>AI is good at math it won IMO g-ACK!
>>107685186>1 green close to 2 blue = mostly blue5 blue apples>green apples = any apple with blue 7 green>blue is the new red 13 red
Can it render an apple that actually looks real?
>>107685344Can you? >>>/i/ >>>/ic/ :^)
>>1076852113 green3 blue10 red
if it can solve erdos problems it's good enough
>>107685566I can take out an apple and photograph it. It'll look as real as the real apple. So what about the AI?
>>107685704No, it doesn't. No matter what camera you use, it won't look exactly like a real apple.
>>107685186Don't sweat the details. That's what humans are for. AI is the big picture guy.
>>107685739>That's what humans are for.Not for free it isn't.
>>107685735When's the last time you photographed an apple and looked at the photo?
>>107685186the model that won IMO gold and the one that can generate pics are different and I'm pretty sure that the first is not even fully released to the publicwhen you tell it to generate a pic, it just routers your request to the image model
>>107685796When's the last time you took a photo of yourself and looked at the mirror?
>>107686083It took you half an hour to come with that?
>>107686105I was afk. Now what's your response?
>>107686134I'm not bumping your thread
It's always just guessing, it gets the "gist" of things, but it's imprecise. A simple old calculator will never miss that 2+2=4, not in a million tries, it simply cannot do anything differenother t than getting it right. An LLM can get 2+2=4 right 99% of the times, which may be close enough for an idiot, but this 1% is absolutely crucial and makes the whole thing useless for calculations, seems like a waste of computational power to even try.
