Suggestions
Gerhard Pilcher
CEO at Elder Research and Author
Gerhard Pilcher is the Chief Executive Officer (CEO) of Elder Research, Inc., a consulting firm specializing in data science and predictive analytics. He has held this position since June 2016, succeeding Dr. John Elder, who transitioned to the role of Chairman of the Board at that time.14 Under Pilcher's leadership, Elder Research has continued to expand its capabilities and services in the field of analytics, serving clients across various sectors both domestically and internationally.
Professional Background
Before becoming CEO, Pilcher served as Vice President and Senior Scientist at Elder Research from 2011 to 2016. His extensive background includes roles in technology and engineering, such as Vice President and Chief Technology Officer at Pulse Communications and Director of R&D Operations at Bell Northern Research.23 He has also been involved in academia as an adjunct faculty member at Georgetown University and currently teaches data strategy at the University of California, Irvine.2
Contributions and Expertise
Pilcher is recognized for his expertise in predictive analytics, machine learning, and data mining. He co-authored the book "Mining Your Own Business," which serves as a guide for executives on leveraging organizational data for better decision-making. The book is frequently used as a textbook in graduate-level courses.25
In addition to his role at Elder Research, he is actively involved on several university advisory boards and serves as Chairman for three start-up companies he co-founded.2 His work has earned him recognition, including being inducted into the NC State Computer Science Department’s "Hall of Fame" in 2017.2
Personal Interests
Outside of his professional endeavors, Gerhard Pilcher enjoys spending time with his family, engaging in outdoor activities, woodworking, and running.2
Highlights
AI isn’t new. 55 years ago, a computer helped steer Apollo 11 to the moon. But technologists are STILL forgetting to ask: “Is AI being used for the right purpose?” Case in point: generative AI. Gen AI is being treated like everything is a nail it can hammer, but it’s best suited for generating or summarizing unstructured content (natural language and images) using the vast information it was trained on—without any knowledge of the correct answer to any question. This means AI is less suited to applications that require a precise and consistent answer. That may change in the future. While I applaud the exploration of Gen AI, there are other ways AI can unlock great value: 1️⃣ Prediction This can be uncomfortable for us because it’s hard to think a computer can beat human intuition, but it’s where AI can have the most value. We tend to apply different standards to the same decisions, from hour to hour and day to day, introducing a lot more uncertainty than a computer. AI’s consistency and ability to process vast data make it more adept at prediction, reducing uncertainty and enhancing decision-making. In fact, we can measure the uncertainty in a mathematical model but must rely on a person’s best guess of uncertainty for a human forecast, creating even more uncertainty. 2️⃣ Process Automation It’s also hard to let machines take some of our tasks because we naturally resist change. Maybe because we think our way is best or we’re afraid the machine will remove the need for us. We’re also typically more comfortable asking machines to do things the same way we’ve done them. But in most cases, we must reimagine the process to optimize the potential of the person and the machine. TRIGGER WARNING: I’ve lost count of the times in my career I’ve been pressured by people saying I was wasting time by taking a step back to more fully consider the process instead of accepting the status quo. For whatever reason, this work isn’t considered progress. Without fail, the investment of time up front created higher velocity and more predictable results. Don’t think this is just an old guy stuck in a waterfall world; this is critical regardless of how you approach development. I love this scene from “2001: A Space Odyssey.” The film premiered in 1968, but it illustrates what we should be considering today: “I’m afraid I can’t do that,” says HAL, the AI. HAL was being snarky because Dave planned to shut HAL down. Ironically, we’ve experienced snarky AI in the modern chatbots, like ChatGPT, five decades later. We need to stop and think about what we are asking AI to do. And we need to use discernment about the claims made by all the shiny AI tools. Some have little to nothing under the hood. Look closely. The real value of AI is in how much it helps humans. Does it address business problems and fit into processes? When AI addresses those things, it frees us to use our creative brains the way they were designed.
Are we building artificial trust in artificial intelligence? Generative AI tools like ChatGPT are helping us with many things:
- Greater productivity at work ➡️ More time with family
- Quicker answers to challenging problems ➡️ Faster and better decisions Generative AI models are continually being trained … and perhaps we’re slowly training ourselves to trust them. Here’s why we should be circumspect:
- Generative AI can generate truthful-sounding nonsense. These tools are only as knowledgeable as the information they’re trained on. For example, ChatGPT 3.5 is trained on data through 2021. That training is unsupervised, which means it hasn’t been taught the “right” or “best” answer to questions, whether or not they’ve been asked before.
- Generative AI can interpret context and sentiment, but it can’t speak for everyone. ChatGPT 3.0 was trained on 160,000 dialogues between two human participants and several other conversational datasets. But those dialogues and datasets can’t accurately represent the more than 8 billion people in the world.
- Generative AI can flip its “values” and “personality” instantly. When new training data is introduced, the algorithms behind gen AI tools adapt to the new information. But that information might lack context, be factually incorrect, or be manipulated to push one viewpoint over another. We’re naturally less likely to question information from ChatGPT, Dall-E, and other large language models. Why? Because that information comes from mysteriously intelligent algorithms that must somehow be smarter than us.
And because the algorithm responds in a human way, kind in most cases, we build trust with it. Even though it has no real human character. So where does that leave us? Does it mean we need to abandon the gen AI ship and swim to safer shores? No. But it does mean we should be vigilant and discerning about how we use AI in our professional and personal lives. What safeguards do you have in place around how you use generative AI? #GenerativeAI #GenAI #ResponsibleAI
