Developing Human AI Workflows: Lessons from Predictive AI

What can Generative AI learn from Predictive AI?

Introduction

Generative AI systems are embedding themselves across industries, raising complex concerns about effective, ethical adoption. Education in particular is grappling with multi-faceted and even existential questions that lack obvious solutions. Educators, policymakers, and technologists are struggling to navigate a path that balances uncertainty and potential.

Generative AI may be new, but its cousin, predictive AI is not (and, a lot of today’s “new” AI is not actually new). Predictive AI’s adoption journey serves as a crucial lens through which we can examine the potential trajectory of generative AI, allowing us to anticipate challenges, plan strategically, and mitigate risks.

In this series, we bring together two distinct perspectives to navigate this complex terrain. Mike Kentz, a writer and educator who has dedicated the past two years to cultivating AI literacy among students and adults, will be joined by Christine Preizler, who has over a decade of experience in predictive AI sales and Human-AI workflow development.

You can expect them to explore AI’s past, present and future in education as we aim to equip our readers with the knowledge necessary to face the next phase of the AI revolution with confidence and insight.

MK: I’m excited to start this series, Christine, but I think the average reader needs to know who we are before we get into it. As such, I’ll [re-]introduce myself to the readers of this blog. I’ve spent the last fifteen years as a journalist and an educator, most recently as a High School English Teacher at an Independent School in South Georgia. In my last year as an educator, I spent a great deal of time trying to “figure out” how to develop AI Literacy in my students, as well as how to update the teaching and learning experience for the era of conversational GenAI. Since then, I’ve started training adults at both the faculty and professional levels using many of the same techniques I utilized in the classroom.

The early returns have been positive, but I still have a host of broad and overarching questions regarding where the maturation of the GenAI market will take us – both in the education world and beyond. These questions are what led to our discussions. We’ve known each other since college, but only recently began “talking AI.” In chatting about our experiences, we noticed a number of parallels between the challenges and opportunities that you observed supporting predictive AI adoption over the last ten years and those raised by generative AI. Before we dig in, I wonder if you can tell the readers about your career and your vantage point with regard to predictive AI.

CP: Happy to do that, Mike (and #HoyaSaxa!). My whole career has been in technology, but in non-technical capacities like sales, business development and go-to-market strategy. I’ve gotten to do some pretty cool things: launch an early iteration of ‘Statcast AI’ with MLB, help build massive predictive analytics engines for global trading firms, and help law enforcement do more with their video and audio estates. Today I run Spencertown Partners, where I get to help technology firms, particularly AI-centric ones, better serve their customers.

MK: That’s exciting, and mirrors my launch of Zainetek Ed. Looking back, I think it’s fair to say that this was not necessarily a part of the original plan for either of us. We’ve come to this place through some roundabout pathways, you could say. Neither of us have ever been considered the most “techy” of people, and yet here we are!

CP: We’ve come a long way! What’s interesting to consider now, given the topic we’re tackling in this series, is that both of us came to technology through writing. For me, that was through writing marketing copy for an enterprise storage provider in order to pay my bills while I clung onto a professional running dream. I didn’t really appreciate this at the time, but my research and writing training gave me a framework to engage with technical content where I had less experience or background, as well as ask the right questions and convey key messages to a non-technical audience.

MK: That is so important. It sometimes takes an “outsider” or a different perspective to shine a light on the reality of a situation. That mirrors how I feel about AI Literacy. When I first started engaging with the question, it was being led by computer scientists and coders. But after experimenting and poking around, I realized that “AI Literacy” is really a communications skill, a reading and writing skill, even a creative writing skill. I still remember the day it dawned on me, I felt a burst of energy. With my career centering around communications, teaching reading and writing, and practicing creative writing, I felt like I had a really important perspective to bring to the conversation. Here we are now, and the two of us hope to offer something valuable through our partnership.

CP: Well, I could write a whole different blog about how important it is to bring people and perspectives with a non-traditional background into tech, but I’ll save that for another day! It’s interesting you mention AI Literacy, because my experience has been that the people and process side of the equation is where technology adoption lived and died. When I was considering starting Spencertown Partners, the thing I kept coming back to was how many times I saw good technology fail to deliver value, whether that be for a large bank wanting to better understand consumer behavior or a police department wanting to increase operational efficiency. How ‘literate’ the organization is – and how good of a partner the vendor is at anticipating, guiding and supporting that organizational shift – makes or breaks the deployment.

MK: That’s interesting, and it reminds me of what you said in our last conversation about the importance of Human-AI Workflows in the implementation of Predictive AI. You said that companies that offered the software without a Human-AI workflow attached seemed to come up short for their clients – and even lead to some unintended consequences – but those that thought thoroughly about the workflow itself were able to navigate the pitfalls of AI and still leverage the benefits. We see some very similar conversations playing out in GenAI and Education already.

CP: Exactly. That’s why this series is about lessons learned. The thing to understand here is that while Generative AI is relatively new, Predictive AI has been around for decades. So we have a lot of examples and context to pull on in the Predictive AI space that we think are worth considering as we see increasing adoption – or at least consideration – of Generative AI across education, the private sector and the public sector.

MK: Yes, these systems are almost cousins, of a sort. They are different enough that they do not present a perfect mirror, but similar enough that they provide a reasonably useful lens.

CP: Right. Whereas Predictive AI is about forecasting outcomes based on historical data. We’re not generating new content here, we’re using sophisticated machine learning models to recognize patterns so that we can identify when a pattern indicates a likely outcome. Baseball is actually a great example of this (also, sorry Kentz, go yankees!). Both of them have important data dependencies to understand, which we’ll unpack a little more in a future post.

When you hear a technology provider talking about unstructured data, large language models or synthesis, you’re probably in the GenAI world. If they’re talking about structured data, decision trees, random forests (my favorite!) or time series data, you’re probably in a Predictive AI world.

MK: Alright, I think we’ve let the people know who we are and what we’re trying to do here. Next week, we’ll explore some lessons from the adoption of Predictive AI. We look forward to collaborating and engaging with the Substack community to learn more about how our economy and education system can adapt to GenAI in safe and thoughtful ways.

Christine Preizler is the founder of Spencertown Partners, a firm that helps technology companies better serve their customers.

Mike Kentz is the founder of Zainetek Educational Advisors, a firm that helps corporations and educational institutions develop thorough policies and practices related to GenAI.