If you are watching the news here in the US, we again have the side in power proposing fixes to problems and the side out of power blocking those fixes. The show and tell aspects of their presentations go back to the 1980s, when we were still doing flip charts or having graphic arts departs making exhibits. Politicians in the US Congress generally don’t even use the 1990s PowerPoint tools, which have arguably also aged out.
I’m using the government as an example because they tend to spend money by the hundreds of millions of dollars, which should justify doing real modeling and simulation. In the Ronald Reagan era, you could argue something like Trickle Down Economics without knowing whether it would work or not. But with the tools of today, you can model concepts like this and anticipate the impact. This capability is essential because, while the idea might be right, it would help refine an implementation that assured success or point out that the idea doesn’t work, which, sadly, turned out to be the case.
Even before we had today’s computing power, military organizations used simulations (war games) to understand past and future battles. Still, these efforts were historically hampered by the limitations of technology. Simulations are limited by the assumptions surrounding them. They can be biased but, when tried by video game-based rendering tools, you can not only create more compelling and exciting examples, you can adjust the parameters in real-time to address counter-arguments or see if altering variables could improve the outcome.
Let’s talk about applied AI and Simulation this week.
The Missed Advantages Of Applied Simulation
Currently, simulations are being used to develop airplanes, train pilots, develop military gear and train operators, develop autonomous cars, prepare them for the real world, and walkthroughs for buildings that haven’t been built yet. But they are rarely used to flesh out and sell ideas to VCs or politicians.
Years ago, when I was in the competitive analysis sector, part of my budget went to the best presentation tools I could find, and I regularly outperformed my peers who didn’t see the value of telling a rich story. They not only allow you to simulate the impact of the decisions you are advocating, but they also present those results in a far more compelling way than just words or static slides.
Let’s take the $15 minimum wage argument. Just looking at both sides of it, they are incomplete and flawed. A simulation, based on the CBO analysis, could immediately showcase and help create a bipartisan plan that could be far more effective than either side is currently proposing. For example, in California or New York, you couldn’t live on $15 an hour, so the liberal plan isn’t enough. Still, in the middle of the country with massively lower living costs, $10 an hour might be more than adequate, suggesting the liberal plan is too much.
This discrepancy between locations on cost would suggest that minimum wage should be indexed to local living costs allowing it to increase or rarely decrease when those costs change. So the result would be something that could evolve naturally rather than taking up legislative time that could be better used on other future projects.
But you can model everything from the initial impact of a decision to its likely outcome, and, depending on the complexity, accuracy. How far in the future the model is asked to forecast, you can get a far better idea of what program is best. Chances are, like I pointed out, with minimum wages, the best solution will be a hybrid of what the two sides believe depending on location and cost of living.
Why Models Aren’t Used More Often
The issue is that we are all status focused and endemic in the human race is a need to appear right. You can state a belief and defend it without facts but, when you bring in a modeling program, you will almost always find that some part of what you believe to be right isn’t.
Let me give you an example. I was in a Competitive Analysis group at at large corporation, and they sent a product that was created for us for review. We did a detailed study and concluded the product would not only not sell, but it might cause the company to fail. We convinced the engineering executive that we were right; he went back and was reassigned. The company sent another executive out, and we convinced him we were right, and he went back to the company and was reassigned; the company shut us down. The problem behind this behavior is called Argumentative Theory. It suggests that sometime in our past, we developed a genetic need to be seen as right, regardless of the facts. Being right equated to more power and getting the best mate, breeding this foolish behavior into our makeup.
So models can help you avoid mistakes, but only if you accept that you may be wrong and are more focused on taking the correct path than appearing infallible.
Wrapping Up: AI-Driven Models
Digital assistants will continue to gain capability and eventually use AI-driven models to help us make better decisions. Becoming comfortable with taking advice from an AI will still likely take time. This behavior suggests we should move these models and simulation tools aggressively into education. Hence, students learn how to make unbiased models and learn how to identify unbiased models and simulations and trust them.
Particularly in politics, we need to move to something less about beating the other side up with questionable facts and more about developing bipartisan efforts that have a higher probability of success. This outcome will only happen if we learn how to develop unbiased models and simulations coupled with a willingness to work together to find the best answer rather than roll over anyone who disagrees with our poorly founded ideas.
In short, it is time that both PowerPoint – and the people that use it – evolved to use models and simulations to reach better decisions, resulting in better outcomes.