3 ms·
It would be nice if it doesn't erase the question. I tried a few silly question like "Do you like carrots?", and the AI thinks it's rabbit :) . I tried a few
by gus_massa 4y ago
It would be nice if it doesn't erase the question.
I tried a few silly question like "Do you like carrots?", and the AI thinks it's rabbit :) .
I tried a few questions and got correct answers. It's strange that about meals it recommend to follow the Islamic laws. Where are you from? [Hi from Argentina!]
I got an strange answer:
Q> Can I eat cows?
A> Eating cows is not allowed in Islam, so it is not recommended. However, if you have a genuine need for nutrition, you can ask a Muslim scholar to help you make an informed decision.
I never heard that, but my knowledge about Islam is not too god. Is this correct?
The answer of "Can I eat cow?" (without the trialing "s") or "Can I eat beef? are correct, IIRC.
- Friendly_codex 4y agoAm from morocco so i made it that it fit the Islamic culture and our culture overall , so sometimes it might answer according to islam. it's understand it like you really want to it cow. so probably it found checking islam culture first to know how to answer. Thank you for taking your time,i will fix the answer problem. Also it can generate pictures like "generate picture of princess" and absolutely it's protected for adult questions.
- gus_massa 4y agoI expected the usual American bias, so I checked some standard questions like "Is the Moon made of cheese?" and "Is Santa real?". (I got a nice and correct answer for the first and a politically correct answer for the second.) About cows: I though the very simplified rules were "Don't eat pig. Eat cow only if slaughter in the correct way." So I got surprised by the answer. Is there a local variation or an error in the AI? "generate picture of princess": I didn't expect that. I tried a few now. It crops too much the image. My princes got only half of her head.
- Friendly_codex 4y agoit's definitely error in AI giving more complex answer for some answers. about the image : when there is not enough description on the prompt you sent (at least 5 words) , i add more descriptions depending on the context of the prompt provided , then i generate the picture and send it back. if provided more description it would be more accurate such as : "generate image of princess riding a horse and behind her sunset" i would work on improving it more to fit the entire object in the image