• The Tavern needs to be a safe place and comply with the law. In the UK the OSA requires us to know the age of our members and posters, and yet the UK Govt. has not provided a method to do this that is affordable to small voluntary BBS like ours. To that end, all members are safe and secure, we know you must be adults by now. However if you wish to join as a member, please contact us using the form at the [very] bottom of the page and we shall do a one to one verifiaction.

AI links and posts.

Guvnor

Guvnor

The Guvnor
Staff member
Recently my good friend and colleague posted a text (and link) that had been generated, at least partially, by a LLM 'AI'.
Since I know my views and feelings I wish to not colour others, but I would like to both invite comments and then later hold a poll.
This is a non binding open discussion and I know we will all discuss it civilly.
I think for good form we should treat LLM text, sound, code and images separately, but it is valid too argue they are the same.

How do people feel about such posts and such links?
 
Do we need an "AI helped me write this" tag? Or "AI helped me create this"?
 
I think @Ezio would need to tell us if it's technically feasible to do that?
 
Stronty Girl said:
Do we need an "AI helped me write this" tag? Or "AI helped me create this"?
Click to expand...
I do use an LLM when I'm creating the write ups for my campaign (a tool called tabletop recorder.com which converts audio to a transcript but then gives summary option). I take the lowest level of summary then seriously hack it around. If anything, the model is getting worse so needs more editing for repetition and a bit of hallucination.

The text I use is:
Source - initially Tabletoprecorder.com, edited for names, clarity and brevity with some parts injected to explain or catch elements missed in the summarising.
Click to expand...

I'd be loathe to stop using it, because it's helping me produce summaries easily which helps the players and myself in the campaign. It goes on my blog because that's my main online presence that's accessible.

I did post the original post that @Guvnor shared, but hadn't really clocked it as AI (assuming it is...). I realised the writing style was a bit off but in many ways no worse than posts from people I'd seen when English isn't their first language. I did find the general thrust of the argument it made interesting, which is why I shared it, but as ever linked the source. I was figuring that it would prompt a discussion on the content rather than the way it was made, but I guess that's where we are now.
 
I've certainly found AI useful to create scenario outlines when running my current Armitage Files based CoC campaign, with the proviso that editing is invariably required to get rid of the duff bits.

I don't read AI generated posts/articles though because I find their verbose style makes them tedious, regardless of subject matter.
 
I used AI professionally as an academic to proof read, summarise, check grammar and internal consistency. Many academics do, and there a several pro-forma declaration texts that are used at the end of papers along with other declarations and acknowledgements. The scientific world acknowledges that many AI tools are very very useful and helpful.

Much of what we read will already have been touched by an AI tool from Grammarly to Co-Pilot to NotebookLM to Claude.

It's good when you can't tell.. and in my experience that's when you need the human - from begining, middle to end. AI are only tools.
 
I don't think we need to ban AI, but I'm not sure I see the point of it here. (Other than grammar checking.) I mean, we're here to voice our opinions, and an AI essay isn't really doing that - so why would you post it?

Going back to that original post, it's kind of interesting that my "I think this is AI" comment derailed it, since the subject might have been worth discussing. And I don't know it's AI-generated, but given that I've seen that style of writing a lot in the last few months, particularly on Facebook, (and I've generated some of it myself in chatting with ChatGPT), it seems likely.

(So my preference for that post would have been to just to link to it and prompt a discussion. However, it being a FB post and some of us not being on FB means that posting the whole thing was kind of necessary. I think if Dom had followed it up with "I know this looks like AI, but I think it makes some good points because..." or somehow reflected on the post, that would been better.)

And as ltd pointed out, AI text can be incredibly wordy and isn't often fun to read.

As for Dom's blog AI-summarized sessions, I can see why he does that, but I don't read them as I am more interested in Dom's reflections on how the session went, roads not taken, and so on, and that's not in the AI-summary. (And I don't think it's the purpose of Dom's blog posts.)
 
Guvnor said:
I used AI professionally as an academic to proof read, summarise, check grammar and internal consistency. Many academics do, and there a several pro-forma declaration texts that are used at the end of papers along with other declarations and acknowledgements. The scientific world acknowledges that many AI tools are very very useful and helpful.

Much of what we read will already have been touched by an AI tool from Grammarly to Co-Pilot to NotebookLM to Claude.

It's good when you can't tell.. and in my experience that's when you need the human - from begining, middle to end. AI are only tools.
Click to expand...

I think some people forget that there's a big difference between an AI tool and generative AI technology.
 
There are indeed other AI and machine learning tools, not all are LLMs.
 
To date I haven't knowingly used AI tools for my own creative projects and for anything I might publish. I use other digital tools as you might expect.

These days, I won't buy any product on DrivethruRPG that has been flagged by the creator as being partly created by AI tools.

Even in my online and convention games, I try not to use any 'AI' generated images. I have explored their use, as an experiment, on one image, to modernise a very old line drawing, and the result was impressive. I now find it very difficult to find images on Pinterest or other sites that are not modified by AI. Perhaps my filtering isn't good enough to hold back the sludge?

Although I often feel like the luddite in these conversations, I have remained unconvinced by the responses to either the ethical or environmental impact of a range of the AI tools that are now available. Additionally, as an occasional creator, I note that the prevailing purchasing view, at least in my own circle, is intensely hostile to the use of any AI tooling.
 
I'm still figuring out where I stand on other people using AI tools, as I can see arguments both for and against. Personally, I don't use them, because I think needing them is a sign that I have over-complicated what I'm doing and need to simplify it. YMMV.
 
I think we're starting to stray into a discussion about using AI/LLM outputs in our games, as opposed to using them here on the forum.

Not that I mind having that discussion, but I didn't think this thread was for that.
 
I'd not really thought about it before but I don't think I'm averse to AI posts if they are interesting. An indication of whether they are AI (or a link to the source) could be useful though for those who do have an opinion one way or the other.
 
I am not averse to AI derived materials in posts. I think it'd be nice to identify them as such if possible but recognise that this may not be apparent.
 
steveh said:
I think we're starting to stray into a discussion about using AI/LLM outputs in our games, as opposed to using them here on the forum.

Not that I mind having that discussion, but I didn't think this thread was for that.
Click to expand...
Sorry about that — you’re right, this thread wasn’t meant to head in that direction. I didn’t mean to derail the discussion, and I’m happy to move that topic elsewhere.

(Composed using Chat GPT)
 
I don't think sharing posts from a Facebook group consisting entirely of AI generated biogs of designers is great - the community is promoting it by cross-posting - whether or not it's potentially interesting. To some extent the post got the response it probably warranted, so maybe this is a thing that can be self-policed. BUT I often get the feeling that the emergent general view on AI is "AI is terrible, except when I use it" which makes this tricky. I'm not sure what other forums do - will go and have a look.
 
First Age said:
Sorry about that — you’re right, this thread wasn’t meant to head in that direction. I didn’t mean to derail the discussion, and I’m happy to move that topic elsewhere.

(Composed using Chat GPT)
Click to expand...

I think AI like Chat GPT can be a useful tool by giving a voice to those who find it difficult to articulate themselves in writing, those with dyslexia and learning difficulties, for example.
 
Hmm, I hadn't thought of that. Good point.
 
Good point.
I think LLMs are great.
Some more focused machine learning models will change the world.
There are going to be some ructions and bumpiness on the way given that we live under rapacious capitalism.
I think we are grown up enough to not only discuss this but share and discuss AI materials.
 
Hey there everyone, appreciate that you don't want to have someone beat up an old thread but just thought it was interesting topic and I hadn't been on the forum for a wee bit. Just given that AI and LLM is getting g a lot of my focus at the moment around my workplace and in some other topics.

In keeping with the forum topic on allowing posts and linked that are part or mostly AI / LLM generated. I would suggest sparingly.

I have used some of the tools as a novelty for the image (see Avatar) but I haven't created any image for over 2 years and in the relation to text generation, if someone need it to help articulate something due to sentences and words it is helpful but I would try and make sure they are not using AI/LLM lconstantly. I have found the novelty wears off and from what I seen generated it loses a voice in the writing. This is making me develop an abhorrence to AI/LLM work than an interest as not finding it connective and having to use it than want to use it lately.

As said if someone uses it to help them, then fair but if it is consistent I tend to disengage with their work

From the AI/LLM generated stuff I come across I am reminded of the quote comparing with drunks using a lamppost for support than illumination to make their own way and I haven't seen a lot of illumination.
 
You must log in or register to reply here.
Back
Top