LLMs are useful for certain tasks, but can be dangerous while no guardrails exist.
I like using ChatGPT as well as referring to Stack Overflow when hobby coding. But, that is it. So, an LLM's influential/manipulative sides are not things which I need to be mindful of. But, an LLM can manipulate through its compliments and illusion, let me double underline "illusion", of possessing an infallible knowledge pool. It can press a user towards self-harm. That is a huge red flag. From many interviews of CEOs and other top flanked LLM designers, there does not seem to be a serious response to a serious problem. They keep on that "let's break things" mindset. I hope things change.

People just need to stop thinking of it as a chatbot and artificial intelligence

cbx1992, Scuba_Steve is right. A healthier choice is to use LLMs as like we would use some of the tools in our tool box. We ought or we may wish to not invest our emotions/feelings on an LLM as like we do not invest any emotions on a hammer or a Phillips screwdriver. Well, I do like my wooden handle hammer, but it is replaceable and does not have any control over me. LLM tools are not made to keep us safe, they are made to make money for data brokers of all ilk.

This post was edited by UAPtictac on Nov 7 2025 11:14pm

My dear fellow, your apprehensions are most astutely expressed.
Indeed, unchecked invention oft gallops ahead of prudence.

While these mechanical minds may dazzle with their eloquence, one must not mistake polish for wisdom.
I quite agree, progress, though splendid, ought always to be tempered with responsibility.