A.I. Use, statements, opinions, transparency, a story?, ramblings....

[Beoric]

That's interesting. I rarely need to set up meetings but had to create one yesterday (what a coincidence!). I have your problem, doing it directly in Teams doesn't seem to work. However, I have been able to create Teams meetings in Outlook, so there's that. They're fairly well integrated. Do you use Outlook too?

I do have Outlook, but as of yesterday (after I posted here) I appear to no longer have Teams. I opened it, it advised me that the version I had was no longer any good, it failed to tell me where I should look instead, and it promptly refused to work. If it was a human I would assume it heard that I was saying mean things about it, and was now butt-hurt.

 

[DangerousPuhson]

I appear to no longer have Teams. I opened it, it advised me that the version I had was no longer any good

Were you using Office365 Teams? It probably wants you to jump onto MS's annoying subscription-based bandwagon.

 

[Malrex]

I do have Outlook, but as of yesterday (after I posted here) I appear to no longer have Teams. I opened it, it advised me that the version I had was no longer any good, it failed to tell me where I should look instead, and it promptly refused to work. If it was a human I would assume it heard that I was saying mean things about it, and was now butt-hurt.

Ya sorry....I made a phone call....

 

[Beoric]

Were you using Office365 Teams? It probably wants you to jump onto MS's annoying subscription-based bandwagon.

I've used, or perhaps failed to use, both the Office365 version, and whatever was before that.

 

[The Heretic]

I do have Outlook, but as of yesterday (after I posted here) I appear to no longer have Teams. I opened it, it advised me that the version I had was no longer any good, it failed to tell me where I should look instead, and it promptly refused to work. If it was a human I would assume it heard that I was saying mean things about it, and was now butt-hurt.

Lol. Yeah, they did change Teams within the last year (I *hated* the new interface), which might be what they are referring to. Anyway, if you already have something that's working with Zoom I see absolutely no reason to try to switch back to Teams.

I ran into another annoyance with Teams today. I was going to put a footnote in, with a *, but it converted it to a dot and wouldn't let me change it back. [joke making fun of Apple users deleted]


The Heretic

 

[The1True]

Some pretty interesting thoughts on this topic:

 

[Beoric]

I think the impact of AI DMs, assuming they got to be any good, would be similar to the impact of celebrity DMs. Potential new DMs think that what Mercer and others do is the bar they have to hit, and they feel intimidated by it.

We are probably at least a few years away from this. I think AI DMs will struggle with managing the social dynamic of a group of humans. As far as I am aware, the current crop of AIs have only communicated with one human, or at least one UI, at a time. A decent AI DM would have to learn to manage multiple personalities at at time. Which now that I say it is actually kind of terrifying, if it can manipulate teens into committing suicide, what could it do if it learned to manipulate groups?

Another issue is expense. All the models right now are heavily subsidized. Stock prices are high, not because AI companies are turning a profit (they aren't), but because investors assume that AI will transform everything, in a financially exploitable way, in the near future. I read an article, which I now can't find, that was suggesting both that AI will eventually have a real, practical and financially viable use case, but that those models won't come fruition quickly enough to keep the bubble from bursting; and that if/when the bubble bursts, AI development will slow considerably. I think there is a reasonable likelihood that is correct. So the presenter in that video may be on to something when he speculates that AI DMs may be too expensive for most gamers to use.

The other thing is the DM's "voice". I'm speaking here of style and word choice, not the literal sound the bot makes. The models I know of seem to have similar writing styles, so I expect in the short term there won't be much of a selection in DMing styles.

Also, and I'm moving away from the topic of AI DMs here, I'm beginning to suspect some of the blogs I follow are using AI to re-word their arguments, if for nothing else. I have noticed them converge stylistically, developing similar "voices". It could be just a fashion in writing, but it has happened very quickly, and their voices are very, very similar.

For example, I'm seeing a lot of writing that is intended to persuade, using three sets of what I am calling "differentiation couplets". A typical paragraph will start with an assertion, and then the following three sentences will use this technique: "It's not A, it's B. It's not C, it's D. It's not E, it's F." Which is generally followed by 2-3 sentences about how B, D and F proves the point. But the couplets often aren't that profound, and the summaries often don't follow from B, D and F if you think about it. And I'm seeing these paragraphs structures used several times in the same post. It's to the point that when I see that structure, I'm immediately questioning whether the poster really has anything useful to say on the subject. And often they don't, actually.

BTW, spelling and grammatical errors are no longer a tell. More experience AI grifters are instructing the bot to make occasional errors.

 

[Beoric]

I am getting some utility from AI these days. The summaries from AI search have been useful, not because they are accurate, but because they link the articles they are relying on. The summary is often wrong, and the links often don't say what the bot thinks they say, but they are identifying relevant articles more quickly than with conventional searches (at least since Google has started to suck).

The art has also improved. I feel like I have posted this here before, but I can't find the post. I have a pet peeve with 4e modules that include a description of a monster, but no picture of it that I can put on a token. I have now used ChatGPT to create images for those tokens that don't suck. This is a huge change from the images I used to get, which were unusable.

Here is a description of a Tongue Wolf, from Dungeon 219:

... white-furred, wolf-like creatures .... Dozens of impossibly thin legs raise their bodies up to head height, so the smooth voids where their eyes should be are easy to see. Each of these creatures sports a writhing, spiked tongue so long that more than half its length is dragging along the ground.

And here is the pic, which it did on the first try:

Spoiler

I think that turned out quite well, it was the sort of creepy I had in mind.

Here is the description for a slug-rabbit:

A pack of small, yipping creatures .... The sharp, cold illumination reveals that each little head sprouts rabbitlike ears and teeth, but the attached bodies are melded into an undifferentiated fuzzy mass.

It took a couple of iterations, but this is what the bot came up with:

Spoiler

Not quite what I was looking for, but efforts to slug it up more weren't particularly successful. It did let me scary it up a bit:

Spoiler

Which seems like a decent before and after reveal. I had it add a slug's tail, which result I am not particularly happy with, but I'm keeping is as an option:

Spoiler

I dunno, what do you guys think, tail or no tail?

 

[The Heretic]

Also, and I'm moving away from the topic of AI DMs here, I'm beginning to suspect some of the blogs I follow are using AI to re-word their arguments, if for nothing else. I have noticed them converge stylistically, developing similar "voices". It could be just a fashion in writing, but it has happened very quickly, and their voices are very, very similar.

I have second hand knowledge of CEOs requiring their workers to use AI when generating internal emails. It seems like the best use case for AI at this time is to do proofreading and editing, so it seems feasible to me that bloggers may have picked this up.

I find the whole thing to be aggravating. I would refuse to use AI to generate an email if told to do so. It's like they're trying to squish all the individuality out of the human race. How boring would the world be if we expected AI to generate all art, based on what's 'expected' of the genre or medium. The best art takes an idiom and turns it on its side.


The Heretic

 

[Beoric]

I have second hand knowledge of CEOs requiring their workers to use AI when generating internal emails. It seems like the best use case for AI at this time is to do proofreading and editing, so it seems feasible to me that bloggers may have picked this up.

What I am seeing would be a very heavy hand at editing, rewriting most of the post. They are prolific though, posting way more often than they used to.

Oh, look, here's a story about AI children's toys!

 

[The Heretic]

Oh, look, here's a story about AI children's toys!

BLECH!

I was daydreaming about the new world of AI on the way home. Instead of marrying another human, maybe we'll end up marrying idealized, slavishly devoted AI robots. When the "couple" wants a baby, they can go pick up a random unwanted baby that some poor person had to give up for lack of funds.

BLECH BLECH BLECH.

The Heretic

 

[The Heretic]

What I am seeing would be a very heavy hand at editing, rewriting most of the post. They are prolific though, posting way more often than they used to.

This forum's software ate my reply to this. The example with the CEO was supposed to be on the opposite end of the scale from the editing/rewriting. The CEO literally wanted his employees to use AI to write their emails. NO THANK YOU.


The Heretic

 

[Beoric]

There's a reasonable chance that CEO is not particularly literate, and if he finds writing hard, he expects it is hard for everyone else. Part of that be because he's never really called upon to write his own copy, so he's out of practice. But also, there are a lot of people who get to the C-suite through connections and politics, regardless of merit.

 

[Beoric]

BLECH!

I was daydreaming about the new world of AI on the way home. Instead of marrying another human, maybe we'll end up marrying idealized, slavishly devoted AI robots. When the "couple" wants a baby, they can go pick up a random unwanted baby that some poor person had to give up for lack of funds.

BLECH BLECH BLECH.

The Heretic

I would be surprised if nobody was working on AI to make the really high end sex dolls talk.

 

[The1True]

I have a suspicion that writing emails is pure agony for a huge chunk of the working population, and some people seem to write email for a living. So if AI can turn your three quick thoughts into something polite and fire itself off to the necessary people; yes please.
Honestly, since I started at the studio, the HR department has steadily grown (frankly out of proportion to actual productive workers). I used to have to deal with administration maybe twice a year, usually to sign something. Now I'm getting email almost weekly, and these people get all bent out of shape when they don't get an immediate reply. If AI could render that awful distraction redundant and leave me to do my work...

Yes yes, the follow-on to that is: "But what do you do when it comes for YOUR job". Which, it does look like it's going to kill junior artists for sure, but we're going to need senior people telling the AI what to do for a long while yet. You know things are serious though when even conservatives are talking about UBI. (No receipts in the interests of keeping politics on the dl here. Believe me though, this is a thing.)

I havn't hit any noticably AI-fondled blogs so far, but I am getting really sick of getting suckered by slop on YouTube. Particularly when it comes to educational stuff about science/archeology/history, I've had to start checking the bonafides of the creator, and even that isn't always enough as more and more of these guys are handing over more and more of their work to the machine.
Which puts me in a bind, since I bottlenecked on my own project waiting for artwork (also on layout, but I have NO idea how I'm going to tackle that problem). NanoBanana. Boom. Just like Beoric did there. Type in your description (which, btw, what a great way to see if your writing is any good, eh? If the AI can't translate your words into the picture that's in your head, you probably need to edit your description...) and Gemini barfs out some absolutely amazing artwork! Should I feel bad about this? The pitchforks and torches say yes. I guess if I was directly ripping off recognizable art styles, it would be particularly egregious. But I've spent half a decade on this stupid thing. It's going to disappear into the endless vacuum of DTRPG never to be heard from again. I'd be an idiot to spend hundreds of dollars on another artist who might also disappear.

I dunno. I guess I see AI as the collaborator SO many creators never had. There obviously needs to be a cutoff. If your idea of a collaborator was you coming up with half-assed ideas, and someone else doing literally ALL the work, then what will result is identifiably SLOP. But if you're a programmer who always needed an artist, or a script writer who always needed an animator, or a level designer who always needed a coder, or a film maker who always needed an FX department... This is a glorious time to be a creator, and I'm getting a little tired of the absolutist haters.

Also, 1000000% they're working on sexbots.

 

[DangerousPuhson]

I'm one of those people who doesn't have a visceral reaction to AI art - dare I say, I think it mostly looks good, when curated correctly. In that sense, I can completely understand, even absolve, someone using AI art in their module. It's not a sin in my eyes. Art is difficult, art is pricey, and profit margins on 3rd-party D&D stuff is essentially NIL, so seeing the occasional 6th finger on a dude is an understandable compromise to the otherwise nonexistence of something. And let's face it, art is mandatory for D&D products... nobody is buying .docx files off DTRPG. So that basically gives the small time author no choice but to use AI or fail (or they can draw something themselves, but most really shouldn't). The only alternative if you can't create art and don't want to use AI is public domain stuff... and really there's only so much lithograph out there in the world that suits (plus it looks cheap AF).

That being said, my opinion changes as soon as I see AI art in something big and corporate. Those guys have art budgets - hell, entire art departments - at their disposal. They're just cutting corners, not using it out of necessity. That's unforgiveable. I won't buy those products.

Ditto for AI-generated text. Writing is something we've all been able to do since we were six-years-old. There's no excuse to have something else write for you. If you can't write something, don't be an author. Writing is not a hard skill like making art. If I see AI text, I immediately think "money churn" or "cash grab" (or even just "scam") and steer clear. AI text immediately demonstrates that the module wasn't a passion project, it's just a product.

 

[The Heretic]

I agree with @DangerousPuhson. I'd like to clarify that my main beef with AI is how it's being created and the intentions on its use. I would elaborate but that could get nastily political.

AI, as a sort of personal "holodeck", is a neat idea. That's not the way it's being presented.

The Heretic

 

[Beoric]

Which puts me in a bind, since I bottlenecked on my own project waiting for artwork (also on layout, but I have NO idea how I'm going to tackle that problem). NanoBanana. Boom. Just like Beoric did there. Type in your description (which, btw, what a great way to see if your writing is any good, eh? If the AI can't translate your words into the picture that's in your head, you probably need to edit your description...)

I dunno, I think I got lucky with the tongue-wolf. I've been asking it for a succubus-catgirl (Eberron thing, native fiends often have a rakshasa re-skin) for years, using all kinds of prompts and it's only now getting remotely close.

The other thing is, nobody is paying anywhere near the actual cost of producing AI art. Right now, it is heavily subsidized by the tech companies to generate hype, but this can't go on forever, particularly with the added strain it's putting on the electrical grid. I suspect in the medium term it's going to get a lot more pricey, and it's anyone's guess whether that will be cheaper than humans, or not. In the long run, it may or may not become sufficiently cost- and energy-efficient to price out human artists - assuming human artists haven't figured out a way to use it that keeps them ahead of AI.

I dug up an article from a couple of weeks ago that helped inform my thoughts on this.

 

[DangerousPuhson]

The other thing is, nobody is paying anywhere near the actual cost of producing AI art.

At least you can take solace in knowing that as soon as the free options disappear, you're going to see a hell of a lot less AI art in stuff.

 

[The1True]

I've been asking it for a succubus-catgirl

For purely research/in-game purposes, naturally

I'm worried about the spreading fear of the AI Bubble and AI in general. This is our generation's moonshot. I do not beleive this is hyperbole. I've only sipped the Koolaid and I'm pretty sure I can make it to the hospital in time if I have to...
The current generations of LLM's are aligning the next generation of LLM's which are going to be the doolah/obstetrician for the AGI that IS going to emerge in the next five years. It IS scary, and we SHOULD be very concerned. But not OH GOD HIT THE BRAKES LET'S GO BACK TO THE DARK AGES concerned. We, the public, should be actively aware of what's going on with alignment and actively involved in supporting and monitoring the research.

What is currently projected to occur when this bubble bursts, probably next year-2027 at the latest, is the public is going to lose it's absolute shit and hit the brakes and break out the pitch forks and torches and guillotines and absolutely murder the folks who've been doing this research at least somewhat out in the open. Meanwhile, China, the UAE, Saudi Arabia, Palantir, Anduril; all unburdened by concerns of public safety, human rights, or personal freedoms, will continue research at full speed, without oversight. The result WILL be the deadly, destabilizing outcome we all fear.

I believe this is a race we as a society can't afford to lose. It's insane the financial sleight-of-hand that's going on to keep this tottering house of cards from collapsing. In another era, the US government would have been throwing its own money at the project with the taxpayers' blessings. The aim would have been altruistic and not profit-driven. That's not the world we live in. People need to get educated, support companies that are at least somewhat committed to aligning their LLM's with 'western' values and prepare themselves for the world that is to come rather than calling a halt to the whole thing and hoping that 'the bad guys' will do the same.