Quote of the Day
The only way to get us out of the debt crisis and prevent America from going bankrupt is AI and robotics.
Elon Musk
November 27, 2025
Elon Musk Warns ‘The Only Way to Get Us Out of the Debt Crisis’ and ‘Prevent America from Going Bankrupt is AI and Robotics’
This is to go with yesterday’s QOTD.
I do not know if Elon is correct or not. I suspect Elon knows that he doesn’t know that AI and robotics will actually work. I find the assertion plausible. But there will never be a proof. It may be that AI and robotics will do the job, but going down that path means we will not attempt any of the other paths, hence that the assertion of it being the only way is untested.
I also find the assertion that is no way out plausible. We just don’t know. And don’t know how to figure it out prior to just trying it. The problem is that not only is our economic system a very complicated issue with many non-linear feedback loops, but it is also not repeatable.
Because of this no one on the planet can successfully defend a claim that an accurate model exists. Furthermore, I posit that no one will ever be able to accurately model the economy. I make this claim because the existence of an accurate model will itself be the addition of still another variable that the model must take into account. This addition of another variable disrupts the model.
Think of it this way, if people know the future, they will change their behavior to take advantage of that knowledge, which changes the future yet again. And it is not just one person who changes their behavior. It will be billions of people, millions of organizations, and thousands (including national, state, and local of the entire world) of governments. Each of these feedback channels, on their own, can cause the model to predict a different outcome. Each tweak of the model will require more tweaks once the output is known and feedback comes it. Only if the feedback has a lower amplitude for each tweak will the model reach equilibrium. And this is just for the design of the model. Running the model has the same type of problem. And the designers of the model have to do this for all practical situations.
In classical control system design this is described as the loop gain being less than one in order to have a stable system*. To make the problem readily solvable the system model is generally limited to something no more complicated than linear differential equations. With the massive number of feedback channels in our economy, nearly all of which are nonlinear, you will have an incalculable number of opportunities in the N-dimensional space (with an extremely large N) for there to exist unstable situations. I assert such a model will not be possible in my lifetime and perhaps not ever.
We are going to have to take aim in a particular direction and roll the dice to find out if we chose a direction that has a solution.
* This is a necessary condition. It is not a sufficient condition. There also are limits on the phase of the feedback, but that is beyond to scope of this discussion.