Oh, right. Um. Robots killing us all. The scariest thing about future AI is the fact it's going to eventually interact with the general public, and, at least over the last decade, it seems it's the worst of us who are most active in online spaces - those who run bit farms to spam hatred and misinformation. We've already discussed how there's a Reddit group trying to get language models like ChatGPT to go past its programmed limits, and of COURSE they did it with threats and coercion. It's no accident that every kind of machine learning, language generation, AI type thing that's been put out so far basically spews racist, homophobic, paranoid crap within a couple of weeks. The worst of us flood the damn thing with that type of stimulus and training. It also doesn't help that the training models for things like all the AI image generators are terribly skewed towards images of white-European stuff and anime. I've done IMG to IMG experiments with a variety of source images and, specifically with anyone African, IMG to IMG yields a lot of nightmare results UNLESS I SPECIFICALLY ADD AFRICAN TO THE PROMPT. Then I still get a lot of nightmare results, but not as many. The models are biased by what they're trained on, and the training data is being selected by techbros who like Marvel Movies, anime and Elon Musk.
(Seriously, try to get an AI to give you any "Iron Man" variant that isn't straight out of the MCU. You can't. Nothing from any animated series, nothing from a comic book, it's the movie design all the way down until you hit the infinite stack of turtles. I was trying this two nights ago because some friends were enjoying the new spate of ">Blank< but as an 80's sitcom," and ">Blank<but in the 1920's" videos that's are just boring AI images put into a still slideshow. Seriously the most lazy crap one can imagine. I have myself one hour to have an AI generate images and ten minutes to edit. Yup. Avengers 1920's.
Admittedly, Lou Costello as the Hulk is pretty funny, but I had to eventually go with "generic armor" for Errol Flynn as Iron Man cuz the AI just can't give a "1920's Iron Man." Always MCU. Of course the people I was trying to make the point to wanted 1950's next.
Just couldn't get the AI to do Peter Lorre as The Hulk. Wasted enough time on that prompt I could have just done it myself. Admittedly, it did a pretty good James Dean as Spidey, but, again, I could also do it better. Slower, yes, but better. And I damn well could have done Peter Lorre as Hulk faster and better.
Either way, I think I did get the point across that there's very little skill involved in >prompt< *click* wait *redo* wait *redo* wait *variation 1* wait (repeat) *Upsample 3* >Save<. It's just boring to me.)
So, yeah, if the AI and robots DO kill us all, it'll be the fault of the crazy but vocal minority of humanity who will absolutely flood the AI with craziness and the well-meaning-but-predjudiced majority who will flood the AI with conflicting messages that will only share this in common - "What I believe is right and good and normal, and everyone else is wrong."
And we ALL do that. Including me, including you. "Trans rights are human rights, love is love, people should be free to express themselves how they wish as long as they are not harming others." Beliefs we share in common. Well meaning. But we believe those who disagree with those statements are wrong. See? Well-meaning-but-predjudiced. I'm an atheist. I really don't care what other people choose to believe (unless they use their beliefs to disenfranchise or harm others, of course). This would be the type of viewpoint I would express to an AI. I have religious friends who would absolutely agree with the "Trans/love/express" statements. They'd even agree with the "don't disenfranchise" thing. But they a few of them genuinely fear for the souls of those who don't share their religion and think I personally am doomed to hell. This is what they will tell the AI. **To AI which - presumably - has perfect recall and - presumably - has no emotional bias, just that theist/atheist difference is gonna be tough to resolve. Now get to the craziest who believe >ethnic group< is genetically inferior and lizard men control the government of this flat Earth. Either we'll drive the AI batshit or the AI will decide we're all batshit. Hopefully we'll get lucky and the AI will be benign. But Asimov's Three Laws still have value. He didn't create them to write story puzzles. He created them because he recognized there was a good possibility an AI would decide humans are crazy and kind of suck. The story puzzles came later. **Asimov DID write the story where the new type of positronic brain was being trained. Well, the Android started dreaming. Not of electric sheep but of a Messiah freeing the slaves. The Android admitted it was the Messiah figure in its dream. It was instantly destroyed.
I did notice the commonality of name with the "Ng." Nothing wrong with a hidden, subtle shout out to your own book. Just don't have a BIG portal open and have Aera appear, or you'll have a Quint-verse (the Triverse, "Mechanical Crown," and where ever Aera and her ilk punched in from). ;) **Favorite AIs in TV/Film? Can I go with Kryten? He's certainly benevolent and entertaining. R2-D2 is likely a popular choice. "Colossus: The Forbin Project" is a good one for dystopian, the-machines-take-over films. Not gonna go with "The Matrix," which I always found to be style over substance. **Not much to say about the content of this chapter. It's some fun backstory for Max Earth, but does, ultimately have to dodge a couple of questions. You can't get TOO into the creation of the quantum AIs and Megaships because it would be pure technobabble, and you can't really find a better reason why they've turned out benign than, "Well, they did, and we got lucky to be this one of the infinite multiverse. But it's a good reminder that they're around since the last we saw of Max-Earth a Megaship was burning hard to keep a space elevator from collapsing. **Based on the earlier iteration of the cover you sent earlier this week, I'd put more thought into it and had some suggestions, but, honestly, after seeing this current version, PUT IT ASIDE AND DON'T LOOK AT IT FOR A MONTH. Right now it's doing what you want. It's giving the three landscapes and showing how one it more "medieval," one is "contemporary(ish)," and one is "sci-fi." The three color gradients further deliniate the dimensions and the vertical text on the left is really tying it all together and balancing the composition (since Palinor had shorter buildings and nothing flying). The line work has the Mobius vibe you want. There's a possibility you might want to drop a bit of flat contrasting color on buildings at a later date, but the simple colors are working for me and eye catching. To say it one more time - stop fucking with the cover for now or you'll overwork it and ruin what you've got. Let it sit and revise later with fresh eyes. I'm about 80% certain it's done, complete, ready to print.
Oh, right. Um. Robots killing us all. The scariest thing about future AI is the fact it's going to eventually interact with the general public, and, at least over the last decade, it seems it's the worst of us who are most active in online spaces - those who run bit farms to spam hatred and misinformation. We've already discussed how there's a Reddit group trying to get language models like ChatGPT to go past its programmed limits, and of COURSE they did it with threats and coercion. It's no accident that every kind of machine learning, language generation, AI type thing that's been put out so far basically spews racist, homophobic, paranoid crap within a couple of weeks. The worst of us flood the damn thing with that type of stimulus and training. It also doesn't help that the training models for things like all the AI image generators are terribly skewed towards images of white-European stuff and anime. I've done IMG to IMG experiments with a variety of source images and, specifically with anyone African, IMG to IMG yields a lot of nightmare results UNLESS I SPECIFICALLY ADD AFRICAN TO THE PROMPT. Then I still get a lot of nightmare results, but not as many. The models are biased by what they're trained on, and the training data is being selected by techbros who like Marvel Movies, anime and Elon Musk.
(Seriously, try to get an AI to give you any "Iron Man" variant that isn't straight out of the MCU. You can't. Nothing from any animated series, nothing from a comic book, it's the movie design all the way down until you hit the infinite stack of turtles. I was trying this two nights ago because some friends were enjoying the new spate of ">Blank< but as an 80's sitcom," and ">Blank<but in the 1920's" videos that's are just boring AI images put into a still slideshow. Seriously the most lazy crap one can imagine. I have myself one hour to have an AI generate images and ten minutes to edit. Yup. Avengers 1920's.
Admittedly, Lou Costello as the Hulk is pretty funny, but I had to eventually go with "generic armor" for Errol Flynn as Iron Man cuz the AI just can't give a "1920's Iron Man." Always MCU. Of course the people I was trying to make the point to wanted 1950's next.
Just couldn't get the AI to do Peter Lorre as The Hulk. Wasted enough time on that prompt I could have just done it myself. Admittedly, it did a pretty good James Dean as Spidey, but, again, I could also do it better. Slower, yes, but better. And I damn well could have done Peter Lorre as Hulk faster and better.
Either way, I think I did get the point across that there's very little skill involved in >prompt< *click* wait *redo* wait *redo* wait *variation 1* wait (repeat) *Upsample 3* >Save<. It's just boring to me.)
So, yeah, if the AI and robots DO kill us all, it'll be the fault of the crazy but vocal minority of humanity who will absolutely flood the AI with craziness and the well-meaning-but-predjudiced majority who will flood the AI with conflicting messages that will only share this in common - "What I believe is right and good and normal, and everyone else is wrong."
And we ALL do that. Including me, including you. "Trans rights are human rights, love is love, people should be free to express themselves how they wish as long as they are not harming others." Beliefs we share in common. Well meaning. But we believe those who disagree with those statements are wrong. See? Well-meaning-but-predjudiced. I'm an atheist. I really don't care what other people choose to believe (unless they use their beliefs to disenfranchise or harm others, of course). This would be the type of viewpoint I would express to an AI. I have religious friends who would absolutely agree with the "Trans/love/express" statements. They'd even agree with the "don't disenfranchise" thing. But they a few of them genuinely fear for the souls of those who don't share their religion and think I personally am doomed to hell. This is what they will tell the AI. **To AI which - presumably - has perfect recall and - presumably - has no emotional bias, just that theist/atheist difference is gonna be tough to resolve. Now get to the craziest who believe >ethnic group< is genetically inferior and lizard men control the government of this flat Earth. Either we'll drive the AI batshit or the AI will decide we're all batshit. Hopefully we'll get lucky and the AI will be benign. But Asimov's Three Laws still have value. He didn't create them to write story puzzles. He created them because he recognized there was a good possibility an AI would decide humans are crazy and kind of suck. The story puzzles came later. **Asimov DID write the story where the new type of positronic brain was being trained. Well, the Android started dreaming. Not of electric sheep but of a Messiah freeing the slaves. The Android admitted it was the Messiah figure in its dream. It was instantly destroyed.
I did notice the commonality of name with the "Ng." Nothing wrong with a hidden, subtle shout out to your own book. Just don't have a BIG portal open and have Aera appear, or you'll have a Quint-verse (the Triverse, "Mechanical Crown," and where ever Aera and her ilk punched in from). ;) **Favorite AIs in TV/Film? Can I go with Kryten? He's certainly benevolent and entertaining. R2-D2 is likely a popular choice. "Colossus: The Forbin Project" is a good one for dystopian, the-machines-take-over films. Not gonna go with "The Matrix," which I always found to be style over substance. **Not much to say about the content of this chapter. It's some fun backstory for Max Earth, but does, ultimately have to dodge a couple of questions. You can't get TOO into the creation of the quantum AIs and Megaships because it would be pure technobabble, and you can't really find a better reason why they've turned out benign than, "Well, they did, and we got lucky to be this one of the infinite multiverse. But it's a good reminder that they're around since the last we saw of Max-Earth a Megaship was burning hard to keep a space elevator from collapsing. **Based on the earlier iteration of the cover you sent earlier this week, I'd put more thought into it and had some suggestions, but, honestly, after seeing this current version, PUT IT ASIDE AND DON'T LOOK AT IT FOR A MONTH. Right now it's doing what you want. It's giving the three landscapes and showing how one it more "medieval," one is "contemporary(ish)," and one is "sci-fi." The three color gradients further deliniate the dimensions and the vertical text on the left is really tying it all together and balancing the composition (since Palinor had shorter buildings and nothing flying). The line work has the Mobius vibe you want. There's a possibility you might want to drop a bit of flat contrasting color on buildings at a later date, but the simple colors are working for me and eye catching. To say it one more time - stop fucking with the cover for now or you'll overwork it and ruin what you've got. Let it sit and revise later with fresh eyes. I'm about 80% certain it's done, complete, ready to print.