Dallas Fed AI Projections - A Curious Mix of Wow, Hmm, and ... Well That Can't Be Good

There was an old Dilbert cartoon in which Dilbert learns how to use PowerPoint for presentations. In the first panel, he shares this fact. "Bad news: our product is killing people." In the second panel he presents this instead. "Good news: drop in the number of dissatisfied customers."
Here is a curious graph from the Dallas Fed, which offers four scenarios to capture the possible effect of AI.

1. No change. Per capita GDP continues to grow at 1.9% a year as it has for more than a century.
2. Significant change. Per capital GDP begins to grow at 2.1 to 3.4% a year. (And this is not to be scoffed at. Given the size of the economy, this is a difference of trillions of dollars, a rate of progress equal to or significantly more than what we've had since the Industrial Revolution.)
3. A benign, or positive singularity in which our lives are so transformed by AI that we gain entirely new capabilities and productivity soars beyond what one might reasonably project now, changes more profound than what percentages could capture.
4. Or this singularity results in our extinction, AI deciding that we're either superfluous or detrimental to progress and ridding the planet of us the way we have tried eradicating various viruses and diseases.
So, the most plausible scenario is that because of AI we will be making ten thousand or more dollars each year than we otherwise might in a decade (an outcome bounded by imperceptible change on the one hand and remarkable progress on the other) and in the worst-case scenario we are extinct. But as Dilbert might point out, in this last scenario there are no complaints. I mean, can it really be called a risk if there is no one left to complain?

https://www.dallasfed.org/research/economics/2025/0624