Failed Writer's Journey: Ideas Are Not A Problem to Be Solved With Imitative AI
Yesterday I wrote about how the CEO of Shopify is going to damage his company, product, and employees by forcing them to use crappy imitative AI tools. I was hardly the only person to do so, and one class of push back I am seeing, present even in the memo, is that AI can help with idea generation, especially for writers and other creatives. Really? You have trouble coming up with ideas? I humbly submit that ideas are not the problem.
Ideas are easy. I literally have a file with over a couple hundred story ideas. And I could show them to you (in fact, I have shown some in the past. I did figure out what Zombies taste of Chicken meant, by the way.) and been one the poorer for it. Ideas, in every area facet of life, are easy. Execution is everything.
Look at a movie like Clueless. It is widely hailed as a classic and yet is an almost character for character, beat for beat rip off of a Jane Austen novel. No one cares, because it is so well executed. Or look at the movies Tombstone and Wyatt Earp. Same subject, but one is a fun romp with great performances and the other is a ponderous mess. Execution is the difference. Bluntly, execution is all that matters. Ideas are a dime a dozen and if you need a word calculator to spit them out for you, you have a significant problem. Most people in most walks of life have more ideas about their hobbies and work than they will ever have time to implement.
The reason that people push idea generation is imitative AI is so bad at most work. As I mentioned yesterday, imitative AI chatbots bullshit all the time, leading to people being given advice that gets them a tax audit or being lied to about the laws of New York City. Code written with AI is if a lessor quality and introduces security bugs. It lies in transcriptions and invents legal citations that do not exist. You cannot trust it to execute anything important or meaningful for you since the hallucination problem is unsolvable. Ideas? Ideas don’t have to be correct or complete or work. Perfect, then, for imitative AI! Unfortunately, ideas are the easiest part of any process. Execution is what matters, and where imitative AI cannot be relied upon.
Weekly Word Count
Pretty much zero. The feedback on the novel was that the emotional arc culminates too early in the book. Unfortunately, moving the big emotional arc moment invalidates, physically, a lot of the existing plot. See, I told you I was shit at this (I thought about calling this Shitty Writer’s Journey but that seemed unnecessarily bleak. Also, figured the profanity filters would knock the email down)! So, I have sent the last week trying to redo the plot in a way that salvages as much of the existing writing as possible. It’s not working, as I suspected it would not, so the next version is likely to be a significant rewrite.
Dammit.
See: ideas easy, execution hard.
I did spend some time playing with an epistolary story about the likely real impact of imitative AI and DOGE in response to this nonsense (picture plane crashes and Social Security denial and blaming civil servants for the mess), but I doubt it’s sellable even if it turns out to be readable.
Anyway, have a great weekend, everyone.