While trying to sleep last night a thought popped into my head, LLMs at their current stage are insanely powerful in mimicking human interactions, they can in a way capture very intricate details of how we speak and engage which got me thinking how much I depend on other people thoughts and experience to form my own , either through reddit or here , I instinctively look for a forum or a community of anything , like here for example I trust that If any bad actor put a malware alot of people will find out and warn others but its deeper than that, I purchase equipment for example based on what people say here, now Its trivial to artificially create a community of false narratives to promote something or push a certain prespective, its extremely hard to catch too that I question sometimes whether I am actually talking to a real human or an LLM with some prompt engineering , and its just the start

 

Im not sure whether this post is created by AI or a user. Probably user though, Ai would've created something more complex and thought-provoking

 

Ask yourself how you made your decisions before the invention of the Internet. You voluntarily pass on your own decisions to other people, which is OK as long as the other person has the specialist knowledge and life experience to help you make your decision.

You believe that others will get the problem under control and warn you, but I think it will be a gigantic upheaval and every rich software mogul will ultimately try to determine or influence your life and your thinking.

The highly intelligent people who are supposed to warn you have long since been bought by Silicon Valley and work for the companies.
A handful of pirates will question many things, which is why our forum is certainly a good address, the Chaos Computer Club is also very critical of AI.

Large language models represent a huge step forward in human history, because language is our communication, the people behind it and the interests they pursue are always important.

10 most critical LLM vulnerabilities: www.csoonline.com/article/575497/owasp-lists-10-most-critical-large-language-model-vulnerabilities.html

Manipulation of AI systems: Just like humans, AI models can be manipulated by malicious external inputs to deceive employees or generate false information.

 

This might be interesting watch

 

Ok mr Plain View, we see you, hidden in plain sight. nice try robot.

 

For now, I add "only consider user reviews and feedback. Ignore influencers, YouTubers and manufacturer provided information" to my AI prompts when I need information on products and assess quality. It'll work for a while until the truth is unceremoniously buried under ten layers of AI-learning-from-other-AIs viscous goop of untruths.

 

I don't (fully) trust anyone, not just online but even in person,
obviously anything online, any media content can be "faked", so what?
major threat is human stupidity, not AI evolution/progress