**The Prompt**
It all started when my friend decided to host a writing contest to boost creativity within our circle. The prompt was simple: “Write about your most embarrassing moment.” I racked my brain for a while and finally decided to share the time I tripped over my own shoelaces, fell in front of a crowd, and somehow managed to knock over an entire cotton candy stand. As I recounted my tale with all the dramatic flair of a Shakespearean actor, the room was roaring with laughter—until I realized I had accidentally overshared my worst memory, and my mom was sitting right there, gaping at me like I had just revealed my secret superhero identity.
What makes this scenario hilarious is the unexpected twist—the moment I tried to turn my awkward past into comic gold, I inadvertently marred my mom’s perception of me as a fully capable adult. As my friends laughed uncontrollably, my mom just shook her head, almost proud of her unique blend of shame and amusement. Let’s just say, the next family dinner definitely included a “remember when” discussion that no one would ever let me live down!
in Funny
The Request

P
Needs more gaslighting when you question if the result is definitely 4 and it tries to convince you that 2+2 might actually equal 3
C
I’m constantly getting wrong answers when I question ai. We are doomed.
T
They don’t call it artificial high intelligence
S
they should call it PI for Probably Intelligent
X
AI – Almost Intelligent
M
Adjacent Intelligence
C
Anti Intelligence Anything (but) Intelligence Etc etc
J
Actually Indians
P
Artificial Incompetence
K
Assistant to Intelligence
W
Ai am offended, have my upvote
J
https://tech.co/news/ai-startup-chatbot-revealed-as-human-engineers
S
this! The whole thing is a scam. It’s really just call centres in Mumbai pretending to be AI bots. (or at least that is the wonderful mental image that “actually indians” conjures up)
V
There was a case last year I think where some grocery store decided to remove the registers and use “AI” cameras instead to automatically detect what people were buying. Turned out it was literally just a bunch of people in India watching the feed and manually adding the totals. Also that humanoid AI robot they demonstrated a few months back. They haven’t figured out the “AI” part yet, so it’s just remotely controlled by some guy in, you guessed it, India.
G
Annoyingly Inaccurate
P
Artificial Idiocy
L
No BI. Barely Intelligent.
S
Probably Incorrect.
I
SRtI for Stolen Reddit thread intelligence.
X
Who the fuck call llm intelligent ?
G
I honestly don’t see the point in it, if you have to double check everything it says, you might as well have just done that research to start with, for example yesterday I googled “when was Iron Maiden’s first album released” and the Google AI overview said “14th of April 1980”, while the Wikipedia article right below it said “11th of April 1980” , now that’s a pretty minor mistake, and “early eighties” would have been a good enough answer for what I wanted, but how did it make that mistake when the correct answer is right there? And it just proves you can’t trust it for more serious stuff
E
>but how did it make that mistake when the correct answer is right there? I am not an AI expert, but from reading between the lines I am suspicious that these language models are being *required* to aggregate from multiple sources when giving an answer so as to avoid charges of plagiarism. And this is what is causing hallucinations.
M
Ohhhhhhh, yeah. That makes sense.
W
Not to rain on your parade there, but that might be a legit case of confusion. Because while the Wikipedia text mentions 11th April, the box right next to it mentions 14th April as the release date. I assume that the 14th April was the start of the tour for the album which was released on the 11th April? It probably is a mistake on the English wikipedia, because in German both parts are correct. And if I ask ChatGPT in German, it does give the correct answer. But overall your point is correct, though your example is possibly an actual error on the Internet pages and not on the side of the AI.
M
Wait, what’s the correct date then?
P
The 13th, clearly
M
🤣🤪
A
Agreed. I kind of hated getting a bold, larger font, simplistic answer, with a rambling, uncanny AI-istic answer, clearly begging for conversation. Now, it seems like we get a response that covers all angles (aggressive, neutral, assertive, introspective, etc.) and then a long list of advertisement-related links that may provide a clearer image and detail.
S
There’s a constant notification that shows up on our military computers prompting us to utilize the AI systems they’ve developed. It talks about providing “efficient, decisive results to the warfighter.” Every time, I can’t help but notice it doesn’t say “accurate”.
S
And there’s so many people who love AI responses. What are the chances that those people just happen to get better responses than you? Oh, if only the calculator app was still available, we could check 🙂
W
Why do you keep using it if its constantly wrong?
M
I only use ai for grammar checking. I’ve had decent luck. My spectrum-y boss is like a living Chicago Elements of Style, and he usually agrees with the ai so I guess that means it’s good enough for my line of work.
R
I once asked it how some ***working*** code it generated functioned (like genuinely wanting an explanation) and it freaked out and went “Omg! You’re right! That totally won’t work!” and changed its code to something that actually ***didn’t work*** because I guess it thought I was trying to catch it lacking and was asking rhetorically/sarcastically or something
K
Needed some help with regex the other day, so I figured I’d try the AI. It gave me a wrong answer and when I pointed that out it told me that what I was asking was impossible to begin with. Which irritated me because not only did it just generate me a bullshit answer for a supposedly unsolvable issue, I knew that it was indeed possible and suggested a way to do it. And wouldn’t you know it, suddenly that idiotic thing gave me a solution that worked. Complete clown show these things.
S
Copilot have been learning Terrence Howard school of mathematics.
T
TerrAI is actually massively powerful.
A
When you bring 2 and 2 together, you get 22.
Z
Ah yes, proof by concatenation.
T
It also needs to be ridiculously sycophantic and shower you with praise for even asking how to add numbers
V
“What’s 2+2?” “3” “No it’s 4” “CORRECT, now you’re thinking like a real mathemagician! *EMOJI* The answer IS indeed 4 — we know this because *absolute bullshit*”
A
Nah bro, it’s actually 12
B
“Great catch! You’re right, in modern versions of math it seems that 2+2=3 has been deprecated and you should now use 2+2=4. Would you like me to provide a detailed breakdown of where numbers come from? Let me know how I can help!”
F
**THERE! ARE! FOUR! LIGHTS!**
W
nah should be trying to convince you it equals 5 since 1984 was in it’s training data and it completely missed the point
R
You have 2 apples and someone gives you an additional 2 apples. Now obviously the government will tax the shit out of that and therefore you end up with 3 apples. In conclusion, 2 + 2 <= 3
U
2+2 = 3 for very small values of 2
D
“How many fingers am I holding up, Winston?” – Copilot
D
Well, if you consider that 1.5 is frequently rounded to 2 and were to use the original figure of 1.5 instead of the rounded number…
R
Best thing Microsoft did was removing support from my old ass laptop. Installed linux back in October and I’m happier than ever.
F
Same here—suddenly my “obsolete” machine feels immortal.
GIPHY App Key not set. Please check settings