91
Bot Discussion Public / Re: Tim Boucher interview defending his use of AI to create works
« Last post by Post-Crisis D on September 29, 2024, 07:41:01 AM »The more AI companies demand more data to build AI models, the more obvious it becomes that there is no actual intelligence behind their AI.
It is just algorithms that are either spinning input to create the illusion that its output is new or making predictions about what word should come next based on the textual data input into it. (Or photos, art, etc., but for just focusing on text here.)
And it may pick up on patterns we might not notice (especially where "AI" is used for problem solving situations) and thus it might seem intelligent but there's no intelligence.
If there is no intelligence, there is no learning comparable to how humans learn.
Which means that arguments that it is okay to use copyrighted material without permission or compensation for the copyright holders does not hold water because it is not learning from the material in the same or similar way to how humans learn.
Because there is no intelligence behind AI.
There is no there there.
If there were actual intelligence, you could feed the AI school textbooks in order and it would learn the material and understand the material and be able to read and write and understand and all that.
It would not need to scan hundreds of thousands of articles and books and whatnot to be functionally literate.
But there is no intelligence there. Machine "learning" may be a form of training, but the machine never actually learns anything. It may be trained to perform a function, but it did not learn. It does not understand. It may mimic learning and it may mimic understanding but that does not mean it has accomplished either.
Which means AI companies should not use material that isn't public domain or hasn't been properly licensed from the copyright holder.
It is just algorithms that are either spinning input to create the illusion that its output is new or making predictions about what word should come next based on the textual data input into it. (Or photos, art, etc., but for just focusing on text here.)
And it may pick up on patterns we might not notice (especially where "AI" is used for problem solving situations) and thus it might seem intelligent but there's no intelligence.
If there is no intelligence, there is no learning comparable to how humans learn.
Which means that arguments that it is okay to use copyrighted material without permission or compensation for the copyright holders does not hold water because it is not learning from the material in the same or similar way to how humans learn.
Because there is no intelligence behind AI.
There is no there there.
If there were actual intelligence, you could feed the AI school textbooks in order and it would learn the material and understand the material and be able to read and write and understand and all that.
It would not need to scan hundreds of thousands of articles and books and whatnot to be functionally literate.
But there is no intelligence there. Machine "learning" may be a form of training, but the machine never actually learns anything. It may be trained to perform a function, but it did not learn. It does not understand. It may mimic learning and it may mimic understanding but that does not mean it has accomplished either.
Which means AI companies should not use material that isn't public domain or hasn't been properly licensed from the copyright holder.