4 pillars of generative AI
4 Pillars of Generative AI
January 10, 2024 | By Dean Sevin Yeltekin
In the first installment of this Q&A series, strategy consultant Jeff Sigel ’01S (MBA) describes four pillars of generative AI usage and warns against common pitfalls.
Sevin Yeltekin: You started your own consulting firm, Proprioceptive, several months ago. Can you walk us through your career journey to this point?
Jeff Sigel: I was originally a math and physics teacher, which I like to describe as the hardest marketing job I’ve ever had. After several years of teaching, I joined a consulting firm that introduced me to the world of marketing and inspired me to pursue a business education. Simon gave me the training and credibility I needed to grow in a new direction. I discovered brand management in the first year of my MBA program, landed an internship at Kraft Nabisco, and started a journey through food marketing and innovation that included stops at The Hershey Company and Cracker Barrel.
At Cracker Barrel, I noticed a lack of communication between the finance and data engineering teams. No one was translating consumer data into actionable insights. I raised my hand and volunteered to build an analytics function to bridge the gap. Our CFO would often bring up the topic of integrating generative AI into our operations, in addition to machine learning (ML) tools we were already using for forecasting purposes, but like many organizations, we were not prepared to use those tools in a strategic way. At the end of 2023, I left Cracker Barrel to found a consulting firm called Proprioceptive with a vision for helping companies activate strategy, not just put something to paper. Generative AI is increasingly part of this work.
SY: How have you built expertise in generative AI?
JS: I am an avid listener of audio books. In 2023, I spent much of the year listening to books on topics related to AI and machine learning. My approach is to listen to everything at 2x speed and fly through without being too worried about picking everything up, because what I glean from one book will help me understand the next one better. Using an application called DataCamp, I have also taken weekly classes to learn to code with Python. To supplement this independent learning, I took a class offered by Dan Keating at Simon that brought a fascinating perspective to the table. I learned more about what ChatGPT can do in terms of running Python code and was inspired to explore multimodal generative AI in greater depth.
SY: How can generative AI enhance human intellect rather than replace it?
JS: When developing materials on generative AI for new clients, I walk them through four pillars of application:
- Knowledge task assistance—Tools like ChatGPT add incredible value when it comes to tasks like coding and report writing. I attended a recent conference looking at generative AI in the pharmaceutical industry, and most presentations touched on report writing. These reports are onerous and cost a fortune, but they are an integral part of every drug trial. Generative AI has the potential to streamline this process.
- Enhanced idea generation—Generative AI can help my clients create in different modes: images, words, numbers, and sounds. Coming up with an image that fits a word is much simpler than taking a piece of music and generating an image that fits, but both are now possible. I also use AI tools for simple brainstorming. Today, I use ChatGPT to create a list of questions I should ask a prospective client about or generate images for marketing materials. Back in my food innovation days, I might have asked ChatGPT to come up with a way of packaging chocolate that allows someone to carry it around without it melting.
- Accelerated personalized education—Generative AI can help my clients bridge the gap between functions. Imagine a salesperson who doesn’t understand what a data analyst does. They could ask an AI assistant to listen in on a conversation and explain what that analyst is saying, or suggest better questions to ask. There are endless examples of ways that generative AI can become a tour guide in unfamiliar territory and help us work more effectively across disciplines and functions.
- Automated human-like interaction—With the help of video generation platforms like Synthesia, it is now possible to type in a script and watch a video of a computer-generated person reading it in several languages. If a client is creating a series of training videos, this tool could dramatically improve accuracy and reduce cost.
SY: What are some common pitfalls you help your clients avoid?
JS: You would never want to start with a hammer and look for ways to use it. In the same way, it is a mistake to take a generative AI tool that seems interesting and look for ways to apply it to business operations. As a marketer, I believe that you always start with the problem. Define the problem and search for a tool that can address it, whether or not it is related to AI.
Just like clients can become too enamored of generative AI, they can also become too cynical. Maybe I’m too much of an optimist, but I view generative AI through a positive lens. Even with the tremendous social disparities in place today, think about the ways that innovations in fields like medicine have vastly improved living conditions. I’m currently consulting for a data analytics company that is building a new AI-powered app to help doctors create better patient care reports. Another client is working with machine learning models in the drug discovery space to predict new drug candidates. Humans will find a way to abuse every technological advance, but the general arc of history bends toward progress. The good will outweigh the bad.
Sevin Yeltekin is the Dean of Simon Business School.
Follow the Dean’s Corner blog for more expert commentary on timely topics in business, economics, policy, and management education. To view other blogs in this series, visit the Dean's Corner Main Page.