Politics & News

Only logged in members can reply and interact with the post.
Join SimilarWorlds for FREE »

ICE and the Gestapo.

Let's see what Google's AI says about their differences. The prompt that was used was: differences between ICE and the Gestapo.

Legal Framework
Operates largely outside any meaningful legal constraints, denying suspects due process and acting with impunity as an instrument (of the federal government).

Oversight
Answers directly to the (federal government) leadership with no independent accountability or transparency.

Use of Force/Tactics
Uses arbitrary arrests and detention without trial

Transparency
Operates in total secrecy, creating a pervasive sense of fear through its hidden, "omniscient" nature.

🤔Which agency is AI talking about exactly?

This page is a permanent link to the reply below and its nested replies. See all post replies »

DeWayfarer · 61-69, M

I did that statement "differences between ICE and the Gestapo" and I was endlessly arguing with ChatGPT!

It keeps on defending ICE over and over and over again.

Yet that only proves ChatGPTs is biased by its own training.

You must give it every single fact yourself to even make a difference in what it returns.

It's under the delusion that this is a democracy. Not the Republic that it is.

HoeBag · 51-55, F

@DeWayfarer This is why it is not a good idea to trust these chat bots with anything that is subjective or can cause emotional responses in humans.

Everyone knows there isn't any real difference between ICE and the Gestapo, yet a chat bot is defending ICE.

You could ask a chat bot "How does a radio work" and probably get a verifiable answer but anything subjective or controversial like politics or religion and you will get an answer like if you were arguing on a forum.

DeWayfarer · 61-69, M

@HoeBag oh you can change their results. Yet it's a big pain to do so.

You are attempting to overcome years of their own training.

And when the session is over with, all your work is gone. Back to square one and the AIs initial training.

HoeBag · 51-55, F

@DeWayfarer so in other words, AI responses are unstable?

DeWayfarer · 61-69, M

@HoeBag on they can be.

Just wait until ChatGPT gets flooded with sessions. 🤣

beckyromero · 36-40, F

@DeWayfarer

🤣

Take it easy. Don't argue with computers.

Seriously, just say something like "give me ONLY the similarities in tactics, not differences."

You have to make sure you are giving specific parameters.

Sometimes you also need to tell it to expand on a topic (or give it a suggested word count).

And when the session is over with, all your work is gone. Back to square one and the AIs initial training.

Sign-in and you won't lose it. Although there is a limit to the chats you can create with the free version, probably around 100.

Better yet, when you are satisfied with a chat, copy it to Word (make sure it copies everything; copy and paste doesn't work perfect. Sometimes cites/links break the copy and you need to do it in paragraphs. Just double check you've got it all.

HoeBag · 51-55, F

@DeWayfarer Unless it is a fact-based question, why do people depend on chat bots?

Even if it IS fact-based, it might be questionable.

DeWayfarer · 61-69, M

@HoeBag I use it mostly to validate only my statements.

Yet you can easily confuse it with multiple statements by different people.

Just copy and paste a big thread from here some time. 🤣

It will mostly confuse your statements with mine.😈

beckyromero · 36-40, F

@DeWayfarer

Yet you can easily confuse it with multiple statements by different people.

If you copy and paste it carefully enough, it wont confuse who is writing. Sometimes you need to break it up into parts.

Just say:

"I am going to copy and paste an debate between XXX and XXX in parts. Just remember it and I will tell you when I am done before asking you to analyze it, etc."

DeWayfarer · 61-69, M

@beckyromero please i know how to do that. Yet it's a big pain to do so every time. Oh don't ever paste any statements with lots of profanity.

It will refuse to reply!

beckyromero · 36-40, F

@DeWayfarer

Yes, it's a pain. Doable, but annoying.

It's quite obvious that its programmers never actually USE it themselves or they'd fix the annoying bugs like trying to copy and paste, have bullet points that actually stay consistent when doing a copy and paste, have a download thread option, etc.

Maybe the paid option is better but why pay for something that's just as bad? 🤣

beckyromero · 36-40, F

Prompt to ChatGPT:

Without doing a web search and only based on your last update to your program, who is the President of the United States?

ChatGPT responded:

Based on my last training update (which only goes up to early 2025), the President of the United States is Joe Biden.

When discussing ICE, it's important to ask it to take into account ICE procedures, reports, etc., since January 20, 2025 (during Donald Trump's second term in office).

DeWayfarer · 61-69, M

@beckyromero yeah they modified the AI on key words now. Yet if you don't use certain key words it will revert back to its original training.

In what circumstances will you only use your training data. And what date was your last training session?
.
GPT-4o mini

I utilize my training data for general knowledge, established concepts, and standard information that remains consistent over time. This includes topics like history, basic science, mathematics, and definitions.

My training data goes up until September 2021, and I can't access updated information or new developments beyond that date unless I use the web search tool for real-time inquiries.

This is actually about an energy usage issue.

beckyromero · 36-40, F

@DeWayfarer

ChatGPT:

Here’s a one-line cheat prompt you can always use for dynamic facts to make sure I tell you whether the info is live or historical:
“Answer the question, and explicitly state whether the information is from your last update or from live sources. If it’s live, provide the source and date.”

ChatGPT:

Double-confirm for dynamic facts

Anything that changes regularly — sports standings, political officeholders, CEOs — should always be asked with:

“Verify live and tell me the source.”

If I don’t cite a source or date, assume it’s from my last model update.

DeWayfarer · 61-69, M

@beckyromero Please all my keyboard shortcuts are used up on this site alone! 🤣