I’ve been writing about AI for a while, and there are more AI topics I could cover. However, from my perspective, I feel like I’ve done enough and am starting to repeat myself. I also feel that my ideas on the topic have gelled into a usable form now, at least in terms of my own work, so that thinking is what I’m going to share today. Then I’m done with AI discussions until something new occurs to me. Back to more obviously gaming-related stuff on a Monday.

So where have I got to?

Well, I think the following statements are true about our current crop of “creative” LLM/AI tools, and here are my thoughts on each point:

Point 1: It’s here to stay whether I like it or not.

Therefore, I should have a considered opinion about it, even if it ends up being to ignore it. 

Point 2: It’s everywhere.

I don’t have to use it.

Point 3: Raw AI output can be competent, but is generally unremarkable. It needs (often extensive) additional work from a human to bring it to a high-quality professional standard. 

This may change, and I may look again if it does. Other people may also have different standards than I do for what constitutes an acceptable professional quality, and that’s their choice. I’m happy with my standards where they are.  

Point 4: Once the prompts have been refined (which may take a long time), AI output is often suggested to me as a solid basis to work over. 

Most people I know who use AI use it like this. In a way, it’s like having a very fast junior to do the foundational work. Personally, I want the best foundations possible for a project as everything is built on top of them. If I know in advance that I’m going to have to overwrite their poor work anyway, then why not do it myself in the first place? I feel like me doing it is going to be a faster and better quality end result when you factor in all the remedial work rather than looking only at how fast AI does the initial output. As the saying goes, “If you don’t have time to do it right, when will you have time to do it over?”

Point 5: AI can’t copy what it hasn’t previously stolen, so it cannot innovate. 

Average isn’t good enough. And, if I have to bring the innovation myself, why don’t I bake it in from the start? Then there’s the ethical question of using a tool that’s based on such a blatant moral vacuum. 

Point 6: AI can mimic my voice, but it isn’t me. 

I create stories, experiences, and new worlds to play in. I do things that interest me, that make me smile, and which aren’t the same as dozens I’ve seen before unless there’s a reason for them to be, in which case I’ll be doing something else different. Their faults and excitements are products of my experience and my character, what amuses me, or terrifies me. Working on my ideas is far more interesting than working on someone else’s. So why would I want to hand off a chunk of that fascinating process to some software? Baffles me. “Here, have a big slice of my fun. No, no, I’d rather do the boring tidying up your mess at the end. You do the exciting bit.”  Now there’s an idea that does not compute. 

Overall, AI isn’t for me. Not in its current form, not with my current workflow, not until it’s offering a benefit that comes without so many downsides. 

I know people who use it occasionally, and a few who lean more heavily upon it, and that works for them. I’m not evangelical about people not using it. I can see how it works in some use cases. You choose what works for you. 

Critical for my own decision is the fact that I’m not required to use it by my boss. I work freelance, and if someone hires me, they’re presumably after my 50+ years of writing and game design experience rather than any nascent prompt-writing skills. If your boss decides that your job now needs to integrate AI, then you don’t have the same choice. 

Other than that, the most important aspects of this question for me are around skill and emotion. Delegating the creative part of the process means dulling my skill and missing out on the fun. Neither feels like a benefit.

Asking someone or something else to do the work means that I don’t learn and my skills atrophy. As true for AI as it is for delegating anything else. I learn to delegate, not to do the thing. And here, doing the thing is infinitely more fun than learning to delegate. I love the endless challenge of game design in all its many permutations. I think about it all the time. As far as I’m concerned, AI can’t have that fun. It’s all mine. 

The main potential upside I see of using AI is that it arguably does some of it faster, which is an economic argument rather than a creative one. I say “arguably” because by the time I’ve gone over its output and brought it up to the standard I’m happy with, I’m not sure that it saves me any time. I’ve worked with many less experienced people over the years, and polishing the work of others is both less interesting and often no faster than starting from a clean sheet. 

If it’s a human junior, then part of the process is mentoring and teaching them where they could improve and how to do so. I’m happy to do that as it’s a rewarding process. Fixing AI’s failings is not. 

In the end, I’m unconvinced by the time-saving claims, and have no interest in letting my skills die or foregoing the joys of creating my own stories. So, thank you very much Mr Billionaire, but I won’t be using your AI.