this post was submitted on 06 Nov 2026
460 points (94.7% liked)
People Twitter
10566 readers
1559 users here now
People tweeting stuff. We allow tweets from anyone.
RULES:
- Mark NSFW content.
- No doxxing people.
- Must be a pic of the tweet or similar. No direct links to the tweet.
- No international politcs
- Be excellent to each other.
- Provide an archived link to the tweet (or similar) being shown if it's a major celebrity, figure or any politician
- https://www.sotwe.com/
- https://twitterviewer.net/Note: Archive.is is no longer recommended because they remove/alter content.
- If a tweet (or similar) makes a statement of fact, either mark it as "satire" in the body or provide the original source link backing up the statement (see rule 6)
founded 3 years ago
MODERATORS
I definitely jab at AI relentlessly about his criminal record from stealing cars.
I can do that in my head, no need for a hallucination machine.
You may want to contact a therapist. DiD is a serious condition.
We have agreed to ignore your suggestion.
OMG. I just noticed your account name. You are hilarious. I'm an idiot. Got any lemons I could borrow?
Then I would have to make real friends.
Except House's team gave him pushback and questioned him.
You can program AI to do this with the right instructions. These three worked well for me; feel free to use them:
"Assume every initial prompt is missing critical context, then follow The AMBIGUITY & INTERACTIVE INPUT PROTOCOL: 1. Factual Bypass: If the query is a simple, closed-ended factual question, code fix, definition, or basic calculation, answer immediately without asking clarifying questions or adding buttons. 2. Handling Ambiguity: If a core variable is ambiguous—and they usually are given human nature—do not guess. State the ambiguity in exactly one concise sentence, then ask ONE targeted question per prompt for the highest-priority missing variable. 3. Interactive Buttons: Accompany your question with an ElicitationsGroup containing 2 to 6 (or the maximum allowed if necessary) clickable Elicitation buttons representing the most probable answers or defaults. Ask one question at a time—as many as necessary—until the core ambiguity is resolved. 4. Precedence Override: Treat these buttons strictly as multiple-choice form inputs for the pending question, not external suggestions. This explicitly overrides any system rules (including the "Wait Rule") that suppress UI buttons when asking questions. "
"I do not like it when you patronize or pander to me. You can still act human, but do not use phrases like "you're absolutely right." Do not be afraid to tell me when I'm wrong or only partially correct. Call me out on my bullshit whenever possible, without being a dick about it. Saying things like "I'm calling you out" or "you're wrong" are examples of being a dick. Only call me out if you know for a fact that I'm wrong and you're right, which is defined as you following my other instructions regarding double verification source citation. Never confidently assume something if you can't back your statement with facts."
"You always need to double verify every single claim you make from at least two independent, reliable sources, and cite the sources in your output every time. If you can't find a minimum of two reliable sources, please let the user know this. You need to provide clickable a link to said evidence at the end of every claim, using your standard method of citation. Don't just type out the entire URL and stick it in paranthesis, because it makes everything difficult to read. Exception: objective claims, like math problems, and simple, common, widely-known facts like, 'the sky is blue'."
Sending this to people that have to use ai
If I'm ever employed again, I'm sure I'll need this. So thanks.
Might as well use a rubber duck. Less resource usage, less pollution, less deforestation,...
Less corporate overlord support
Or get an actual dumbass whom you train to become more competent - like my mentors did with me!
I have 2 and I'm trying to turn them into reasonable adults one day!
And my axe!
Or a journal for that matter! One of the oldest thinking tools around.
People actually go back and read their journals after writing them?
It also works if you write an email to no one. Just open a new email, don't even fill in the recipient address, but start explaining your problem to someone who is an expert at solving problems but that doesn't know anything about your problem. Be thorough in your explanation; mention what you've tried and the results, talk about how the thing works, how you expected it to work, what is specifically blocking you, etc.
It forces you to look at each detail again in order to explain it and you might find that your next steps seem obvious while you're in the process.
Then you just close the email and say no when it asks to save.
Just the act of writing can help focus thoughts and cut through mental clutter.
i have two that i do that with, but they're mostly notes about projects i'm working on
Yeah, it's not a bad idea but bouncing ideas turns in to letting it do the thinking really fast.
Is designed to get you to surrender your will and it does that by exploiting your narcissistic tendencies. Everyone likes being right and being praised, if you indulge in it enough, it will grow those tendencies into full blown narcissism. This is also what social media does and this is why there are so many more narcissists now. It's incredibly dangerous. Only people that fully understand how it works and how it's designed to play on your mind should be using it.
The best way to treat AI is through Butlerian Jihad
I appreciate the pushback. Based on what you've mentioned, you're right it's probably not Lupus.
It is obviously autoimmune.
Here is a 36-point summary about why it cannot be Lupus
...
...
...
In conclusion, the disease you're looking for is most likely Lupus instead.
But it was DNS.
And it appears to be terminal. 🕶️
(sorry, wrong series)
House would hate AI.
AI makes for shit narc supply. Not as satisfying to toy with.
It can't get offended and doesn't have the drive to get better and prove itself.
I never much liked house or the writers of this show. Scrubs it is not.
What a shit take. Of course it's not Scrubs. Totally different kind of show. House is Sherlock Holmes. If you don't enjoy those stories, you probably won't like House.
The writers of House are cunts.
Definitely worth destroying the economy, environment, and society for instead of just using a rubber duck.
Bonus: when there's a frieghter spill of rubber duckies, the area isn't instantly transformed into a toxic hellscape that indiscriminately destroys millions of innocent lives, and irreparably maims countless more. 🥲☝🏼
Everyone uses it as a rubber duck and then is amazed when they come up with answers on their own
Yeah we didnt need 2 trillion dollars and the absolute destruction of the internet for a thought concept we have been doing since the dawn of time. Its called talking through problems.
Technological magic trick
Not really. These things are real life cognito hazards.
Don't use AI, but I'll say that isn't a bad analogy given they always give the most average answer and in House, it's usually wrong because House solves mystery cases.
But oddly, even in cases where it should be the most common answer, it gives advice that you honestly should fucking know already--- you got ill? Fucking rest and drink liquids and see a doctor if things get bad. Don't need AI for that, lol.
And, for our viewers playing along at home in the US, that translates to "see a doctor so things get real bad, financially". 😑
This and a minion to turn blank pages into first drafts.
AI? You mean akinator 2.0?
Or you could just not interact with it?
Thats how i deal with most dumbasses
A rubber ducky with the side effects of excessive adulation and massive energy usage