How would you create your own Generative AI?

[redcrazykid]

I personally would use a program written in PHP to scan an online encyclopedia. That is free to use, regardless if you're using a program instead of using the website in person.

Codes like this:

for($i = 0;$i < strlen($pageparagraph); $i++) {
}

So that it memorizes all the geometry of an elephant's dimensions, to save what it would look like in real life.

So then you can type in "Draw an elephant.". And then it would make a picture of it, according to the infromation provided on the online Encyclopedia such as "www.britannica.com".

 

[dos2unix]

The turn "AI" is a little misleading to me. Artificial intelligence isn't sentient or self aware. It only knows what someone told it ( programmed into it ) To know evertrhing about everything would take a massive server farm. Thousands of computers nuero-linked scanning thousands of websites and documents everyday.

Once you have all that data, then there is the matter of making an intelligent user IA, in this case usually a chatbot that understands certain key words and language nuances.

Think of a 3-d scanner, it needs all the dimensions of the object you want to 3-d print. If it's a square cube that's pretty simple, if it's say.. a human face, it could be literally thousands of data points just for that one face. Now imagine I need all those data points for another face, and another... maybe 10,000 faces. How much data is that?

 

[osprey]

The turn "AI" is a little misleading to me. Artificial intelligence isn't sentient or self aware. It only knows what someone told it ( programmed into it ) To know evertrhing about everything would take a massive server farm. Thousands of computers nuero-linked scanning thousands of websites and documents everyday.

One you have all that data, then there is the matter of making an intelligent user IA, in this case usually a chatbot that understands certain key words and language nuances.

Think of a 3-d scanner, i needs all the dimensions of the object you want to 3-d print. If it's a square cube that's pretty simple, if it's say.. a human face, it could be literally thousands of data points just for that one face. Now imagine I need all those data points for another face, and another... maybe 10,000 faces. How much data is that?

The idea that AI "only knows what someone told it", has been supplanted by "learning from itself". Think for example of the creative AI that creates an original visual image from some manner of its own random or specialised processing, and then is able to identify and use aspects of that image in it's analysis of other images which creates something entirely new. The same sort of process applies to data. AI specialists already describe how unknown and unpredictable outputs occur, though that doesn't mean there aren't limitations of one sort or another, but it does mean that some are unknown which has been a source of some discussion and even concern.

 

[KGIII]

Think for example of the creative AI that creates an original visual image from some manner of its own random or specialised processing, and then is able to identify and use aspects of that image in it's analysis of other images which creates something entirely new.

Then the AI figured out its strategy for a game called Go. It was nothing seen before and nothing programmed into it. This is unlike chess, which has determinate rules. The math for that alone is amazing - and it was done all by the computer training itself.

 

[Condobloke]

This may muddy the waters a touch......but it does involve AI in use...

Arc from The Browser Company

Experience a calmer, more personal internet in this browser designed for you. Let go of the clicks, the clutter, the distractions.

arc.net

 

[redcrazykid]

Oh well when you program every word in the dictionary into a script and use it to respond to text messages they've been calling that AI lately. Sry k

 

[redcrazykid]

Here's a cool picrtures.