70 Comments

I can't say I'm sold. A lot of the use cases you describe are things I already have my own, human systems for, or are muscles I would like to continue exercising in my own brain. At this point given everything known about AI today, I wouldn't upload my own work to one for analysis, not even a small section. This topic is a little difficult because, after all, it's developing so fast that the facts today may not be the facts tomorrow or next week and almost certainly not next year - so I don't say "never". So far, every time I tried to ask AI generators to do something for me, the result was either useless or I could have better spent the time that I took to come up with prompts on just doing the actual work. Now, if an AI can clean my house for me or order my groceries when we're running low, that will be a very different proposition that I will embrace eagerly!

Expand full comment
author

I’m not entirely sold either but I continue to dabble and experiment because the technology is evolving so fast and I’m curious about the mechanics. I don’t ever end up using the generated summaries but they give me a starting point to edit. I think I just hate the process of summarizing and writing synopsis so much that I’ll look for anyway to make it feel less painful!

Expand full comment
Apr 7Liked by Ben Wakeman

I've been waiting ages for a dusting robot.

Expand full comment

I listened to a podcast recently in which the interviewee, an author, said he submitted his 350 page manuscript to Claude 3 (presumably the paid version) and within 20 seconds received a detailed editorial analysis that was virtually as good as the one he received from a professional editor.

Expand full comment
Apr 7Liked by Ben Wakeman

You've tempted me to tiptoe into this Ben, if only for experimentation. But it still makes me feel queasy. Say I use AI, as you did, to assist with forward progress from a stuck point in my novel. And later I use the AI to write the novel summary I put in my query letter to agents. Say I use it to write the letter itself. When an agent's submission guidelines say, "We will not accept AI generated work" where does that leave me? I have a very large (hyperactive) conscience. I CAN'T LIE. And I think what freaks me out about using AI in any part of my creative practice is it feels like lying. Like if an agent asked me point blank if I'd used AI for any part of my writing process - I'd start sweating. And I'd ultimately say, yes. And then I'm out. Any advice for retraining myself NOT to see AI assistance as a form of dishonesty when it comes to selling the work it helped create?? I want my life to be easier to. But ... queasy... ☹️

Expand full comment
author

Now you’re just making me feeling like a godless sociopath! Let me not lead you into temptation. In all seriousness, only do what you feel comfortable with. The ways I use the technology don’t feel like cheating to me because I am still the one doing the creating. As I said in the essay, the writing is the joy part for me, so it would be counterintuitive to give that to anyone else much less a machine.

Expand full comment
Apr 8Liked by Ben Wakeman

You're fine!!! 😇 I'm just a wet blanket, goody two shoes. Not proud. Just owning it. 😂 Honestly, I'd be so much farther along in my career if I could just learn how to scheme properly!!!! Part of the reason I'm so tired of the grind is because I take the long way to EVERYTHING. I do every step of every process myself either because A) I can't afford help or B) I don't trust the "help" to have my best interests at heart. But I would like to learn a few shortcuts. Lord knows, the gatekeepers I'm trying to impress are probably using them to navigate their slush piles. Why shouldn't I even that playing field!?? 😈Now you've done it, Ben. You've corrupted my hard drive.

Expand full comment
Apr 7Liked by Ben Wakeman

Brilliant read, Ben. I have been using ChatGPT daily for my work job for over a year to streamline various tasks on the research side and also to generative creative ideas on the teaching side, and it's become invaluable.

I hadn't consider using it as an editorial assistant, but this makes complete sense. I guess my only slight concern here is that by uploading the entire story so far, is it getting mined and fed into the bowels of the LLM? I haven't read up on any of that. It's a drop in the ocean of course, but it does make me stop to wonder.

Also: "But that was already happening - see the Marvel universe." hehehe. 100% agree.

Expand full comment
author

Thanks Nathan. I had a hunch you would be a pro user given the nature of your day job. I’m sure you could teach me a lot. As for the concern about AI learning from my writing, I’m not overly worried. The way these models work, they don’t pull wholesale plot lines, or characters from training data. I may feel differently a year from now, if some punk AI generates harmony house. But then it’s a copyright case and I have not just a copyright, but a lot of witnesses here on Substack. 😁

Expand full comment

Nah I reckon you'd be teaching me. I'm a regular user, but I wouldn't call myself pro.

Lol, so true re witnesses.

I think the AI voice over scene is more concerning because of how quickly replica voices seem to be able to be trained.

Expand full comment
Apr 7Liked by Ben Wakeman

Wow, Ben, your piece on AI really resonated with me. It’s cool how you went from skepticism to seeing AI as a useful tool, almost like finding a guide in uncharted territory. The way you describe AI as more of a creative partner than just a tech gadget makes the whole concept feel more approachable and personal. Reading about how AI helped unblock your novel’s progress was especially inspiring. It's a fresh perspective, blending the practical with the imaginative, and reminding us that AI can be a companion in our creative journeys, helping clear the path for our ideas.

Ok. So, I was full on in the Taylor Swift camp. Utterly distrustful and suspicious, bordering on dismissive. But you made an excellent case, and the way you used it for Harmony House is really, very clever. Anyway, long story short, I made a ChatGPT account and practiced a bit—I even gave it your post (I hope that’s ok?!) and it wrote the comment above for me…! 😨 I need to mess about with it a lot more, but I’m really grateful for the encouragement to use it to save time. I could actually really use an assistant to do boring but time consuming things, so, I’m open to giving this a proper go! Thank you for making me step out of the dark ages. I absolutely love having my mind changed :)

Expand full comment
author

Ha, touché! You must have given it a pretty good prompt to generate your comment. I'm glad I've made you curious. As I mentioned to Kenneth in his comment above, the more kind, intelligent, and thoughtful people who learn to use this technology, the better our outcomes will be, because it's not slowing down. The big capitalist machine will look to make it's pound of flesh.

Expand full comment
Apr 8Liked by Ben Wakeman

It required some nudging, the first comment was so robotic so I said ‘please could you make this more casual?’ and then it sounded like I was 17, so I said ‘please make this sound as though Mary Oliver wrote this’ and it made it so flowery & OTT, so I said ‘could you please make it 90% less Mary Oliver?’ and this was the result! I also am recognizing that without coaching it seems to have a particular voice of its own, which is interesting as I can see where I’ve seen it in other places where I’d assumed I was reading a ‘real’ voice.

I’m going to have it come up with recipes for me today, and to tell me how to order my day so I can get all the things done. If this thing can make me even 5% less scattered as a human, I am in. Thank you, Ben.

Expand full comment
author

You’re right, each of the LLMs has its own default voice and particular set of skills out of the box. What you did to get the result you wanted is essentially the way prompt engineering works- trial and error.

Expand full comment

I learn so much from you, Ben. Ever so grateful.

Expand full comment
Apr 5Liked by Ben Wakeman

You are right about Swift but not about airfryers. I promise you won't be disappointed.

As for AI, that prompt you gave is brilliant! I have used chatgpt in my day job where I have to write copy for products and ecommerce newsletters (it's my own business). Writing these is like the field plowing of desk jobs. Gotta do it, but don't want to. I have no problem asking it for help with mundane writing tasks. I also like to ask it questions like deciding which writing project to tackle first. Or bullet points to explain a philosophical principle.

I have also used ai to write my own bio, a synopsis and so on, because this is the stuff I'm garbage at. When I was a music journalist, I never ever wrote my own headlines. As a novelist, I am crap at blurbs. I can write 80000 words but nothing under 25. I agree with you - do what you're good at and get a gardener if you don't like watering plants and mowing.

Lastly, do you actually tip it?

Expand full comment
author

Hi Jo, it sounds like you’re make the most of the tools, that’s great! Regarding the tipping thing it does tend to make the LLM work a little harder. If you weight your prompt with consequences from real life situations like getting fired or getting a promotion, the AI factors these in and might go the extra mile to generate a better response.

Expand full comment
Apr 6Liked by Ben Wakeman

I see. Do you actually pay it?

Expand full comment
author

No, it’s just a trick of language. Stories are catnip for LLMs and they’re wired to “think” like humans so they can be motivated by things that motivate us.

Expand full comment

Oh my. That's great Intel.

Expand full comment

I thought you might enjoy this ... My problem with Ai is simply honesty... New tech is open to truth abuse, particularly in my industry ...https://open.substack.com/pub/leighgbanks/p/8-billion-reasons-ai-is-the-art-of?r=drr6n&utm_campaign=post&utm_medium=web

Expand full comment

Thank. I'm going to check it

Expand full comment
Apr 6Liked by Ben Wakeman

Going to save this one and give it a good think and test run for some experiments - thanks Ben!!

Expand full comment
author

I hope some of it will be useful to you.

Expand full comment

Thank you for this summary and insight! I have been completely resistant to AI but can definitely see potential for its use in my full time job. Your article gave me more understanding and also sparked my curiosity to learn more.This also makes me wonder if an AI finance assistant may be available to manage my bills and spending! Ha! Now that would be useful.

Expand full comment
author

Thanks Sandy, I’m glad it was helpful. The tools are really great at analysis, but right now they don’t have long-term memory or direct access to your systems to do something like help with accounting or finance. You can bet that QuickBooks and most major banks are working with AI to do exactly what you’re asking for.

Expand full comment
Apr 6Liked by Ben Wakeman

AI is also excellent for planning family meals for the week, and associated dietary needs, and grocery list. 😁

Expand full comment

I've gotten some great perspective from this piece.

And I have a few thoughts. Digital recording has made rap, as we now know it, possible. Some think that's great. I don't. I respect musicians whose instrument is, well, their instrument. I'm not sure that the software will ever exist that has an instinct concerning when to play 'behind the beat', when to vary tempo, and so forth. I've tried recording with the metronome, but it doesn't suit me. Rap is what I call 'mechanical'. Nuance is difficult to pull off, when software is calling the shots. But what I've said here more supports than contradicts what you've said, Ben.

You said "There are concerns around the potential stasis and even degradation of intellectual creations whether it’s art, photography, film, music, or writing. After all, if everyone is just using AI to create based on what’s been created in the past, we will end up in an echo chamber of bad reproductions." I could write a book about this. (Maybe I will, right after all the OTHER stuff that I'm going to write!) I spent a career as an architectural woodworker, and am very knowledgeable concerning production techniques since before the beginning of the industrial revolution. The changes have been somewhat reminiscent of AI, but for mechanical systems. When everything was done by hand, the craftsman's 'fingerprints' showed. Individuality could easily show itself. But as streamlined production took hold, individuality fell by the wayside. There's no time for individual flourishes, when a piece is moving down the assembly line. Machines cut straight lines and square corners very efficiently, but generally suck at curves, odd angles, etc. So, designs got modified to be more straight and square than they had previously been. The result was more uniformity in design and construction, and less individuality. Even function has been compromised, in deference to efficiency of production.

Like I said, I could write a book. There is an expression I came across a few years back: Mediocracy thrives on standardization. Even without AI, conformity of thought has become chronic to the point that conformity of thought is more accepted than rational thought and examination.

Ben, you mentioned that doctors and lawyers shouldn't have to waste their valuable time doing housework. I agree, to an extent. And to the extent that AI can take out the trash and mow the lawn, I'm pretty much for it. But let's not let it be too reliant on AI for developing our thoughts and our perspectives. Yes, it can even help with that. But human nature seems to encourage lazy thinking. That leaves AI wide open for abuse and misuse. Be careful.

Expand full comment
author

I think I may have struck a nerve, Chip! You contribute a lot of good points to the conversation. The only thing I don’t really agree with is what you said about rap. I believe rap artists have turned software into an instrument to create a whole new genre and some of the production techniques they employ I’ve been adopted across all genres when it comes to recording, and even performance.

Expand full comment

Perception is everything, they say. Yes, rap is innovative, but the emphasis on monotone lines and percussion gets pretty old for me. Or maybe it's just that I'm old!

Expand full comment

When man discovered how to master fire, I'm sure there were also many people who were afraid or critical of the use of fire. AI is doing an excellent job in the field of health, for example.

Nothing replaces what is human, but I'm sure we'll be better humans if we know how to use AI thoughtfully

Expand full comment
author

Thanks for reading and commenting Rolando.

Expand full comment

The point is truth and how Ai facilitates lies and cons ... I accept it's uses in other areas ... But as a writer, journalist or any kind of literary or art purveyor the fear is misleading facts, representations, the blurring the edges of fact and fiction ... Ai isn't the problem, it's the people who use it ...but the tool facilitates them..

Expand full comment

Agree with you

Expand full comment
Apr 7Liked by Ben Wakeman

Agree. I think the efficiency increase in tasks allows humans to move towards what humans do best, too.

Yes it's going to hurt certain job sectors a lot, but in the long run the advancements will be seen as a critical leap. Until the singularity, of course, when humankind then incinerated by the machines ;)

Expand full comment
author

It’s getting warm in here, right? Nah, said the frog. 🐸

Expand full comment

Hahaha.

Expand full comment

This is really useful Ben. I’ve tried using AI a bit here and there (I had it draft some instructions to my family on better password practices the other day), but your detailed advice on framing your prompts is especially good. Thanks.

Expand full comment
author

Glad it was helpful. Thanks Tom.

Expand full comment
Apr 8Liked by Ben Wakeman

Hey, got a question and thought I'd ask it here since others might also like to know. I tried uploading my entire memoir onto ChatGPT and it's too long. Is this the right place to do this? I think you mentioned in your essay that you did this with Harmony House?

Expand full comment
author

Hi Kim, there is a hard limit of 512KB for a single file but that should be big enough for a novel. Harmony House is 254 pages and the file size is 410KB. Here's a link to the FAQ: https://help.openai.com/en/articles/8555545-file-uploads-faq

Note that you can only upload and use your own document sources with the paid premium version of ChatGPT and you have to create your own "GPT" (mini app) to take advantage of this feature. It takes 2 minutes to create one. Here's a link to show you how: https://help.openai.com/en/articles/8554397-creating-a-gpt

Hope this helps!

Expand full comment

Very useful, and some of it is remarkably similar to an article I've recently published myself, for a different publication.

Expand full comment
author

Thanks Terry, I'm glad this was helpful.

Expand full comment

A bit hard to embrace it all while being left without a job since November (my longest draught) thanks to ai.

What I give to this tech though is that, hopefully, it’ll revive interest in theater, for example. And maybe someday 100%-human created art will be so rare that it’s value will see a history-high increase.

Expand full comment
author

Sorry to hear that Victoria. AI will be changing the landscape of jobs across the planet in the next couple of years, so you're not alone. Hopefully this little essay inspired you to find some small ways to make it work for you.

Expand full comment

I’m reading everything, from both POVs )) But it’s also problematic for me to smooth my scepticisme because I worked at the startup where they came up with the idea to create your digital twin. I witnessed the reckless jumping in the hype wagon and total neglect of persona data theft, collecting very persona information without any consent. I also worked in web3 environment for 6 years, meaning I was very closely working with tech bros and their minions )) Knowing how things work in there left me speechless too many times. And also I was involved in Women in ai movement that tries to promote ethical ai. I read tons of research and documents they share with members and again, I was left speechless.

Maybe if I didn’t see the “inside” activities I’d be less worried.

Expand full comment
Apr 7Liked by Ben Wakeman

I have been dismissive and distressed at the thought(s) aswirl about and around AI, wondering if I’ll ever know what or who has created the many texts that come before me. Only so reasoned and carefully framed an employ of a chatGPT as yours, and by someone who writes so well, is enough to make me think harder, to re-visit my pretty full dread and dismissal. I should explore . . . But I still so worry about the creep, creep in the practice, if not in your or may case, then in so many. The improved training of the thing making it more and more alluring to hand off more and more (aspects getting re-defined as) “drudgery,” more “tasks,” quite beyond summarising, prep-work, getting eye colour consistent, etc.

Expand full comment
author

You're right to be wary. There's a lot here that could (and will) go sideways at least in the near-term as the technology finds it's place and we all adapt. I feel like the more intelligent and good-hearted people who get involved early and understand the technology, the better. It's not stopping and all we can do is influence how it's used at this point.

Expand full comment