Pieter Abbeel, Zsolt Katona and Matthew Stepka - The Business of AI

  Home | Research | Teaching |  
   

Syllabus - The Business of AI (EWMBA267-11 - Spring 2024)

Note: this document may be updated in the future. Please check back for the latest version.
Last Update: January 23, 2024

Dates, Times, Location

2/4: 9am-5:00pm, N370

2/18: 9am-5:00pm, N370

Instructors - Pieter Abeel, Zsolt Katona and Matthew Stepka

Email: pabbeel@berkeley.edu, zskatona@berkeley.edu and mstepka@berkeley.edu - please use [EWMBA267] in the subject.

Introduction

From self driving cars to chatbots and humanoid robots, and with the recent buzz around generative technologies, Artificial Intelligence (AI) is here and changing the way we live, work and do business. The class is designed to introduce future managers to AI technology and its many business applications. Students will walk away with a foundational understanding of AI and its near and long term applications, explore the myths and realities surrounding the technology, and delve into the legal, social and policy implications of AI. Professor Katona will be joined by Haas Executive-in-Residence Matthew Stepka, to provide valuable insight into the current AI business landscape.

Pieter Abbeel is Professor at the Department of Electrical Engineering and Computer Sciences, Director of the Robot Learning Lab, and Co-Director of the Berkeley AI Research (BAIR) Lab. His research is in machine learning and robotics. He co-founded Gradescope and Covariant. He is also the host of The Robot Brains Podcast and an investment partner at AIX Ventures.

Zsolt Katona is the Cheryl and Christian Valentine Professor of Marketing and has been at Haas since 2008. Zsolt has a Phd in Marketing and Computer Science. His research focuses on the several topics related to business analytics such complex networks, search advertising, network economics, social media and digital marketing.

Matthew Stepka is Executive-in-Residence and Lecturer at Haas. He is Managing Partner of Machina Ventures, an investment firm focused on early stage, artificial intelligence and data science enabled companies. Previously, Matthew was VP, Special Projects at Google, where he led and incubated strategic initiatives, especially mission-driven projects with high social impact.

Participation

The success of this course depends on everybody's effort to prepare before class. Learning is the most effective when you are actively engaged in thinking about and doing rather than passively absorbing the material. Everybody's contribution is important. It is rare that there is one right answer to a particular issue. Do not expect one. Rather, you should expect to learn from seeing how others address the problem that you have thought seriously about. The better prepared you are the more you will learn. Therefore, we will try to call on everyone, not just volunteers to contribute and defend their viewpoints. Lectures and cases will be interactive.

Preparation/Pre-work

Required Readings

For each day, please read the case study(ies) listed under sessions before class. These are required readings. You can access the case studies via study.net.

Required Video Lectures

For the first day of class there will be about one hour of video lectures to complete. These contain material that is somewhat more technical in nature, but still accessible without prior training in machine learning. Depending on your experience and training you may be familiar with the material so feel free to fast forward, but make sure you know the content before you come to class. If the material is new to you please pay careful attention to the videos, they will be fairly dense. The videos are labeled as VL.D1.1-12 , and can be found under Modules in bCourses.

Online Class Participation Questions

There are a number of discussion starter questions/polls intertwined with the videos on bcourses. These all need to be answered/completed. They are labeled as CP.D1.x , whereas the questions for day two are labeled as CP.D2.x

Textbooks

We do not use a textbook in the class. If you want to learn more, we recommend these two:

Assignments

Deep Learning Homework (Individual)

You will learn to code deep learning yourself! Don't worry you don't have to write code, rather you will be given a step by step guide and the homework will involve changing bits of the code to experiment with a classification task. Details will be available on bCourses after the 2/4 session and you will have until Saturday night, 2/17 to complete the assignment.

The assignment will be run on Haas' Jupyterhub platform. The coding language will be Python and we will use the Tensorflow and Keras libraries. Learning/knowing python is not a prerequisite. However, if you have no experience with Python, it is recommended that you complete the Jupyterhub Python tutorial before our first session, if you don't have any prior python experience.

To access the tutorial, go to https://jupyterhub.haas.berkeley.edu/ and click on R & Python Tutorials in the bottom of the page under Research Hubs. After you log in with your calnet account, click on "Start Server" if it does not automiatically start up and then click on the Python-Tutorials folder. Then click on the 1_Start_Here folder and the mp4 video file will walk you through the rest. The tutorial contains problem sets from the Data and Decisions class and goes through the solutions in Python. Completing the tutorial will give you a basic understanding of Python and the Jupyterhub platform, both of which are widely used by data scientist.

Due date: 2/17 11pm: Submit deliverable on bCourses

Final Quiz (Individual)

There will be a short multiple-choice quiz that you need to complete after our second meeting. The quiz will assess your knowledge of the most important points we learn in class. The quiz will be open book, but timed and you can only use materials that are readily available on your laptop/computer. No use of Internet or other outside resources are allowed (This rules out most language models, unless you manage to run it on your computer locally - try at your own risk). You can start any time before the due date, but obviously do not discuss with anyone else.

Due date: 3/10 11pm: Take on bCourses

AI Application Business Plan (Group)

In this project, you will develop a plan for an AI application. Your final report should be a business plan much like you would make a pitch as the founders of a company to a venture capital investor. To facilitate their review, you will be asked to create and submit a 2-3-minute video pitching your idea. The format of the final report should be: PowerPoint deck with 15-20 slides including Appendix. Write out speaker notes as if you were presenting.

Click here for a detailed description of the project

Due date: 2/17 11pm: Submit a document containing

  1. Names of the members of your group (4-6 members per group)
  2. The Title of your project (you can change this later if needed)
  3. A brief, one-paragraph description of the AI application idea that your group wishes to pursue
  4. Even though, you have until our second meeting to submit this document, we encourage you to start serious work on the plan earlier

Due date: 3/10 11pm: Submit the final report.

Submission of Documents

Submit all the assignments on bcourses in .pptx, .pdf format or as a link (to your video or a Google slides presentation). If there is a problem with bCourses, email us a copy before the deadline. Your slides should be self-explanatory as much as possible, but you can add presentation notes if you need to explain something in more detail. Once project groups are assigned bCourses will have a feature for groups to submit assignments. Until then, one submission per group is enough. If multiple members submit from a group, the last submission will be graded.

Generative AI Tool (aka ChatGPT) Policy

Except for the final quiz, generative AI tools are allowed to be used. However, any time you use the text/image/video output of such a tool in a submission, please add a footnote identifying the tool and a brief description of the prompt used.

Grading

Your grade will be based on the number of points you get from the following assignments. The maximum total score is 100 points.
  • Deep Learning Homework: 25 points
  • Final Quiz: 15 points
  • Group Project: 60 points
Note that Haas policies require the average GPA in elective classes to not exceed 3.5.

Session Details

2/4 - AI and what's behind it. AI Applications

Prework

  • Read Case Study: "Zebra Medical Vision"
  • Watch Video Lectures: VL.D1.1-12
  • Complete Online Class Participation: CP.D1.1-5
  • Research Companion.ai and Polyrgraph . Answer the corresponding questions outlined in CP.D1.4-5

Topics:

  • What is AI? How does AI work?
  • Supervised Learning and Neural Networks, Deep Learning
  • Where is it useful? What are the limits, limiting factors?
  • Computer Vision Basics and Applications
  • Generative Language Models and the ChatGPT hype
  • From idea to application
  • Future Applications of AI and the Startup Space

VC Panel

  • James Hardiman, DCVC
  • Shaun Johnson, AIX Ventures
  • Vivek Gopalan, 8VC

2/18 - AI Strategy, Algorithmic Bias and Fairness

Assignments Due

  • Group Members and Project Description Due date: 2/17 11pm
  • Deep Learning Homework (individual) Due date: 2/17 11pm

Prework (due before class)

  • Read Case Study: "Osaro: Picking the Best Path"
  • Read Case Study: "SLB: Disrupting the Traditional Energy Industry Through AI Drilling Innovations"
  • Read Case Study: "LOOP"
  • Complete Online Class Participation: CP.D2.1-3

Topics:

  • Deep Learning Homework Debrief
  • Robotics and Reinforcement Learning
  • AI and Automation
  • Artifical and Human Intelligence Working Together
  • AI and Business Strategy, Competitive Advantage
  • Implementation of AI use-cases. What can go wrong?
  • Algorithmic Bias and Fairness
  • The Future of AI and Humans