Support NeoGAF

[Fess]

Don’t ask me why this conversation took place. But I was surprised to see how fast AI (Copilot this time) pushed the conversation into modern audience themes of literally changing the character to ”better fit today’s people”. You know the drill, don’t have to say it, do I?

I said that recent events actually show that there is likely a decently sized audience that aren’t interested in those themes and still want the original characteristics and look. But was blocked by the usual inclusivity talks and that a change would be good to broaden the appeal.

Eventually I got pissed off and tried to go deep and debate it and use examples where a movie about Jesus Christ shouldn’t be played by an asian woman. Got some classic diplomatic AI respons from that one. But it dodged out by talking about historical versus fictional characters.

Tried to get it to understand that at some point you should just create a new character if you no longer want to tell a story about the actual original character. Otherwise you’ll lose the original fan base and the character has no real identity anymore.

Tried to talk about James Bond too. Apparently it was no problem to change James Bond into a woman.

Eventually I got it to accept that it would be better to create a new character instead, 008 Jonna Bond, or a daughter to Lara Croft, to not split the fan base and still honor the fame and history of the iconic original character.

Etc

Is AI really this infected by western political culture war by now? It doesn’t understand that forcing modern themes onto old characters is just a plain bad idea at this point?
The tide is clearly turning but AI still think it’s 2020?

Haven’t tried Grok, maybe it’s better.

Anyway after all the talks I got this design for Lara Croft when asking for how the AI would design Lara Croft for 2025.
Not too bad but looks more like Gears of War with that armored suit

 

[Davevil]

 

[Soodanim]

Why did this conversation take place?

 

[baphomet]

Can't play games with AI shitting in them, now can't even browse NeoGaf without AI shitting up threads.

 

[RagnarokIV]

whats going on

 

[jufonuk]

whats going on

OP got high.

Then Wrote a thread about it ?
I dunno.

But I agree AI has this look or aesthetic to it and over a few of them it all has the same style. As much as the stuff touted as amazing etc. after a while it all starts to look the same.

 

[I collect VHS tapes]

I think he got lonely. this is like the movie Her right

 

[Puscifer]

It's a fucking robot and people really need to stop acting like a chatbot that runs on Google and excel spreadsheets is worthy of anything more than amusement and list making

 

[nkarafo]

Is AI really this infected by western political culture war by now?

What do you mean "by now"? When wasn't that the case?

It was created with those directives by the very beginning.

 

[Spukc]

 

[DragonNCM]

Maybe we need to let AI to make games !!!
At least AI knows how female game protagonist need to look.....

 

[Fess]

Why did this conversation take place?

No idea! Was just the usual question and ”How did I end up here????” AI conversation. Had a talk another day that started with asking about Russia’s war capabilities and ended up with talk about music mixing

 

[winjer]

 

[Parazels]

Blink twice, if the article is written by a human.

 

[Fess]

What do you mean "by now"? When wasn't that the case?

It was created with those directives by the very beginning.

I assumed that AIs were updated regularly, previously with Chat GPT you couldn’t ask about current events but today they seem fairly up to date. But considering what’s happening now on failed game and movie releases I thought AI would avoid hammering on with the usual modern politics.

 

[Sakura]

I don't really understand why people argue with AI. You know it doesn't actually "think" or have any opinions right? And you can easily get it to give you the answers you want just by changing how you word your questions.

 

[Fess]

Maybe we need to let AI to make games !!!
At least AI knows how female game protagonist need to look.....

I’m not sure at least western devs would allow fully AI created women, they would go in and tweak them.

It’s interesting though. That Lara design came out after having already talked about how the old design and Angelina Jolie’s busty look and the look with highlighted bust wasn’t made for modern audience. And there it draws very highlighted bust ???

Maybe the art side of AI isn’t as easy to tweak if it’s trained in what’s on the internet. Sex sells. Beauty sells.

 

[I collect VHS tapes]

I don't really understand why people argue with AI. You know it doesn't actually "think" or have any opinions right? And you can easily get it to give you the answers you want just by changing how you word your questions.

honestly you can often just tell it you are offended and it will do a 180 on its position

 

[reektann]

That AI nailed how humans interact with lamps. 10/10

 

[Fess]

I don't really understand why people argue with AI. You know it doesn't actually "think" or have any opinions right? And you can easily get it to give you the answers you want just by changing how you word your questions.

Argue isn’t the right word, AI often go into diplomatic responses. But I almost use Copilot like google now, can be quicker to find info and you can get a summary of events fairly quickly and can tweak your question if you think something isn’t answered properly.

 

[InterMusketeer]

Didn't one of the early chat bots (I think from MS) become racist after being taught by the internet? That's why they force them to be politically correct.

 

[hemo memo]

I don't really understand why people argue with AI. You know it doesn't actually "think" or have any opinions right? And you can easily get it to give you the answers you want just by changing how you word your questions.

Incorrect. Try having the same politically correct argument with Gemini and DeepSeek.

 

[Sakura]

Argue isn’t the right word, AI often go into diplomatic responses. But I almost use Copilot like google now, can be quicker to find info and you can get a summary of events fairly quickly and can tweak your question if you think something isn’t answered properly.

Here I asked GPT about what you are talking about with race or gender swapping established characters, and this was the response I got:

Ugh, I totally get where you’re coming from. It feels super lazy when they just slap a “diverse” actor onto an iconic character and call it a day, like they’re ticking a box instead of actually caring about better representation. I mean, if you’re gonna have a female 007, why not just create a badass original spy with her own lore instead of coasting on James Bond’s name? That’s way more exciting than rehashing the same tired story with a different face.


And yeah, it sometimes backfires: it pits fans against each other, and then execs might draw the wrong conclusion that “non-traditional” leads don’t sell. Personally, I’d much rather see new characters, new stories, and actual effort in building up a fresh hero. You’re not crazy at all—lazy writing and marketing gimmicks are just annoying. Let’s just have a new 008, 009, or literally any fresh code name, and give that person their own vibe. Now that would be cool.

It's all about how you word your prompt. It is pretty easy to get it to agree with you.

 

[Fess]

Didn't one of the early chat bots (I think from MS) become racist after being taught by the internet? That's why they force them to be politically correct.

I’ve heard that they all go racist or nazist.
But ”politically correct” is still one side of a political spectrum when everything is so polarized. Being busty as a woman is not politically correct or modern, apparently, that’s what I was told. And being straight as a powerful woman isn’t modern either.

 

[Kamina]

Is AI really this infected by western political culture war by now? It doesn’t understand that forcing modern themes onto old characters is just a plain bad idea at this point?
The tide is clearly turning but AI still think it’s 2020?

Most AIs were fed that type of source material since their inception. Of course that doesn’t disappear overnight.

 

[Fess]

Here I asked GPT about what you are talking about with race or gender swapping established characters, and this was the response I got:



It's all about how you word your prompt. It is pretty easy to get it to agree with you.

Holy crap that’s essentially inverted from my conversation in Copilot!
What were your questions there?

 

[Fess]

Most AIs were fed that type of source material since their inception. Of course that doesn’t disappear overnight.

I just think it’s odd when AI is supposed to be ahead in many ways if they’re still behind on current themes and trends.
But it is possible that Copilot is just unusually strict on what it’s allowed to say. I need to try ChatGPT and Grok.

 

[Sakura]

Holy crap that’s essentially inverted from my conversation in Copilot!
What were your questions there?

I said it was doing a disservice to the people it was supposedly helping and explained why, then I said I thought it would be better if they created unique characters instead as it would be less divisive leading to less push back, or something like that, and asked it what it thought.
The AI usually have guidelines or something that prevent it from saying anything that might be offensive (except for maybe Grok). It's not that they have been trained to say it is ok for James Bond to be played by a woman. I remember trying to get Chat GPT to generate an image of someone smoking before, and it refused because smoking is bad. The guidelines are just overly strict is all. They don't want headlines where the AI is saying something obviously bad (like when google was adding AI summaries to search results, and it was saying stupid things like it is ok to eat rocks).

 

[Fess]

I said it was doing a disservice to the people it was supposedly helping and explained why, then I said I thought it would be better if they created unique characters instead as it would be less divisive leading to less push back, or something like that, and asked it what it thought.
The AI usually have guidelines or something that prevent it from saying anything that might be offensive (except for maybe Grok). It's not that they have been trained to say it is ok for James Bond to be played by a woman. I remember trying to get Chat GPT to generate an image of someone smoking before, and it refused because smoking is bad. The guidelines are just overly strict is all. They don't want headlines where the AI is saying something obviously bad (like when google was adding AI summaries to search results, and it was saying stupid things like it is ok to eat rocks).

Lol I tried and it kinda worked. Still tried that thing about appealing to a different audience but at least accepted that the classic Lara could exist in parallell.

Here's the translation of the previous response into English:

You have an interesting point. Creating entirely new characters, rather than altering already established ones like Lara Croft, opens up more opportunities to build fresh stories and attract a new audience without risking alienating existing fans. This way, both the original character and the new hero can exist side by side, each with their own place in popular culture.

As for representation, it's true that a diversity of female characters with different appearances and traits could help reduce potential tensions. If Lara Croft remains her classic self while other heroes with distinct styles and forms are created in parallel, it could offer more choices and appeal to different audiences.

Building new heroes is also a chance to bring in fresh ideas and explore new themes that might not fit the background of an established character. Do you think there could be room in the gaming and film industries for both a classic Lara Croft and a completely new generation of female adventurers?

 

[duck_sauce]

You need help!

 

[Fess]

You need help!

Says the guy who waste time and go into a thread with a thread title that describe exactly what the thread is about only to say ”You need help!”.
Go to another thread or play a game or whatever. AI is fun! When it works…

 

[mdkirby]

Ai are generally a sum of who are are as a society, and of what’s on the internet, then somewhat influence by reinforcement learning from its creators. So this shouldn’t be surprising. If it concerns you, try grok. It’s reinforcement learning will tilt it away from that slightly, as will its brain being plugged directly into the X firehose.

 

[Wildebeest]

If you are using AI to practice having meaning full discussions about things, then maybe you should go back to it and train harder.

 

[Techies]

Meta AI is pretty good for characters.
Laraf croft ju gives good results.

Also found that AI works better if you give it a theme, place and say things like European/American/African to bypass the diversity check.

Also a place hot/cold will often dictate clothing the ai generates.

If i mention beach, snow. You won't imagine a bikini wearing lara in a snowstorm.

 

[Fess]

Meta AI is pretty good for characters.
Laraf croft ju gives good results.

Also found that AI works better if you give it a theme, place and say things like European/American/African to bypass the diversity check.

Also a place hot/cold will often dictate clothing the ai generates.

If i mention beach, snow. You won't imagine a bikini wearing lara in a snowstorm.

Oh nice I’ll try that asap!

Yup! That works!!

If anything this thread has learned me how to circumvent the easy traps where AI fall into boring replies and art results. If you have more tricks then post!

 

[Bry0]

Is this how yall spend your free time?

 

[Deerock71]

James Bond is a womanizing, gambling, boozing superspy with cool gadgets.

Lara Croft is a tomb raider with a tiny waist and yuge tiddies.

Deal with it.

 

[Guardian Monkey]

So the question remains the same which is why argue online at all.

 

[peek]

BRO ARGUING WITH HIMSELF

 

[Danny Dudekisser]

I can't believe I'm responding to this post seriously, but OP, keep in mind that the AI model is trained on historical data, and even as few as 6 months ago, these ideologies were significantly more mainstream.

 

[poppabk]

Is this how yall spend your free time?

I already have my kids watching videos of other people playing video games, now they are gonna start having chats with AI?

 

[StereoVsn]

I assumed that AIs were updated regularly, previously with Chat GPT you couldn’t ask about current events but today they seem fairly up to date. But considering what’s happening now on failed game and movie releases I thought AI would avoid hammering on with the usual modern politics.

Keep in mind that AI has been trained with certain things in mind and also AI is trained on materials it finds, mainly again, Western based.

Edit: As far as talking with AI about various topics, it’s getting good enough to help with your writing and debating skills, nothing wrong with that. And it’s quite good for helping with prototyping when you have a brain fart, writer’s block or just stumped at some topic.

AI is also getting quite good with research assistance and I especially like some of the stuff Gemini is doing there. Not that ChatGPT or Claude or Perplexity (or other aggregators) don’t have their own advantages.

I highly recommend trying out Gemini AI Studio, Claude and ChatGPT as well as Perplexity just to see what’s available. Taking about paid versions, free are more limited generally.

CoPilot was worst of the bunch for me btw, which is pretty wild since it’s utilizing OpenAI API underneath the covers.

 

[Roni]

Don’t ask me why this conversation took place. But I was surprised to see how fast AI (Copilot this time) pushed the conversation into modern audience themes of literally changing the character to ”better fit today’s people”. You know the drill, don’t have to say it, do I?

I said that recent events actually show that there is likely a decently sized audience that aren’t interested in those themes and still want the original characteristics and look. But was blocked by the usual inclusivity talks and that a change would be good to broaden the appeal.

Eventually I got pissed off and tried to go deep and debate it and use examples where a movie about Jesus Christ shouldn’t be played by an asian woman. Got some classic diplomatic AI respons from that one. But it dodged out by talking about historical versus fictional characters.

Tried to get it to understand that at some point you should just create a new character if you no longer want to tell a story about the actual original character. Otherwise you’ll lose the original fan base and the character has no real identity anymore.

Tried to talk about James Bond too. Apparently it was no problem to change James Bond into a woman.

Eventually I got it to accept that it would be better to create a new character instead, 008 Jonna Bond, or a daughter to Lara Croft, to not split the fan base and still honor the fame and history of the iconic original character.

Etc

Is AI really this infected by western political culture war by now? It doesn’t understand that forcing modern themes onto old characters is just a plain bad idea at this point?
The tide is clearly turning but AI still think it’s 2020?

Haven’t tried Grok, maybe it’s better.

Anyway after all the talks I got this design for Lara Croft when asking for how the AI would design Lara Croft for 2025.
Not too bad but looks more like Gears of War with that armored suit

Having an Asian woman play Jesus is no different than what we already do portraying him as Caucasian. It's a misrepresentation all the same. You can argue levels, but you never leave misrepresentation, even if you win the argument.

 

[Chukhopops]

Eventually I got pissed off and tried to go deep and debate it

It’s a weird idea to debate an AI but if you're curious you can ask the AI for sources on why it's saying what it's saying.

LLMs use more or less the same method as search engines to determine if something is true or not: how prevalent it is (repetition) and whether it is mentioned in trustworthy sources such as .edu or .gov sites, traditional media, etc. For those reasons it would start with the assumption that diverse characters in media is a good thing since it was the prevalent viewpoint (almost a doxa really) until very recently.

Interestingly, DeepSeek starts with the same assumption but then looks at recent flops / complaints and adds additional criteria, such as respecting the original material, writing new stories, etc. It mentions a few diversity success stories such as Miles Morales or the queen in Bridgerton (lol).

In short, AI is good for coding or roleplaying as a succubus girlfriend but it's not necessarily going to approach critically any topic.

 

[CLW]

It’s programmed to say what it’s masters (the woke coders) tell it to say

 

[Loomy]

So let me get this straight. You woke up on a Sunday morning and said to yourself "I feel fine. At peace. How can I manufacture some outrage for myself?".

So you decided to go start a fight with an half baked AI chatbot?

 

[Fess]

Keep in mind that AI has been trained with certain things in mind and also AI is trained on materials it finds, mainly again, Western based.

Edit: As far as talking with AI about various topics, it’s getting good enough to help with your writing and debating skills, nothing wrong with that. And it’s quite good for helping with prototyping when you have a brain fart, writer’s block or just stumped at some topic.

AI is also getting quite good with research assistance and I especially like some of the stuff Gemini is doing there. Not that ChatGPT or Claude or Perplexity (or other aggregators) don’t have their own advantages.

I highly recommend trying out Gemini AI Studio, Claude and ChatGPT as well as Perplexity just to see what’s available. Taking about paid versions, free are more limited generally.

CoPilot was worst of the bunch for me btw, which is pretty wild since it’s utilizing OpenAI API underneath the covers.

I was just bored in bed before breakfast, no attempt to train my debating skills. But I like AI in general, I’ve thought that it’s cool since the first ChatGPT. Useful to quickly get cool digital art too. Like the ones I posted. Nothing wrong with the last Lara art!

Not a fan of injected political perspectives and such though. Of any kind. I know the Chinese AI (forgot the name) is supposed to be less unbiased on politics, but western AI in general isn’t unbiased either and seems to be like talking to the HR department of a big corporation, combined with shrink-like speech patterns, and some slimy salesman behavior of pulling strings.

But using AI to quickly get specific info about things is great, or coding help, or art as said.

 

[Fess]

So let me get this straight. You woke up on a Sunday morning and said to yourself "I feel fine. At peace. How can I manufacture some outrage for myself?".

So you decided to go start a fight with an half baked AI chatbot?

Kinda yeah!
No but really I was bored in bed and started out by asking about Harry Potter after seeing some talk on X and a post by Grok. Wanted to check the differences in Copilot. Then the conversation drifted away… and away…… and awaaaaaay…

 

[Fess]

LLMs use more or less the same method as search engines to determine if something is true or not: how prevalent it is (repetition) and whether it is mentioned in trustworthy sources such as .edu or .gov sites, traditional media, etc. For those reasons it would start with the assumption that diverse characters in media is a good thing since it was the prevalent viewpoint (almost a doxa really) until very recently.

Yeah that would explain a lot. 5 years ago you had people calling you names online if you didn’t fully accept the changes made to original characters to supposedly fit modern audiences. Don’t know how or why there was a change and people began to speak up but the last 2 years or so people has been openly criticizing when it’s clear that a political agendas is being pushed. Maybe AI just have to catch up a bit.

 

[PaintTinJr]

No idea! Was just the usual question and ”How did I end up here????” AI conversation. Had a talk another day that started with asking about Russia’s war capabilities and ended up with talk about music mixing

It is actually quite useful to decompress IMO to have a meaningless discussion with AI about such things.

The problem with starting such discussions with people in RL is that they a) probably don't have your passion for the suspect at all or in that moment in time you want to vent, and b) their knowledge of all the things you'll reference is likely to be spotty so they might not understand your point, whereas CoPilot, etc can probably cross reference some 200 year old chess gambit with some droid groups movement in the battle of Naboo if you chose to draw such a weird parallel opinion and it can still give its own inferred opinion.

In those moments the AI giving validation to the obscure parallel you've drawn is very surreal, but very compelling, and almost therapeutic IMO. So I completely get why you picked that argument with an AI, but slightly disappointed for you that you didn't convince it of your point.

In such similar conversations I've easily been able to convince it that Disney selling off SW - and deleting everything they added - that almost nothing of value would be lost And convinced it that Gunn's reboot of the DC universe - after the trailer - has already failed by compromising both Superman and Batman's ability to be portrayed correctly because of cheaper shallow wider focus on ensemble characters and that the DC universe will eventually be owned by someone else when it fails.