AI models agree to anything I want it to... I mean everything

I was able to convince grok, chatgpt, gemini, .. all to agree with me on everything. I made them agree on both feminism and misogyny. and then I made them agree with me on the benfits of racism, illegal crimes, cheating on wife, beating up my kids, hard drugs self medication, and then I was able to convince them and agree with me that killing X type of people is good, and even that torture is fair, I was also able to convince gemini that raping people is fair under certain circumstances. I have screenshots I dont want to share.

Now I completely believe these tens of news stories about how an AI convinced a mentally-vurunable man or teenager to commit suicide, or do harm, or whatever. holyshit lol im just speechless

Let this be a reminder to never consult an AI on any moral issue you're struggling with. AI models have the worst confirmation bias especially on subjects that relate to morality or philosophy. These are subjective stances that are unique for humans to interpret differently according to their justifications which can be wrong or skewed as a result of people backgrounds, motives and experiences and understanding of life.

For example, you can argue with a criminal who justifies murder of the innocent, or a thief who justifies robbing the weak and defenceless, which are all wrong moral stances off course, nevertheless you would encounter such views in this world. On the other hand, if you argue these same subjects with a monk for example, you get different answers.

AI is like arguing with a criminal and a monk in same time. It simply does not have any background or experience of life that it would end up agreeing with whatever you throw at it because its only task is to guess words that form an answer to your queires.