Blaugust and Large Language Models

Good Morning Folks! My morning has been a bit on the odd side. I had to get up super early, run into the office, do a few things, and then get back home before an 8:30 meeting. As a result, I am getting a bit of a late start to the day. There is a topic that I have been mulling over in my head and I thought I would use today as a time to open a discussion. This year with Blaugust we have had a significantly larger reach than we have in previous years. At the time of writing, we have one-hundred-seven participating blogs and fifty-nine of them are completely new to the Blaugust event. This means our community is growing, but also as it grows some of the collective cultural tenets of the community are changing a bit. Right or wrong this was an event born out of the MMORPG Blogging community, and most of us old-timers have created blogs where we ramble on about our thoughts regarding the games we are playing and the new games on the horizon that we find ourselves pining for.

I appreciate a fresh perspective as folks arrive who have been wholly disconnected from this original community. I think there has been a good deal of shared growth since we opened up the call to pretty much anyone who creates regularly syndicated content in whatever form that takes. However this year I have seen a few entrants that are very clearly using Large Language Models as part of their daily routine. This feels like we are approaching a slippery slope here, and I wanted to share some of my thoughts. I’ve screwed around a lot specifically with Large Language Models that generate images off prompts. I’ve spent most of my time playing around with Stable Diffusion, and that is precisely what I think of it as… “play”.

It is a toy that I can fiddle around with and get some curious results from, but nothing I am doing do I truly consider to be a creative endeavor. I know other folks who have been in this same blogging pod for a while do the same. Sometimes we need an image to break up the text of a post, and can’t find anything that fits our exact requirements so it is pretty easy to turn to a picture generation machine and have it poop out something to fill this void. The thing is… when I do this, I don’t take any sense of pride in “creating” something because I did not. I typed some words into a magic box and it spit out a visualization, and while that visualization might be appealing it is a largely unrepeatable event. What I take pride in are the words surrounding that visualization that I pulled from my head.

So I am finding myself in a weird state looking at the prospect of folks having LLMs spit out the text for their blog posts, even if they are tweaking them later. Blaugust originally was an event centered around writing. That is humans who are sitting down to commit the thoughts that they are thinking to the digital page and then share them with peers. Typing some prompts into a machine and having it generate thoughts for you… feels like a bit of a violation of that original goal. I don’t care what ChatGPT regurgitated from stolen data… I care about what YOU think as the person behind the screen typing the thing that I am reading. I find zero value something that an LLM generates that is formatted to look like a number of posts that it consumed that were written by someone originally.

So I find myself in an awkward position. Blaugust 2024 as the rules of the contest stand, does not have any text forbidding the use of large language models to generate posts. As a result, I find myself having to tolerate some things that seem distasteful to me personally… and given the number of back-channel comments that I have received other Blaugustans also find it distasteful. Going forward for 2025 and beyond there is likely going to be an “Anti-AI” rider that the mentors will have to work through. Again I don’t so much care about the “AI as Clipart” trend of using it as something vaguely pleasant looking to break up large blocks of text. What concerns me is using Large Language Models to generate text posts instead of sharing your own thoughts.

I’m not purposefully trying to be a Luddite, but also believe there is value in human creativity that cannot be replaced by a regurgitation of consumed source material. Maybe I am off base here, and folks don’t mind the concept of LLM content mills… but given the number of sidebars I have had with folks I am guessing that will not be the case. Anyways I am opening this discussion, feel free to drop me your thoughts and feelings below.

24 thoughts on “Blaugust and Large Language Models”

  1. Wait… 137? I obviously haven’t been keeping up with the meta/organisational side of things.

    I guess around the AI… I’d only want to use it sparingly myself, and only likely in a way that’s clear it was AI and being used to joke/comment about AI in some way. But if people wanted to use it, and people wanted to then read it… then whatever I guess. If it’s not for me, then it’s not for me.

  2. After stewing in it over night, I think there are a number of issues.

    Should a blog largely composed by AI be allowed to participate? Maybe. Kind of a gray area.

    Is such a blog post going to be remotely worth the time that it takes to read it? Almost certainly not if my experience with LLMs is any guide.

    Would allowing large numbers of such blogs to participate in Blaugust erode the mean quality of the blogs being promoted, perhaps giving some outsiders a bit of hesitation as to whether clicking on a random blog is from the list is worth it? Almost certainly.

    Could this be solved by flagging AI generated blogs and sticking them in a separate section? Probably.

  3. In the end, does it actually matter how someone else is doing their blog writing? Especially for something like a community event? Maybe they desperately WANT to participate, and due to social pressure, are feeling like complete and total failures. Thanks to AI, they are able to post daily, change how they feel, and think “well, I didn’t write it myself, but it used my prompts, and I feel like a part of the community”.

    This entire process doesn’t actually affect you at all. It makes THEM feel better. This is not school where they are being graded.

    Take pride in how YOU do it, and leave other people to how they want to do it. If you’re a fan of the writing, read it, if you’re not, don’t read it. It does not take away any of your own pride in your own writing if someone else has used AI to write/inspire theirs. This pitch fork gang is exactly why some of us don’t sign up. The amount of pressure to just fit in, to just get posts in, is a lot. There’s SO much judgement.

    I dislike everything there is about AI art. I don’t promote it, and I don’t use it. I also don’t care or shame others who do use it. It’s a moral decision that I have made for myself, for my reasons as an artist. Everyone gets to make those decisions for themselves. I don’t make any secret to the fact that I am against AI art, and I’ll talk and explain about those reasons to anyone who will listen, hoping for change – but if it doesn’t happen, I am confident that I did the right thing for my moral compass, even if it’s not the same as someone else’ compass. There are so many larger issues out there, whether or not AI was used in a blog post is the very least of my daily concerns.

    • I guess the thing is… they are being graded. Awards are handed out as part of this event, and there are folks who push themselves to get that 31+ posts during the month badge to add to their blog. Does it not diminish the effort that they put forth by letting a fully AI generated blog participate? I don’t know. Honestly that is why I threw this post out there to get a discussion going. I did not name the blog that sparked this, but on their about page they clearly state that it is an AI Generated news blog. I kinda feel like using LLMs as an inspiration engine is valid, and hell having it throw out some ideas for how to approach a topic that you are thinking about is also valid. I am not going to demonize folks wanting to use it to help their process, but I still feel real odd about someone entering a blog that is entirely and publicly denoted as being machine generated.

    • I think it absolutely matters. Blaugust has always been clear that there is no pressure to write 31 posts to take part and “win”. And we also don’t ostracise folks for not taking part, if they want to be part of the community, they are! Inspiration from AI is fine, but having an LLM generate your entire blog piece for you isn’t.. writing. Of course, if somebody else feels differently that’s their opinion and sure. But ultimately, writing daily blog posts as part of Blaugust is not the only way to be a part of the community. Else I’d never have taken part, I hope you know how much I struggle to write. 😅

      But I have to refute the judgement point. I’ve never felt judgement, or pressure to fit in and fit in posts. People have been great at uplifting and welcoming in my experience, I’m really sorry if you’ve felt differently though. But Belghast, at least, has never done this that I’ve seen.

  4. I made a post about WordPress’ AI “writing block” and despite it not being useful for me, because writing per se is what I enjoy about blogging, after all, I also don’t think it would do me any good at all if I ever used it seriously. It’s not a good writing style to begin with. Not one you would expect from a blog. AI doesn’t know what I feel and want to share with my readers and definitely can’t express my own thoughts better than I can.

    Besides, for the purpose of Blaugust, it doesn’t make any sense. The goal is that I want to challenge myself to write more than I usually do. So it definitely goes against the spirit of Blaugust.

  5. Absolutely agreed. I’m not a Luddite either and have come around to some good uses for AI, but filling the internet with more colorless and soulless word paste squeezed out by ChatGPT isn’t one of them. There’s no value to it unless you’re trying to get more clicks and generate sales or ad revenue, but that’s certainly not the point here.

  6. Blaugust should be about humans writing- posts should be written first and foremost by a person: that seems like a simple enough standard to get behind.

    I’m less concerned about “AI augmentation”. For example, I’ve puttered a bit with using the Jetpack “AI-review my subject line” and “AI- review my content” features. Checking for tone, suggesting some ideas for improvement after I’ve written my content but before I publish- that seems like a fair and reasonable use. Likewise using AI to generate an image or two to break up the text a bit seems okay to me: it isn’t the content, but a supplement to it.

    But writing the whole article? That seems well outside the original intent of Blaugust. This is coming from a ‘first timer’, so I don’t have the full history to go from.

    I am somewhat less clear in terms of other ‘non-AI’ concerns like those Bhagpuss brings up re: Blaugust growing beyond the bounds of gaming posts. I’m a gamer, yes, but that does not define the entirety of my interests. I like the broader topic range, and can ‘filter out’ the things that don’t appeal to me.

    I for one would not have been able to participate this year if the event was restricted to gaming posts as I haven’t posted about games for a while. I might be able to next year as I re-organize my blogs, but possibly not. Not being able to participate would be disappointing to me. Do with that what you will.

    • I’m certainly not advocating restrictions. Just some tags would be nice. I’d tag my blog MMORPG, Gaming, Music, Pop Culture for example. Obviously that wouldn’t restrict anyone to posting only to the tags – its your blog, you put anything you like in there – but it would give a broad indication of what to expect.

  7. One hundred percent in agreement with the sentiment here, blaugust should be AI-free. (And I am such a luddite that I think it should be blogging, written-words-focused, too, though that may be a bridge too far).

    Bel himself said in his “Blaugust is coming” post a few weeks ago, “in this era of AI-slop content, our voices are needed even more than we ever have been”, and I think that should be enshrined in a rule next year.

  8. Yes, I wondered when this would become a thing in the blogging realm. It’s been a few years since I’ve used wordpress but like Aywren said, even they have AI pieces integrated now on the backend of things. I’ve always believed the a writer should write his or her own stuff, whether that is a blog post or a book, I don’t think an AI bot should be the one to spit the words out for you. I would absolutely support and back you on this going forward.

  9. This is the reason Google pulled their commercial from the Olympics. It featured a dad talking about how his daughter could use their AI to “bootstrap” a letter to her favorite Olympian. People HATED it, saying that such a letter shouldn’t even be outlined by AI because it sucks the soul out of the endeavor.

    I’m not the best writer around, but I’m more than happy to let my own talents AND flaws represent me. Using AI to write even a little bit for a post is just lazy and possibly even cowardly.

  10. I’m not particularly bothered about the LLMs as such and I suspect that problem may already have solved itself by next Blaugust. By this time next year they’ll either be last year’s fad and as forgotten as NFTs and the metaverse or they’ll be so integrated into the mainstream, the idea of avoiding them will be highly impractical.

    I’m a lot more bothered by the sheer size and scale of Blaugust this year, which I’m finding almost more than I can – or want to – handle . I felt somewhat like this last year but at least then I didn’t feel overwhelmed until near the end. This time I’m exhausted from trying to keep up before the end of the first week – and I haven’t even added those extra thirty or so blogs that joined late to my rss feed yet! The way it’s going, I’m wondering whether I’ll even want to do this again next year or if I ought to take a year off, although I know that by then I’ll have forgotten about the stress and be looking forward to it as much as always.

    The real issue from my perspective isn’t AI or LLMs or indeed any specific subject or style or approach. It’s that, as you suggested, the event has drifted a long way from its roots and origins and now some of the new blogs are so far removed from anything I’d normally read that I struggle even to understand them. That could be an opportunity to expand my horizons, if they were providing an accessible introduction to new fields of entertainment or endeavor, but many of them seem to be established blogs that already speak to an extremely specialist audience in language that audience expects. They don’t always feel like they’re intended for a general audience.

    I wouldn’t want to discourage spreading the net wide but I’d suggest next time we might perhaps have some tags or categories, something like Steam does with genres. That way, people might have a better chance of finding the content that interests them – and indeed avoiding anything to which they feel they don’t want to be exposed, like LLMs.

    • Honestly, I don’t think that you (or anyone for that matter) needs to keep up with all the posts. For what it’s worth, I think it’s nigh impossible to read everything and as such it’s important to pace oneself and only read what seems interesting, as you usually would.

      That said, perhaps, we could split the #share-your-content channel into two categories, namely “Gaming-Content” and “Non-Gaming Content” and I think that would probably make it much easier to manage.

      • I think that’s good advice. In previous years, I’ve always added every new blog to my blog roll and made an attempt to read all the posts during the event. It usually didn’t even seem that difficult to do. Last year, though, it began to feel like too big an ask by the end of the month and this year it’s already getting on top of me.

        I’ll at least take a look every blog but I won’t try to read absolutely everything everyone posts any more. I’ll stick to the ones I find interesting or entertaining, which should be plenty!

  11. The quote I keep seeing is “why should I read something that you couldn’t bother to write?”

    I already find it challenging to balance time spent writing my own material against reading the work of others — I liked participating in Blaugust because it nudged me to spend more time on the first activity.

    I would absolutely support an effort to make sure that people were putting actual effort into their posts a.

  12. I did wonder when this would become a thing. Especially with even WordPress jumping on the AI boat by slowly integrating AI bits and pieces into the platform itself. It’s hard to get away from AI tools lurking in every corner now days.

    I feel Blaugust posts should be written by the blogger, and not a machine. That was the spirit this festival was created in, and what we celebrate is the act of human creativity.

    I’m not exactly sure what to look for to determine the use of AI in content since I’ve never investigated it. But I do think that going forward, if you feel this should be a stated requirement, I back you up on this.

Comments are closed.