The Great Ethics in AI Debate (Field Report)
"It was much more civil than I imagined...?" - Delaroux
Somewhere on the internet, a bunch of schizos and weeaboos decided to get together to discuss ethics in AI. I thought we would have a lot of fighting, cussing, swearing at SJW post-modern artists, but it was really civil and gentlemanly...?!
tl;dr Some people are optimistic about AI, that it will be a net good for humanity. Some people are afraid that AI will cause us to lose our humanity. It all ended in despair as nobody could tell AI from humanity. (laugh)
LESSON LEARNT: Have a dedicated autist who will record the audio so other autists can enjoy it.
LESSON LEARNT: We should keep this format of free flow of discussion, seeing how everyone is inclined to it.
LESSON LEARNT: Two hours is the maximum optimum time to have an event.
Pre-Preparation of Event
"Shouldn't we have someone record this...?" - André
Pre-event stream started at 2:30, as I fumbled around my Phyrexian-like setup of cables, a very good question was asked: shouldn't someone record this stream/debate for prosperity?
Shamefurry, I did not consider that since I thought we would be too busy shooting the shit at each other, and it was pretty late to change my setup I had been preparing and testing for today.
We should have a FM radio archive one day or clips of the various voice channel meetups, but today was not the day (due to Author-kun's incompetence ;_;), puresu understando.
P.S: I have no idea how to work OBS so that it records audio without interfering with my microphone or audio, I'll have to test some other time.
Oh, and I was streaming Master Duel on the pre-stream and almost got run over by some shitty kid and his Blue Eyes deck. You didn't think that Altergeists had something over 3000 ATK, did you, bratty Kaiba?!
At this point, Murder and André walked into the stream. And we talked about some random video games until the match ended at around 3:00PM.
The Great Ethics in AI Debate, First Hour
As the guests start to stream in after 3:00PM, with La Generalidades del Autismo, Judgement del Mehico, and Dark Jester entering the chat, along with out guest-of-honor, Giga, who wrestled around with his microphone issues.
"Spill some blood from albino children so your mike will work" - Generalmisso of Autismo
"Giga is trying to enter the industrial age, and my lovely state of California is trying to leave it." - André
Initially, As we were going around doing self-introductions, Giga finally fixed his mike and raised his voice:
"So, when are we going to start fighting?" - Giga, paraphrased
Sides? Are there sides?
Well, Giga, in order to start fighting, there needs to be two rival African tribes... I mean two sides to the debate. First off: nobody knows on which side everyone falls on (laugh).
So I thought a good idea was to get everyone to go around introducing what they thought about our new AI overlords, starting with André. This also led to another good question: "What's the scope of this debate/discussion?"
In reply, I mentioned it could be anything about AI, whether it be generative or general use AI.
Perhaps I made the scope too broad in retrospect, since I guess people came in thinking we were talking strictly about AI art, but seeing the direction the talk was going, I thought it might hinder a more free flow of discussion. This turned out to be a good idea, since the discussion went in some pretty unexpected directions.
LESSON LEARNT: We should keep this format of free flow of discussion, seeing how everyone is inclined to it.
Of course, I would be listening in like a wiretapping glowie and helping steer the direction of the discussion if one person goes on for too long (this didn't happen much), or to help get someone to speak if they didn't get a chance before being cut off by someone else. (this happened, a lot) I didn't really get to share much of my opinions since I was juggling being the host, taking notes in my notepad, but I guess nobody was bored enough to watch my livestream of me prompting AI girls. (it was mucho difficult to multitask)
Current Topic: Will AI Take Over All Our Jobs?
"AI is gonna take their jobs, they gonna take ALL our jobs..." - Generalmisso of Autismo
And of course, the great debate opens off with: "THEY'RE GUNNUH TAKE OUH JARHBS" - Circa 2017 Southpark.
Generalmisso of Autismo rightly points out that one day all of us VILL be VEPLACED by the AI, because it is simply cheaper and easier to get a robot to do a job, and if allowed to, people will do this as a default option.
Apparently, in Burgerland, Darkjester mentions there is a type of store called Wawa (sic) where there is zero human interaction at all, the cashiers are self-checkout, you walk up to a LCD panel, order the food, and it's handed to you. It's nice and efficient, if a little creepy from my POV.
"What about energy?" - Murder
The discussion goes to different industries, farming, computer science, etc... Even music is not spared the scrutiny. Some industries will have to change and adapt, but if they are reborn, the jobs will simply be transformed into jobs we never had before (sic André).
In general, the discussion moves further and further away from AI and more into general advancements in technology, but I don't have the heart to pull it back since it seems to be going so well.
"Won't people replace everything with robots because it's more efficient?" - In many variations
On the other hand, there's the "human element", proposed by Mehico (who finally speaks!) that the human touch will always be a luxury, since fine dining establishments still mostly use human waiters. Surprisingly, almost everyone agrees on this one, nobody contests that human contact will be a Veblen Good in the future. (yes yes, many MBA, much wow)
"Sexbots." - DarkJester
Of course it doesn't take long for the discussion to take a more blackpilled turn as we discuss the people who don't want to have human interactions at all. There are always going to be people who prefer their solitude claims both Mehico and DarkJester.
"... because they don't want to interact with people as a whole" - Mehico
There is a bit of a heated discussion as people discuss whether society would REALLY prefer to live in solitude Death Stranding style, or whether it is an aspect of (((modern life))).
Naturally everyone shits on (((modern life))).
At this point, the discussion has been going on for about an hour, and we seem to be going in a small circle of "some people need human interaction", so headed into the second hour, I took the chance to slightly change course.
The Great Ethics in AI Debate, Second Hour
The Absolute State Of Artists
So far we have been talking about AI usage in practical terms, jobs, the service industry, and how they impact real life. I thought it was a good idea to talk about something that might be missed once Skynet takes over:
"Won't artists lose their status if AI becomes prevalent?" - Delaroux
Having spun around in the AI circles for some time, I see a common concern that a lot of artists (forma East Coasters) are going to lose their positions in the societal hierarchy if the thing that they do and worked on for a long time is completely taken over by AI which is faster, more efficient, and just plain better.
A general consensus is that the "status" part of the artist isn't really that important, and a lot of those who base this as their reason to oppose AI have a really weak case. Besides shitting on (((modern art))) in all its art laundering forms, there was also general shitting on hipster artists. Especially in societies that never valued the arts in the first place...
"Art's always been a joke degree in a way." - del Mehico
mfw MBA with double literature degrees.
An interesting inflection came up as Andre was expounding this point, who claims that status doesn't matter since the "real" jobs like doctors will always need a smart human and be held in high regard.
Well, hang on that doesn't sound right--- Those very jobs which are held in high regard right now; bankers, lawyers, doctors, in my opinion are exactly the kind of high status jobs that VILL be replaced by a machine should they prove capable of doing so one day. (they do, ChatGPT can pass the bar exam)
Obviously I am very in favor of lawyers being replaced by robots, just for the principle of it, and for the laughs because ten years of law school only to get beaten by two thousand lines of code is something that I find very funny. Sometimes, it's not about the society, it's about sending a message.
LPD's Gonna Freak
André explains that there will always be a human hand needed, since they aren't just doing routine jobs, but are constantly under pressure and have to make active decisions. They have to cross reference different cases; the law is very subjective; and each case situation is different and should be dealt with differently. The discussion moves to the finer points of the law system, whether lawyers are only good for getting you a plea deal. (they are)
One of the things I found striking in the argument was that it sounded so very much alike the arguments that "AI will never replace artists" a few years ago. Art is also a very human thing, and very subjective; to my mind the very same arguments that are defending lawyers are paper-thin to the point you can see the pantsu of the artists who wore them first. If they are identical in nature, where is the line drawn?
A good point that was raised by DarkJester is that the people would not accept AI OverJudge#5378.
That is, society as a whole will reject AI in law. (not lawyers, they can defend themselves pretty well)
Can you imagine if during a case involving rape, the victim had the solemn judgement delivered by AI OverJudge#5378, who is truthful and impartial at all times that "she was asking for it"? Even if his judgement was correct and cross-referenced a thousand cases in the past and was absolutely, perfect, nobody would stand for it. It seems cruel and unsightly, so it is better for a human to deliver the judgement.
As much as I am inclined towards the Psycho Pass world for the lulz, I would have to agree. It seems to be a common viewpoint that humans still value their human nature and having judgement passed by a robocop is a line crossed.
Perhaps I am too much of an AI extremist here, but I still think that robo-lawyers are coming, whether people like it or not; (note to self to affix this viewpoint to the antagonist of the Arcadia arc.)
Deepfakes frfr
At around the 16:25 mark, Ferretza and his deep, comfy voice enters the discussion as it turned towards the rabbithole of deepfakes: AI videos and soundclips that are so realistic they could be used to do... stuff.
And I do love deepfakes because just in recent days they are such a fun joy, as clips of Biden squaring off and bantzing with Biden in children's card games make the rounds on Tiktok.
On a sidenote, I think that fake clips of famous people are going to be the new meme industry. 4Chan are going to be left behind in the dust once this truly takes off, and it has, because it doesn't do sound nor clips very well.
Memes are about possibilities, funny demotivationals and all that were the height of possibility in the 00s, and we've come a long way here today to BWEEEEHHHHH memes. Now the entire franchise barrier can be crossed and you get wholesome freaky mixups like BoondocksxGenshin where Uncle Ruckus of all people saves Nahida from the Akademia. It's unexpected, ridiculous and absolutely chaotically funny. Very kek energy. Buy ClonedVoice coin? Yes please.
At some point we minorly discuss about the hypothetical if Hillary Clinton was found eating babies, how could we ever charge her in court because we could claim it was a deepfake. And my computer start whirring up because the FBI is watching me type, (it's only a hypothetical, Agent Yelan!) but Frazzledrip exists frfr.
My notetaking also trails off around here because we been at it an hour and a half so I'm kinda tired too. The discussion of deepfakes moves into the discussion of using AI for media here, as we wonder why AI took off so quickly so fast. Ferretza suspects that these few years it has been pushed really heavily.
"If you push something into the mainstream, there will be lots of support for it." - Ferretza (sic)
Finale: The Sonic Image and The Dirty Trick
"The eyebrows are formed way too well for a child." - Whitey, Famous Last Words
As the discussion grew more heated on deepfakes, Whitey enters the chat around 16:40, who is mistaken for Sayo who left chat, but tanned anime girls are all the same amirite? As we are joking around about Frazzledrip, we talk about a image on 4chan that's been making the rounds.
As /v/ laments the usage of AI in art, some Anon posted an innocuous picture on /v/ (I was there!): It's a picture of a Sonic drawn by a child that is praised by posters as having "soul".
Except it isn't...? That particular Sonic image was made by the Midjourney AI, despite it's very childlike lines and coloring. To which the people left in chat (for it was very late) started analyzing the Sonic image in depth. Some people do agree that at a glance it would be difficult to tell.
For Whitey it was a simple matter of the eyebrows. (not the blue hands?) "Those lines, too accurate for an imperial stormchild." among other things. This went on back and forth while Sonic was dissected on a Phyrexian Altar, and I was sitting back and thinking:
That can't be right.
It's extremely difficult to tell AI art from real art, I know, because I make AI art a lot, and a lot of good AI art really looks like it stretches the possibility of human drawing skill to the maximum, and some artists are so damn good that they can be mistaken for the AI. (already happened)
So I decided to play a nasty trick. Well, not a nasty trick, it was meant to be an honest question, but because of time I couldn't generate a custom image on the spot.
As the conversation raged on about the finer details on how to instantly spot AI art, I took two pictures and put them side by side in the text chat as a fun experiment
If I said one was AI art, which one would it be and why? A (above) or B (below)?
For this experiment, I made our local weeb Whitey go last because it seems like the answer is a little too obvious to anyone who has watched a little anime or play gacha.
Why don’t you try it out yourself too?
The room is divided in opinion, André believes that A is the fake, since there are all sorts of weird artifacting and details that don't make sense. Fair, since the weird machine structures in A are... weird. del Generalmisso shares roughly the same opinion, there's something off about that first image that feels very unreal. Also random machines. Also white background is sus.
"If you told me deviant art tranny made bottom one I wouldn't question it."
On the other hand, Mehico seems to think that B is more fake because of the transparent background and the general oddities in the hair. Which is fair. The pink hair is flowing in weird directions that don't make sense.
So the curtain falls on Whitey to reveal the real answer:
It's B. Whitey believes the girl with the gun is absolutely odd. The lines are really jagged and the resolution absolutely fucked. And isn't the gun a little weird? You don't make guns like that. Also because of the various blurs in resolution, it's the absolute nail in the coffin that the artificing is very AI-made.
At this point, there are rumblings of "well both could be AI-made, and we wouldn't be surprised", since both sides have pretty convincing arguments about what's wrong with the picture and Delaroux is a known trickster...
Except.
That.
It's all real.
I was genuinely wa shocked as I revealed the correct answer: Both are real art by a real artist. No AI was involved.
The artist in question is RedJuice, whom anyone a passing knowledge in anime would recognize the distinctive style, and is very famous for the design of (/Guilty "celebrate Holocaust Day with Shuu" Crown/).
I chose the first image mostly because it was conveniently placed on Twitter, and it dated back to 2018, which 100% cannot be attributed to AI. It's also sorta resembles/is Mana from Guilty Crown btw. I initially thought this was too easy because "oh come on, that's Inori or her evil twin".
Now, the question is: was it AI drawn? You do know that RedJuice's style can be replicated by AI and there is literally a LORA that does that, right? Maybe it's because she's drawn kind of weird, or maybe she's an entirely different character, but she definitely gives very "Mana/Inori" vibes, so I wondered why people are sus of her? Wakaranai, Qiqi pouts.
The second image is another somewhat famous character drawn by RedJuice, the T-Doll M82A1. I thought this would be the one absolutely MOGGED on by Whitey as being real, since this default art is directly from the Gacha game Girls Frontline. It also made me sweat a bit when the Barret M82A1 is described as "nobody would make a gun like that", which is fair, and exactly what everyone said about it when it was first produced. It's also one of the most iconic US-made guns in the world though (sweats)...
Last Word
The room falls into deep silence as the cognitive dissonance starts to set in.
Sweat flows from the brows of every participant,
dripping onto Sonic's not-blue hands.
It seems that telling AI Art from Human Art ain't easy.
The silence went on for a while,
just like my hentai game bad endings.
Punctured by the occasional odd sentence,
I look to the clock my oishii Kronii is holding up at my side,
"Well look at the time, it's almost five!"
"And I think that two hours is a pretty long time, everyone seems pretty tired at this point!"
LESSON LEARNT: Two hours is the maximum optimum time to have an event.
With tired, solemn and painful agreement, the Great Ethics in AI debate event comes to a close.
In retrospect, I would have hoped that the event would end in laughter, as I go "SIKE, THAT's RIGHT, IT'S ALL REDJUICE ART!" or "DAMN BRATTY INORI RIPOFFS", but literally nobody considered that both images were human made.
Oh dear, oh dear, what have you done, Inori...?
Still, the two hours were fun, and surprisingly civil; the ending, not-so-ethical, but I'll try to not play pranks like that in the future, but you never know if Madi-san is going to point her psyopping beam at me and make me do something weird again.
I hope you enjoyed reading this field report.
Maybe there will be another event, at a different time in the future.
Written 230225
By Clover-GPT (DE-Modified)
Or not?