D&D General AI ranks D&D editions

[Snarf Zagyg]

So, you're saying it was trained on boosted articles from twitter/fb/google news and internet discussion boards?

hmmm..... now I want the AI to rank STAR TREK CAPTAINS!

 

[J.Quondam]

One of my favorite pagan and folklore writers just made a post about this. Apparently the #1 book under "celtic paganism" on Amazon right now is by "Mari Silva", who appears to be an AI-based writer or content farm, both from reviews of their text and the fact that they've put out 312 books in the last two and a half years.

Iirc, a few weeks ago, Facebook/Meta withdrew an online interface to a big science AI project, Galactica, in part because it was generating scientific papers, etc, that looked authoritative, but were completely bogus in terms of the fake "facts" they presented and/or citations they invented.

These AIs simply don't understand the difference between a legitimate text and a "yeah, that looks legit" text. And that's a massive problem because we human beings have similar bugs in our own brains which make us susceptible to confirmation bias, propaganda, and the like. This stuff can be used (and almost certainly is being used) to deliberately deceive, defraud, and otherwise misinform on a massive scale, for all sorts of nefarious ends.

 

[HomegrownHydra]

Proof that AI isn't replacing actual writers any time soon lol. Many of those things are...not remotely correct, and some of the most popular discussed features and criticisms aren't even mentioned.

But the same would be true for a response by a person. If you were to provide a single paragraph summary of each edition people would be highly critical that some of what you wrote is wrong and that you left out some of the most significant qualities.

 

[Vaalingrade]

hmmm..... now I want the AI to rank STAR TREK CAPTAINS!

1) The Doctor whenever he was left in charge*
2) Benjamin Sisko
3) Jean Luc Picard
No Others Found

*May be due to AI bias

 

[EzekielRaiden]

That text is frankly Turing complete, if you hadn't said it was AI I would never have known, absolutely looks like something a person would put together.

That's not really how the Turing test works, firstly. More importantly, very short stuff like this has been achievable since at least GPT-2, and I'm fairly sure the above is either GPT-3 or even the model coming after it (which may or may not end up getting called GPT-4.)

 

[Benjamin Olson]

That text is frankly Turing complete, if you hadn't said it was AI I would never have known, absolutely looks like something a person would put together.

Absolutely true, though I would say it looks like what an actual person with no interest in the subject, working on a deadline would say about it. It reads like what a journalist might give as background, in passing, or like the text of some clickbait listical. It has 'I was obligated to fulfill a word count" all over it.

Which will be perfect for this technology's future filling the internet with more terrible listicals and writing student papers.

 

[Levistus's_Leviathan]

I mean ... that's one way to look at it.

Another way to look at it is this- you now have freely available, publicly accessibly AIs that can generate text from prompts that look as if a human generated them. It may not be the best work out there, but it's better than a lot of high school level writing (not to mention the "foreign webpages" we see so often).

This is just the last couple of years, and freely available, and generated from a writing prompt BTW. Something that five years would not have been remotely possible.

Put it this way-

Three years ago, if someone suggested that AIs would be replacing artists in the next twenty years, I would probably have laughed.
This year, we are having serious conversations about ... AI replacing artists (illustrators, etc.) and the ramifications of that in the next few years.

Things are moving really really really quickly- and in a lot examples, we don't understand how these things operate ... just that, unlike a lot of other things, they are scalable.

I've recently become increasingly aware that a "robot apocalypse" where Ultron, Skynet, or the machines from the Matrix brutally take over the planet and kill/enslave all humans is pure fiction and almost definitely not going to happen. Robots will take over the world, but not like that. The robot revolution won't be fast or bloody, it will just be the result of them being better at our jobs than we are, and thus us being replaced by them.

It's a bit distressing, even if it's probably not going to happen in the next decade or two. Eventually we're all going to become Luddites.

 

[Charlaquin]

I've recently become increasingly aware that a "robot apocalypse" where Ultron, Skynet, or the machines from the Matrix brutally take over the planet and kill/enslave all humans is pure fiction and almost definitely not going to happen. Robots will take over the world, but not like that. The robot revolution won't be fast or bloody, it will just be the result of them being better at our jobs than we are, and thus us being replaced by them.

It's a bit distressing, even if it's probably not going to happen in the next decade or two. Eventually we're all going to become Luddites.

Much like how climate collapse won’t be an apocalyptic event, just a gradual process of life getting a little bit worse. Both things are cases of boiling the frog slowly.

 

[Charlaquin]

This is largely pretty solid, though obviously missed that AD&D 1e was far from the original edition of D&D. Also pretty strange that it claims 2e originated character classes, levels, and alignment.

 

[J.Quondam]

I've recently become increasingly aware that a "robot apocalypse" where Ultron, Skynet, or the machines from the Matrix brutally take over the planet and kill/enslave all humans is pure fiction and almost definitely not going to happen. Robots will take over the world, but not like that. The robot revolution won't be fast or bloody, it will just be the result of them being better at our jobs than we are, and thus us being replaced by them.

It's a bit distressing, even if it's probably not going to happen in the next decade or two. Eventually we're all going to become Luddites.

Yeah. In the end, AI are just tools, so how it all plays out comes down to how those tools are used. Of course, humanity being what it is, those tools are at least as likely to be used for bad thing as for good.
And also for unabashedly stupid things.