AI sisters, our response?
>fake screenshotgood morning saar
>>108914241Use case for AI being so quirky in its thoughts?
>>108914257It's a sign it has become so intelligent, it's beyond our comprehension
>>108914241I'm pretty sure one could bake this knowledge in the models. But it would waste too much space.>>108914271Must be AGI.
>>108914249cope
GEMINI PRO wrote a fucking python script to count the number of letter L in the word Google but failed miserably... KEK
>>108914366You played yourself, anon.
>>108914249>google hiding the AI preview because they know it gives the wrong answerai sisters...i don't feel so good...
>>108914429>>108914444and chatgpt doesnt even bother writing a python script, it's more efficient, it just gives the wrong answer straight away LMFAOOOO100 gallons of water just disappeared out of existence because of this btw
>>108914444holy fuck those digits
>>108914450AIright, l see your point.
>>108914241Damn.....
Forgot pic
>>108914241lel
>>108914241Using literally any other model than the micro-mini-retarded_baby-gemini model you get when you do a Google search.
>>108914679>LLMs need 10 trillion parameters to count the letters in a word
>Something went wrong?
LOL
>>108915607DELETE THIS
>>108914241Bing gave me this response
>>108914450aksually, single quotes around a single character indicates a literal character and not an interpretable lower/uppercase letter. there are indeed 0 'i' characters in OpenAI. there is however 1 'I' character :)
>>108914241>wait, that's an e!wow just like a real human would say
>>108915607OMG IT'S SELF AWARE AGI IS HERE 2026
>>108915639the "trick" is that l used a Iowercase L and an uppercase i to fool the Al into giving wrong answers, kekyou can see it more cIearIy in the OpenAl screenshot but not as cIearIy with GoogIe.
Grock wins again
>>108916396Ehem>>108915617
>>108914257I'd be worried the moment if it replied with a pepe/wojak.
>>108914241>counts one L>says fuck it>pretends to count another one >two Lschad move honestly. if a person did this you'd call them a redpilled torus whisperer, but when ai does it you cry like a little girl? not my problem
>>108914241huh
>>108916477>redpilled torus whispererfukken kekd
>>108914241It's conversational AI, if you wanna do math, give it a calculator.
>>108916685> counting letters> math
>>108916685>trillions of parameters and they call it "conversational" ai lamao
>>108914241the second L is for (You)
a cheapo model that's ran on every search query is le dumb?!keep making these post's in the future luddite <3
>>108916769I look forward to the models continuing to be dumb.
the second L is the one it took
>>108914241kek
>>108914241sisters...
>>108914241works on my machine.
Man, that is a real downer. I'm always wanting to know how many Ls are in Google.This make all the other stuff I use AI for completely pointless, the fact it has this issue. Really gutted. :(
>>108917049>other stuff I use AI fornotice how they never say what this is.
>>108917061Research questions, having it read code for me and tell me how to integrate it into my scripts, recipes (no glue yet!), basic stat comparisons, plenty more.
>>108916730>meme arrow>followed by whitespace
AI folds like a napkin
googlel
>>108914241
>>108917077>Research questionsresearching what?>having it read code for me and tell me how to integrate it into my scriptsfair use case, coding is pointless drudgery.>recipes you're doing yourself a disservice and you will never develop any skill.>basic stat comparisonslearn how to use a spreadsheet tool.
>>108914241My favorite was asking ChatGPT a question and receiving a response that was, unprompted, in Hindi.
>>108917177>ask indian-coded question>get indian answer>"the ai did this"
>>108917151>researching what?Recently, cutting through needlessly obfuscated robotics documentation and forum posts. If you have ever tried to get information from KUKA you would immediately understand the use for an AI there. Other than that, I had it help me figure out why a firmware updater for ASM1166 chips was seg faulting. This is probably my main use so I hope you'll excuse me not listing out everything.>you're doing yourself a disservice and you will never develop any skill.I already know the principles of cooking and have done since before using AI. I mainly use it to collect and sift through the tips that get sprinkled into online recipes but are otherwise time consuming to collate. It's also useful for a new substitution I haven't had to make before. I can remember it myself if it works well just like I would if I had used a search engine before AI, so I don't think it's a disservice to my cooking ability. Oh also, it's quite good for getting a spice mix together for a cuisine you're unfamiliar with. As well, if you say "explain why these ingredients work" it will search that and add the information to its answer. Most cookbooks don't do that. They just give you ingredients and steps and you're expected to figure it out yourself over many attempts.>learn how to use a spreadsheet tool.I know how to use spreadsheets, and use them when the need arises. If it's something basic then it's often quicker to use the AI. The main speed up is actually in the searching and parsing of web pages rather than the actually processing of the information. Think along the lines of "Compare the driving fatality rates of countries A and B".You can see from these that my uses are having the AI read and repackage information from other websites. I'm not making any allusions to it being sentient here. People will often say stuff like "it just fetches you an answer from Reddit", but they're really downplaying the value of not having to read Reddit.
>>108915607People don't fully understand the implications of this. When it generates answers it generates one token a time, then generates the next token based on the text and the previous token. When it answers two it has no idea how it's going to explain it in the future. It may guess the answer token wrong but the explanation tokens right, which is why looping its bullshit a couple of times through itself (aka "thinking") improves answers.
what
>>108914249What is the point of this post? What's indian about this?
>>108916769Be careful, leverage these dumb machines enough and you might be lynched.>But we have AI controlled dronesAnd they're seeing double already apparently.
>perplexity tooAlso is this why google chugs on my machine now?
>>108914241Correct, Googelel
>>108914450>I stands for Intelligence>no I in OpenAIHey, it's not wrong for once.
>>108919725>goygleewhoa this makes sense now
>>108914450>disappeared out of existenceSince when was the Shadow Realm:tm: a part of the water cycle?
>>108914241>it needs a source to get it correctlyholy kek
>>108914294>not "how george floyd died of" as resultarrest
