AI and Writing

I have always been a firm believer in education. To keep your brain active and working smooth, learning something new creates new neural pathways, and keeps your brain healthy.

My big research this year is AI and writing. There is a disparity between AI use as an author, and a publisher, and the twain shall never meet. At least not in this decade. Anyone that is writing today is very familiar with AI and the insidious nature creeping into the writing profession growing like a fungus and attaching itself to anything it can.

Can you use AI for research, and grammar and spelling? Can it be used to translate an English novel to another language? If this is done, is the ent translated copy considered one hundred percent AI? How much is reliable information when AI answers a question? With the recent announcement of Anthropic putting in undetectable watermarks in anything AI, is there really any use to even working with AI at all? I do know people have quit their subscriptions to Claud, Gemini, DeepSeek, and other LLM's in droves, less it taint their work. Once a researchers best friend has now turned into the enemy. 

Professional writing organizations have explicit rules concerning AI and do not allow AI in any shape or form. The Horror Writer's Association, The American Author's Guild, The Writer's Guild of Canada are all repeating the same message: no AI writing allowed. And I understand that perspective. With many novels now completely written by AI and being published by one of the Big Five, publishers and agents are making authors disclose if AI has been used and where. Even with the disclosure, your novel will get run through an AI detector and it can kill your book deal. 

There is a recent case going on right now where a first time novel was in a massive bidding war with multiple agents and several publishers. The deal was struck for 2.5 million, only to be shelved days later stating the novel was all AI. Except it wasn't. And finally, an author is suing the publishers and agents. The verdict is still out.

Personally, I have fed stories I had written in the early nineties through an AI detector and discovered my work was AI derived, even though they were written well in advance of AI. Others have had the exact same result. So that questions AI fallibilities. How accurate can it possibly be when work written forty years ago is considered AI?

This is the heart of the question with AI and writing and research. How much can we trust the LLM, if we know it is completely inaccurate?

The writing industry is a tough nut to crack, and AI has just introduced another barrier. Sure, the publishers can use it for their business needs and I wonder how long publishing and agent jobs will become redundant with more and more usage of AI. Although I don't think very many of them are thinking this way.
Then there is the announcement from Barnes and Nobel stating they will allow AI written novels on their shelves. It is a strange time to be a writer.