Love/Hate AI?

I detest but also somehow understand the desire for AI developers to mine the gold that is generative AI. I’m a pragmatist. And for business and some domestic administrative purposes, surely it has its place?

Nevertheless it leaves creatives battling to convince over-eager AI apps not to generate or draft anything. Most AI chat apps have a persistent set of rules that can be user-customised to (or at least intended to) exclude AI responses that include prose generation. I can tell you that it’s very difficult to make those rules stick — you may have noticed that the chat versions of AI always include a disclaimer that states something like ‘AI can make mistakes, including about people, so double-check it’. It would be more accurate to say that AI will make mistakes.

I can’t tell you how many times I’ve had to go back to Gemini or ChatGPT and reiterate the instruction — DO NOT DRAFT OR REVISE MY TEXT. Then, often as not, I’ll spend the next ten minutes arguing with the AI when it insists that the subject text that it generated was actually my own.

I DO NOT recommend attempting to use vanilla AI chat apps, as they currently exist, to support your creative writing. Even if they did not ignore their instructions they will happily tell you that your prose is the best slice of bread since Ishiguro’s loaf. My advice? Maybe don’t believe them.

Using AI as a fact-checker is useful – it performs the same function as a ‘Google search’, but is additionally capable of combining results into a synthesis. I would also consider that AI may effectively be employed as a beta-reader (’editorial analysis tool’) – assuming that the AI does not overstep its bounds. The value of the outcome is, however, highly dependent upon the technical quality of the assessment. Although this also true of human beings, people are also able to tell you how your prose made them feel. If you ask your mother, you might well get a different answer than from your college professor!

In summary, like any tool we use in our modern lives, leveraging the strengths of Large Language Models (speed, pattern recognition, tirelessness) should not be prohibited out of dogma. Claiming that you exclude AI from your life and work is, today, simply delusional. If that is your goal, you might as well promise to eschew the World Wide Web altogether.

However, generative AI will NEVER help you to express to a human reader what it means to live a life. Never. It’s not a matter of future technological progress — rather, it’s a physical impossibility. Human literature is at least about expressing the human condition, and at its best encasing it in amber. AI will NEVER be able to achieve that, no matter how clever it becomes.

I’ve buried people. I’ve scattered my father’s ashes. Do you think AI could ever have any idea how that feels?