Should the presence of AI-generated writing be considered a problem in writing?
AI is increasingly being used to create written content, ranging from editing and rephrasing to generating complete articles. This raises questions about content quality, originality, and the role of the writer.
In this context, should the existence of texts that are partially or entirely generated by AI be considered an issue in writing? When evaluating a piece of writing, should the origin of the text (human-written versus AI-generated) matter, or should evaluation focus primarily on quality and purpose?
From the perspective of writing as a craft and a long-term practice, how does the community view the use of AI in content creation?
2 answers
I do think that language models are a problem, because they don't "do" writing. They mostly emulate the process of writing, chaining up grammatically correct chains of words such that the semantic vector of the output has a similar angle to legitimate responses to similar inputs. It doesn't have any concept "XYZ moved me to create this in hopes that it resonates with readers," because none of those ideas relate to language as such.
Therefore, when you talk about "quality and purpose," AI-generated text doesn't have a real purpose. I compare it to content marketing, where rather than the goal of connecting with inspirations and audiences (even if no more than explaining how to make a thing work), the text exists to keep the audience (the user asking for the work, not that person's audience) "engaged."
And that doesn't even get to research on use of these tools suggesting that they actually slow work down, do a terrible job, and make workers less capable, which presents an entirely different kind of problem.
0 comment threads
AI is essentially just another author. Therefore any time the author is shown or if author reputation could be important to the reader, this must be disclosed. Presenting AI-generated text as your own is plagiarism, just like presenting any other work you didn't personally create as your own.
Whether fiction is generated by AI or not may not matter much. If you're entertained by and like a story, then it doesn't really matter who or what created it. In this case, the author is more of a brand. If you liked one story by a particular author, you might seek out others by the same author. That won't work for AI. Again though, if an author is claimed, then it must be correct. We might in the future see fiction created by AI but guided by a particular human author. That's legitimate as long as it is disclosed.
Non-fiction is a very different matter. Knowing the author is essential in judging the credibility. Since it is nearly impossible to know the original sources an AI used to gather facts from, any non-fiction created by AI is highly suspect. For that reason it is extra important that this be disclosed.
AI has some advantages over humans, such as access to vastly more raw data, and the ability to see connections in that data and to make inferences from them. That's useful. There is nothing wrong with non-fiction that includes insights from AI, as long as this is properly attributed.
In summary, AI is here. We're only at the tip of the iceberg. AI by itself is not a problem, but misrepresenting it as something else is a serious problem.
0 comment threads
1 comment thread