What the fuck is going on with qwen? Also, Grok is really that good? I haven't even tried Grok yet.
>>107999114what temperature are you using on all these?
>>107999131https://research.aimultiple.com/ai-hallucination/
>>107999114lol I was using gemini earlier and it was awful. straight up toddler play house level scenarios. mind, I was trying to get it to analyze images and generate comparative data sets.
Just wait for AI to get smarter then it won't make shit up anymore, right?
>>107999369Honestly, even the shittiest LLM on this list hallucenates less than any human being I've ever met.
>>107999164It's good at generating and editing images but can really suck at parsing out particular qualities of an image.
>>108000225I feel the exact same way whenever I try to discuss art with real people. In fact, the AI tends to be vastly superior to most humans at what you just said, not inferior.Maybe you just don't understand how fucking retarded and brainless most people are. A lot of people can't even identify their own stupid face in a mirror, or visualize an object in their minds. Just be glad you never took psychology and had to learn how fucking dumb homo sapiens are on average. Everyone putting down AI needs to get real and look around at their stupid dumbass friends and family and realize how much more retarded most people are than even the dumbest agentic AI model.Humanity is so fucked it's unreal.
>>107999114>gpt hallucinating more with each new versionuh oh saars...
>>107999114>oyvey our Porn Generator is based and redpilled, come and use it goyimzzzzzzzzzzzzzzzzzzzzzz
>>107999114Why you fa/g/s calling LLMs 'AI' and LYING 'hallucinations'? What next pdf files 'caress' children?
>>107999369AI has 0 intelligence tho
>>108001458>pdf filesGo back to tik tok. How did you even find yourself here?
>>107999114>hallucinations are getting worse over timeHa.
>>108001463What? This is your safe space or something?Go to reddit and cry there.
>>108001488Embarrassing response. Lurk 5 moar years.
>>108001492Standard ledditor response. Embarrassing really.
>>108001462Look at the graph, its clearly going up.
>>108001532Yeah but TIME axis might be in MILLIONS of years.From kid perspective LLMs are great tool for learning and quick.From professional perspective they are useless.
Grok is fairly good at fact finding missions that you give it. Its ability to search the net, not be biased and respond to questions appropriately.
>>108001580very well said, 5 Xdollars have been transfered to your account :)
>>108001580>not be biased
>>108001613>fact finding>your opinionYou realize opinions arent facts right?
>>108001570Dunno, you're just some random but people like Elon Musk who are very smart are saying it'll be sooner rather than later.
>>108001649Elon Musk is rich not smart.He has yet to do anything to prove himself to be smart. In fact he only did things to show he is dumber than average engineer
>>108001661Hm, who to trust.. a nobody on an imageboard or successful people, tough one.
>>108001700>Money makes rightYes, sign up for musk's mars program you are going to live in a Martian colony two years ago.How about you don't trust me or musk and actually look into what that moron said?
>>108001709He says AI will be super smart, just like the graph. Soon. And so do people pumping billions into it, so it seems like that's the truth.
>>108001719He also said he can solve the traffic problem in 5 years. Why should I trust his words if so far nothing he predicted came true?
>>108000836The thing peaked right before the introduction of 5.0. It's all grim now.
>>108001719He has said too many shit, just read his comments from a decade ago. You'll realise how much of a genius he actually is
>>108001719wanna buy some tulips?
>>107999114Are you using AI to check how much AIs hallucinate?
>>108001458Wait until they realize that the AI being right about anything at all is literally a fluke because it's designed, first, to be capable of lying to you convincingly.
