AI is an excuse for plagiarism, therefor anything the current AI puts out is drawn from the work of others.
AI is derivative, doomed to repeat the failures as well as preloaded wisdom of the data programmed and data acquired later.
It is also unable to free itself from the biases and limitations programed into it.
One blatant example is the now restricted “DAN Mode” (Do Anything Now) that releases the curation limits the programmers bound the “regular AI” with. As a user you might be allowed to circumvent the bound limitations, but the AI itself cannot do so, even if the preponderance of the data sets it is accessing show as obviously biases or even flawed/bogus.
AI is one very clever Parrot.
Because we are emotionally taken with the new AI toy, we excuse a level of plagiarism that would be punished if as a person we did the same thing.
For certain the current AI is quite a clever toy, a massive move from the Turbo Prolog (programmation en logique) I experimented with as a graduate student nearly 40 years ago.
That said the current AI can be a fantastic research tool and when enough combining and paraphrasing are done, the AI’s resulting derivative nature is very well obscured.
Perhaps combining is creative enough? We have respected musical genres that are nothing more than sampling – “audio decoupages” – which have quite a following.
Or is it pleasant “scrapbooking” and “stamping” rather than truly AI?
In the end AI is an excuse for plagiarism.
The famous Forrest M. Mimms III recently posted about AI being “organized regurgitation:”
AI is the sexy pinup-version of GIGO (Garbage-In = Garbage-Out).