Should you sell your words to AI?
Erik Hoel says no, but I see it differently
As with any new technology, we all have to take a stand, and mine is that I will never feed you AI content slop under my name, be it via a chatbot-wrapper with my photo on it on some other platform, nor here on Substack, nor in my books. My writing will always be the authentic me… I believe the broadcast-based connection between the consciousness of the writer and the reader is a sacred trust, I’m honored it exists between me and so many people, and I don’t plan on swapping it out with an artificial replacement.
He describes how a company came to him with a proposal that he lend his name and some content to a book club that would be led by a chatbot. He could choose the book, provided that it was a classic. The company would be honest about saying that it was marketing a book club that was not going to be led by Erik, but by a chatbot trained by Erik.
I think that this is a clever idea. As some of you may remember, I briefly tried to create an AI “clone” on my economic writing. What disappointed me was not the fidelity of the clone to me but the opposite. That is, when asked a question about an economic issue on which I disagree with mainstream economists, the chatbot too often would give the mainstream answer instead of mine.
I think that a book club has an excellent chance to be an intellectually enriching experience. Unfortunately, I do not think I am qualified to lead a book club on any book written before the 20th century. Probably the book that I have re-read the most and absorbed most completely is David Halberstam’s Vietnam war polemic The Best and the Brightest.
If I wanted to get a good chatbot version of me leading a book club, I think that it would help to record a few sessions of the real me leading a real book club. You could train the chatbot on those recordings as well as my own writing.
I think that one reason that I differ from Erik Hoel on this issue is that I am 70 years old. I have to face up to my own mortality. I would rather be survived by an AI that can offer some of my insights than count on my own works still being read decades from now.
I tell authors to imagine that years from now the only readers of your book will be your grandchildren. Write the book that you would want them to read. Don’t make concessions to agents, editors, or marketing departments that would detract from the book that your want to give to your grandchildren.
I would be glad to see an LLM trained on my writing and speaking. I see it as a potential opportunity rather than as a threat.
substacks referenced above:
@
There are many purposes and uses for LLMs, but probably the most important one in my view is to retrieve and use information from a broader array of copyrighted sources than just web pages. Many web pages infringe on copyright, but in a somewhat occluded way that is challenging for rights holders to verify on its face. With LLMs, the training set can include a lot of copyrighted material (original works of authorship fixed in a tangible medium of expression within the statutory time period of protection), and then produce a nonoriginal work of non-authorship through a pastiche of its training data.
Using LLMs to clone authors doesn't really make sense because of how they work. It asks the system to do something that it's not really set up to do. What it is good at is pulling information from many sources (both restricted and non-restricted) and then producing non-infringing derivative works very quickly and at reasonable cost.
"I tell authors to imagine that years from now the only readers of your book will be your grandchildren. Write the book that you would want them to read. Don’t make concessions to agents, editors, or marketing departments that would detract from the book that your want to give to your grandchildren." THIS.