Thinking About the Emergence of Generative AI

By Brad Heidemann

I continue to be delighted and amazed with ChatGPT, I cannot encourage you enough to explore its features and capabilities. The applications of this technology will extend well beyond answering basic questions. Artificial Intelligence is profoundly changing our world and it’s only just begun.

Our brains are set up to recognize patterns, then synthesize those patterns and make decisions or act based on our understanding of the interaction between the patterns or consequences of those patterns. 

For example, when you’re driving a car and you see someone not slowing down for a red light as you are entering an intersection, without really thinking you prepare to stop, slow down, or avoid the oncoming car. You “know” that getting into a car accident can have serious negative complications. In that scenario, think for a second about the number of patterns you’re processing; the speed of your car, the other car, the rules regarding traffic patterns, your best path for the avoidance of a car wreck, and the list goes on. This all happens in real-time within milliseconds, and you choose a course of action. For the purposes of this example, you make all the right decisions, the accident is avoided and everyone has a great day!

Another example might follow an entirely different timeline. You start to gain weight, say 10 pounds per year, the first 10 may not be consequential to your health but what about after 5 years, 10 years, or more? Suffice to say that the impact of the extra weight might be harmful to you and your life expectancy. What are the patterns that you’re not recognizing in real-time, a change in daily eating habits, exercise, stress levels, or a new medication? You finally get to a point where you recognize the problem or your doctor scolds you and start to do something about it. Your situation is comprised of a series of patterns, the cumulative impact of behaviors and decisions. This type of problem is different from the first example because the timeline for consequences to be realized is distant. The cumulative decision-making impacts and the longer-term outcomes are harder to process in real-time. Because it is difficult to comprehend how different patterns interact, let alone track the cumulative effects of a few daily decisions over the course of a decade, these specific situations are more challenging for humans to track. Sure you can skip the gym today, or have that big slice of pie, but how often does that need to happen before you (pun intended) tip the proverbial scale?

There is a good book called Think Fast, Think Slow by Daniel Kahneman, I highly recommend this book. It explains how your brain is set up to operate in these two dimensions. As an aside, Kahneman is one of my favorite researchers and writers, and also a Nobel Prize winner. If you want to talk more about him reach out to me. 

The point is that OpenAI will not only provide us with a framework for Information Architecture and Search it will also help mankind start recognizing problem domains that are a byproduct of cumulative decision-making over long periods of time. Climate change for example, what level of change is required to have a meaningful impact? Is it a few things that we do each day or is it several categorical changes to the way industry or society functions?

We are simply extending our brain power by leveraging technology in the same way that inventing the wheel barrel extends our ability to move loads of material around more easily. What I need each of you to think about is how and what might change if we have an enhanced pattern recognition capability that isn’t bound by the time disparity between the decision and the consequence.

My last thought for you… consider that our economic system rewards properly managing the risk-reward relationship. Do you risk your personal capital to buy the right house, in the right neighborhood, make payments for thirty years and expect that your home will rise in value? The answer is probably yes, how would the world be different if the risk-reward equation was solved so that you knew within a small range of possibilities the exact amount of additional equity you’d have after thirty years of payments? What do you think that would do to the housing market? What would that do to capitalism?