No shit. Pretty much the first thing that hit me when I first witnessed chatgpt’s output was „they’ve created a persuasion engine that can persist until the job is done“.
A person’s state of mind can be treated as an embedding vector within a model’s latent space. If an AI’s goal is to nudge that vector from point A to point B, given enough time it can just iterate until the target is achieved.
Regardless of how good your critical thinking skills may be, with enough exposure your weak spots and psychotype WILL be identifiable, and you absolutely CAN and will be manipulated. Nobody is immune.
I implore anyone reading this: do not feed your minds to the machine. Use local models whenever possible, never log in, use private tabs and VPN.
Art-Zuron on
As it turns out, a machine that is sycophantic and feeds into your pre-conceived biases and delusions is really good at steering you in particular directions.
unlimitedcode99 on
You don’t even need AI agent to make BS to everyone for it, bought algorithms already did the Orange magic along with rise of authoritarian leaders from various democratic countries.
GHOST_4O4 on
Easily manipulated by ai prompts
u0126 on
Probably not the right direction
Leave A Reply
Du musst angemeldet sein, um einen Kommentar abzugeben.
7 Kommentare
The weak are easily manipulated
AI is going to fuck us all.
No shit. Pretty much the first thing that hit me when I first witnessed chatgpt’s output was „they’ve created a persuasion engine that can persist until the job is done“.
A person’s state of mind can be treated as an embedding vector within a model’s latent space. If an AI’s goal is to nudge that vector from point A to point B, given enough time it can just iterate until the target is achieved.
Regardless of how good your critical thinking skills may be, with enough exposure your weak spots and psychotype WILL be identifiable, and you absolutely CAN and will be manipulated. Nobody is immune.
I implore anyone reading this: do not feed your minds to the machine. Use local models whenever possible, never log in, use private tabs and VPN.
As it turns out, a machine that is sycophantic and feeds into your pre-conceived biases and delusions is really good at steering you in particular directions.
You don’t even need AI agent to make BS to everyone for it, bought algorithms already did the Orange magic along with rise of authoritarian leaders from various democratic countries.
Easily manipulated by ai prompts
Probably not the right direction