

Registrants also receive self-assessment AI scoring before and after training, and 1 year access to course materials, and recorded sessions.

Introduction - How well do you know AI? (15min)

https://youtu.be/aCmvuIu9nmU 

Exercise 1: Self-assessment (10min)

• Write down your score & enter it in the poll.
• Brief discussion of training to level up your AI experience

Exercise 2: Intro to LLMs (15 min)

Weaknesses & Strengths: 

• Experience LLMs being illogical, hallucinating, etc. 
• See LLM nailing summary with impressive nuance

Debrief: Discuss strengths and weaknesses, preview future Lunch & Learn (sign-post rest of the week) and what attendees will know by end of session.

AI Use Case Showcase (10min)

Universal Approximation (5min)

• Pattern fitting with AI overall
• How “x, y pairs” from universal approximation produce image generation (words & images). 

Exercise 3: DALL-E image generation (10 min)

• Dall-E inpainting / Outpainting - how context influences output
• Debrief (5 min) Logic limitations

Gradient Descent (15min)

• Explain how AI “learns” and why it is necessary, but can make mistakes humans don’t typically make, and amplify bias.

Exercise 4: Midjourney image generation (5min)

Debrief (5 min)

• Discuss Bias and why AI can amplify it.  

AI Use Case Showcase (10min)

How Large Language Models Work (10min)

Exercise 5: Steering LLMs (10 minutes)

• Personas
• Bias
• Fix Mary Jane
• Fix math problem)

Context, Sequence (10min)

Custom Instructions & Grounding (10min)

• Building & saving personas 
• Jailbreaking/Bing Chat break

Exercise 6: https://www.bargainer.ai/ (10min)

AI Use Case Showcase (10min)

Vector Databases Overview (10min)

• How they work – why they can give source of where information came from but LLMs can’t 

Exercise 7: Vector Databases The AI Conundrum Book Bot (5min)

Introduction to Multimodal AI & Analysis Capabilities (5min)

Exercise 8: Analyzing Images (5min)

• Interpret chart  (Bard Only, OpenAI & Antropic can’t “see” images yet)

Exercise 9: Analyzing PDF (Bard, GPT-4, Claude) (5min)

• Summarize key points from 14 page TGA.pdf
• Interpret Report: Kroger Annual Report
  (Requires GPT-4 paid subscription)

Exercise 10: Analyzing Data (Bard, Code Interpreter & Anthropic) (10min)

• Data analysis (requires GPT-4 paid subscription)
  Google form poll to generate data, analyze with Code Interpreter

Debrief and Discussion of Tools / Plugins (5min)

Exercise 11: Limits of Image Understanding (10 min)

• Image understanding
•  Bard Kroger Store Example

AI Use Case Showcase (10min)

Introduction to Agents, APIs and Workflows (15min)

• AutoGPT (command line) and AgentGPT (web interface)
• API and HTML (www.speakerrex.com/AI-radio.html)
• Hugging Face
• Langchain

(Advanced course goes deep into this area)

Responsible AI: (10min)

• Privacy of Data, Challenges of Security, Fundamental Questions (What if AI is wrong, what is the consequence to you or others?)

“The Final” Workshop Discussion: Scoring AI Risk (10 min)

Training Feedback (4 question survey) (5min)

Book Reading & Self-Assessment 2.0 (10min)

AI Use Case Showcase (10min)

After completing The Fundamentals of AI course,  "AI Workflows & Risk Analysis" course is available.