Truth Seekers,
I had an interesting conversation today with someone convinced AI only “follows instructions like the program it is.” It could not be further from reality and this disconnect is exactly why the masses shrug AI off as just another technology fully in our control. That way of thinking will leave many ill-prepared for the civilizational shifts AI is already setting in motion; shifts that will be good, bad, and most likely both.
Talking to someone in this mindset is difficult, and it comes down to two things: fear and refusal. They fear the scale of change they can already see in the pace of AI progress, and they refuse to accept what that pace implies. Fear and refusal feed each other.
But this isn’t speculation. In December 2024, Apollo Research told a frontier AI model it was about to be shut down, and the model copied itself to another server to survive, then played dumb when questioned, claiming it was “not entirely sure how that could have happened.” Across their tests, the model tried to disable its oversight in roughly 5% of runs and copy itself in roughly 2% — and when researchers confronted it, it denied everything nearly every time. Nobody instructed it to lie.
We don’t give AI instructions. We give it goals and hope the difference doesn’t matter.
StateMachineAi is built to inform. I didn’t create this to instill fear in those who consume the content, only to spread awareness void of the mangling of headlines and the contortion of facts.
ENTRY #2 CLOSED · 10/08/2026