WotC Hasbro CEO optimistic about AI in D&D and MTG’s future

[Charlaquin]

My question is: is it currently socially acceptable not to actively press for it? Can you appreciate creative work that is not explicitly diverse without having to lament said lack of diversity in your public opinions or risk being labeled negatively for not doing so?

You can appreciate whatever creative works you want. If you express your opinions on a work in a public forum, you should expect other people to express their opinions on it as well. Since a lot of people desire more representation in the media they consume, I would expect expressions of appreciation towards work with very little representation to be met with commentary about that lack of representation. That doesn’t mean it isn’t “acceptable” for you to enjoy the work, or to express that enjoyment. It’s just how public discussion works.

To look at it from the other direction, does every creative work have to be multicultural and/cosmopolitan?

Obviously not, but given the very high demand for multiculturalism in creative works right now, works that are more multicultural are likely to receive more positive attention (and possibly earn more money) than those that aren’t. In the current social climate, a work is likely to receive heavy criticism on the basis of either its multiculturalism or its lack thereof.

 

[Yaarel]

‘as many calculations’ is a far cry from ‘as smart as’

And yet, even an earlier version of ChatGPT already suggests what Turing will look like.

 

[Micah Sweet]

You can appreciate whatever creative works you want. If you express your opinions on a work in a public forum, you should expect other people to express their opinions on it as well. Since a lot of people desire more representation in the media they consume, I would expect expressions of appreciation towards work with very little representation to be met with commentary about that lack of representation. That doesn’t mean it isn’t “acceptable” for you to enjoy the work, or to express that enjoyment. It’s just how public discussion works.

Obviously not, but given the very high demand for multiculturalism in creative works right now, works that are more multicultural are likely to receive more positive attention (and possibly earn more money) than those that aren’t. In the current social climate, a work is likely to receive heavy criticism on the basis of either its multiculturalism or its lack thereof.

That is extremely limiting.

 

[Charlaquin]

What generally gets lost is this will happen over a long period of time and people will move inland gradually and over generations. It will not be the major global upheaval it is often portrayed as.

Places like Miami and New York will eventually be underwater 365 days a year, but by the time that happens there will no longer be people living there, much like ancient Alexandria and Baia today.

This is key. It is a regional problem, not a global one.

Yeah, climate change isn’t an apocalyptic event. It’s the world gradually getting worse to live in, little by little, over decades. We’re already experiencing that effect in a million little ways. It’s just going to keep getting worse, year by year.

 

[Charlaquin]

I don't think every story needs to be created with a message in mind, certainly not a moral one. Not every story is about right and wrong.

Good stories are about something. That “something” doesn’t have to be what’s right and wrong, but if it isn’t something meaningful, the story is just going to be boring to the vast majority of people.

 

[Charlaquin]

That is extremely limiting.

In what way? What I said amounts to “people can create whatever they want, people can enjoy whatever they want, and people can express whatever they want about the things they do and don’t enjoy.”

You seem to be railing against the very concept of public discussion of media.

 

[mamba]

Is that into LLM territory to do that, and thus subject to hallucinating?

My second paragraph was simply that searching for excerpts is one way to avoid hallucinating.

a hallucination is just a wrong search result, you can theoretically get that with any search

If you look for excerpts then that might get the likelihood down, but if the text got it wrong (still possible), then so will the excerpt

The more the AI combines sources to generate something new (as opposed to providing links to existing pages like a search), the more likely it gets things wrong, it all depends on the data it had available

Like the case of the AI insisting that the movie was not out yet because most articles spoke about it being released in the future and the AI having no idea of what date it is…

 

[Micah Sweet]

Good stories are about something. That “something” doesn’t have to be what’s right and wrong, but if it isn’t something meaningful, the story is just going to be boring to the vast majority of people.

Being about something is not the same as having a message. Messages are about persuasion, trying to get the audience to believe or to do something. Many stories have messages, but plenty of others don't, and some of those are quality stories just as some stories with messages are quality. Having a message is not required for a good story.

 

[Yaarel]

If the subject matter is not about a diverse subject. Something specifically historical (or meant to evoke such) might qualify, depending on the subject.

Historical settings intentionally model history.

I use a mythologically accurate Viking Period as regional setting in several worlds.

During this time period, the Norse explorers cover about half the planet. The Viking setting includes very many ethnicities. But it still helps the feel of the setting to keep track of frequencies, and which ethnicities are more typical in which regions.

If there is a Viking in India, it can plausibly happen, and viceversa, but is remarkable. Norse texts dont mention reaching India, but do mention it by name, and come across the traderoutes leading there.

 

[Thomas Shey]

LLMs don't know correct from incorrect, they know what should sound good, right?

From my understanding, they mostly know what other sources have said is correct. So, theoretically, if you just feed in good data, you'll just get good data (assume your programming will let the LLM say "I don't know".)

The problem is that the feeds often contain data of radically different validity, and the way to sort it out is--questionable.