You are so incredibly sure of these wildly optimistic future predictions, but in fact you have limited understanding of the material: I assure you that the people who write AI models understand how they work.

Now, they might not be able to predict how the AI models respond, but that is hardly the same thing - the first guy who wrote a computer chess program that would beat him, sometime in the 1980s, had the same experience.

The idea that we don't understand this technology is anyway not at all a good argument in the direction of "Huge progress, incredibly fast." If it were true, it would be a cause for caution in predictions.