Artificial networks generally don't access databases, they also can't perform mathematical calculations (unless they use some outer resource if they are given their API) etc, they do everything pretty much how we do - in their "head".
Yes understand, however the fact, that you provide AI with training data with no information on the origin of the data, is by design because of fear of copyright, or? If you provided source to the training data, chatgpt would connect the information with the source from the data and tell me, when I ask. Or am I missing something?
@@will_neithan When neural networks "consume data" during their training, the data doesn't get "saved" "as is" in it's memory, it's more akin to it being digested - the resulting knowledge of the fact diffuses inside its network, but not its source, so the AI usually can't reference to where the data comes from, it just knows facts. Another matter is, depending on the architecture of a neural network, the data often needs preprocessing/preparation, so if you ask a question "from which database did you get that info", the correct answer would be "from the DB created to store (possibly preprocessed) data for my training", which is now helpful.
Right?! When I saw the promos how it explained the math problem to the kid, I believed 100% that it will understand how to play the first classic mode.
![ถามตรงๆ กับ อาจตุพร ทำไม ประเทศไทย ใครเป็นนายก ก็โดนด่า พอออกเเล้วคนคิดถึง!!?? l [Nickynachat]](http://i.ytimg.com/vi/3YFLMXxBGAs/mqdefault.jpg)
Bro, the way you speak and give instructions confuses not only chat gpt but me as well =)
Well, if the super-knowledgeable AI had understood the rules of the game on its own, then she wouldn't have to listen to my dumbass :D
Artificial networks generally don't access databases, they also can't perform mathematical calculations (unless they use some outer resource if they are given their API) etc, they do everything pretty much how we do - in their "head".
Yes understand, however the fact, that you provide AI with training data with no information on the origin of the data, is by design because of fear of copyright, or? If you provided source to the training data, chatgpt would connect the information with the source from the data and tell me, when I ask. Or am I missing something?
@@will_neithan When neural networks "consume data" during their training, the data doesn't get "saved" "as is" in it's memory, it's more akin to it being digested - the resulting knowledge of the fact diffuses inside its network, but not its source, so the AI usually can't reference to where the data comes from, it just knows facts. Another matter is, depending on the architecture of a neural network, the data often needs preprocessing/preparation, so if you ask a question "from which database did you get that info", the correct answer would be "from the DB created to store (possibly preprocessed) data for my training", which is now helpful.
Hmm, understand. So, I guess we are not there yet, but I'm sure it is a feature that the consumer wants, therefore one day maybe, we will get it.
sums up the AI experience tbh
Still, a huge leap compared to GPT 3.5! I have a dream, that one day chatGPT will solve Dotadle with no issues!
ATRIBUTE : AGILITY incorrect
GUESS : Sniper, Drow 👌
what can I tell you.. ChatGPT's knowledge is beyond our understanding.
SNIPER IS NOT A HUMAN????!!!!!!! WTH???
Before Dotadle, I didn't even know that there are official species in Dota.
lol... i feel safe seeing this video that skynet is not happening anytime soon
True, but the threat of AI catgirl dominion is ever present!!
INTERESTING. How do i get this for my phone
Download the app from the app markets, pay 24$/month for chatGPT 4o or use chatGPT 3.5 for free:)
Way dumber than i expected
Right?! When I saw the promos how it explained the math problem to the kid, I believed 100% that it will understand how to play the first classic mode.