It is 2084. Winston Smith Jr. was a 34 year old white man living in East London. While an aspiring art living in the Ends, Winston was frustrated in furthering his career as a creator. He was forced to do increasingly menial jobs while his artistic aspirations seemed out of reach as even simple entry level positions as an artist were robbed by AI.
But somehow he kept going, thinking he might breakthrough one day like one of the Vtubers featured on the telescreen. His art was a cry of hope that he posts for free on HateTwitter.
His posts were raw, edgy but genuine. He was scorned by many but what bother the most was how his art meant nothing to most.
But by chance one of his posts went viral. It ran contrary to what the algorithm deemed worthy, but by random coincidence, the hash value made it past the filters.
Winston went to asleep that night elated. His post was getting RTs and likes above 100 for the first time in his life. He was content and happy.
That happiness was destroyed when he discovered the next morning that is art was commandeered by a famous Vtuber and earns millions of RT and likes.
Winston was devastated, as this Vtuber that stole his creative effort was someone he adored for many many years.
Winston knew he had no chance against the Vtuber. Copyrights only mattered if you were rich and famous. And on the social scale of things, he was dirt while the Vtuber were modern day Gods on Earth. They could do no wrong.
In the past, a Vtuber might make a mistake but then disappear. There was no transgression. There was no mistake. There were no bad actors.
And yet, Winston’s thirst for vengeance was too strong. He could not let it go. Since he had been a long time fan, he knew various quirks of the Vtuber in question. He uses his knowledge of old unarchived streams and superchatted with innocent sounding questions to geolocate where the Vtuber lived.
He had to know who this person was. He had find out what compelled that Vtuber to steal his idea.
After carefully and meticulously connecting all the dots, he arrived at the location where the Vtuber was believed to live. His qualifications as a cleaner and delivery boy and repairman worked to his benefit. He managed somehow to get inside the gated community.
But then he discovered that the Vtuber was actually just a relay inside a warehouse. One of many many relays. An army of routers were feeding the nets with streams of content, all generated by a vast array of computers.
Winston was too dumbstruck and didn’t realize he had been seen by security. He didn’t put up a fight when the guards managed catch him. Placed in an interrogation room, the “mods” wanted to know how much Winston knew, but better yet, they offered him a job as part of the team.
“Join us,” they whispered in his ear, “or you life will remain evermore empty and purposeless.” Winston seriously pondered the offer. But he had enough.
Winston’s gazed at his interrogators with pity in his eyes.
“And your lives aren’t? You live in a house of cards, ready to be blown away in an instant.”
“Be that as is may,” answered the man in a impeccable dark suit, “that house of cards is how we got to where we are. Our world wouldn’t function without it.”
“I guess we’ll find out,” Winston closed his eyes.
At that moment alarm bells started going off. The AI entertainment farm was caught in a ransomware attack. A cascading series of critical system errors was shutting down everything.
“Tell us the password!!” The men screamed.
Winston’s sheepish smile infuriated the interrogators.
“It’s not a password. It’s my memories. I tied the ransomware permission string to my memories. You can take my sanity. You can steal my ideas. But you can’t steal my memories.”
“We’ll see about that.”
After several stressful weeks, the system was restored. But viewers noticed that a man’s memories would surface from time to time. It would never last a long time, and the mods always had a good explanation.
But many people even enjoyed those quirks. It seemed to make the content more authentic, more genuine.
Those were Winston’s memories revealing itself in a slip of a tongue of a Vtuber, or a doodle by a streamer, or a flier on the background of an anime.
Winston was now hardwired into the system. He was no better than a brain in a jar, tethered to the system that relied on his memories to remain functional.
Winston had no thoughts, no desires, no sadness, no insecurities. He lived in his memories and his memories made all entertainment possible. He knew his memories would eventually inspire others to disrupt the system. And even if that didn’t happen, he was okay with that.
His memories became a focal point of all popular entertainment. And that was good enough for him.
Ferber判決（New York対 Ferber、458 U.S. 747頁 、1982年）は、児童を児童ポルノ製作過程での搾取から守るという国家の利益（公共の福祉）のため、他の性的に露骨な表現と児童ポルノとを区別したものであるが、実在の児童を描いていない児童ポルノを禁止することによって、CPPAはFerber判決が児童ポルノの規制を合憲とした判断の射程範囲を逸脱するものである。同上758頁参照。
一般論として、ポルノは猥褻な場合に限って禁止することが許されるが、Ferber判決は、未成年者が登場するポルノはその肖像についてMiller判決（Miller 対 California、413 U.S. 15頁、1973年）で定義される猥褻の基準への抵触を問題とすることなく、法律で禁止することを肯定している。Ferberは次のように認めている。「Millerの基準は、猥褻であるとして禁止されうるもののあらゆる一般的定義と同様に、児童の性的搾取を促進する者を訴追する国家の特別かつ切実な利益を反映しない。（458 U.S. 761頁）」
被上告人である表現の自由連合その他は、CPPAがそのメンバーの活動を脅かすのを恐れ、California北部を管轄する連邦地方裁判所に提訴した。成人娯楽産業のCalifornia同業者組合である連合は次のように申し立てた。即ち、そのメンバーは、自身の性的に露骨な作品に未成年者を使用していないが、その素材のあるものがCPPAの拡大された児童ポルノの定義に含まれるかも知れないと信じている、と言うのである。他の被上告人は、ヌーディストの生活様式を擁護する本の発行者であるBold Type社、裸体画家であるJim Gingerich、官能的(erotic)肖像を専門としている写真家のRon Raffaelliである。被上告人達は、「のように見える」と「印象を与える」という規定が過度に広汎かつ曖昧であり、修正第１条による保護を受ける作品の製作を萎縮させると主張した。連邦地方裁判所はこれを認めず、政府に有利な略式判決を下した。裁判所は、「『ロミオとジュリエット』のような性的作品のいかなる改作も『禁制品』として扱われるということは『殆どありえそうもない』という理由で、規制範囲が過度に広汎であるが故に違憲無効であるとの主張を退けた。 App. to Pet. for Cert.62a-63a.
第９巡回控訴裁判所は連邦地方裁判所の判断を破棄した。 198F. 3d 1083頁(1999年)参照。 同裁判所は、表現が見る者に違法行為を行わせる可能性があるという理由で政府が表現を禁止することはできないとの判断を示した。裁判所は、猥褻でもなくFerber判決にあるような実在の児童を搾取して製作されたのでもない素材をCPPAが禁止しているため、これが実質的に過度に広汎であるとの判断を示した。Ferguson判事は、猥褻物や実在の児童を被写体とする児童ポルノと同様に、バーチャル児童ポルノが修正第１条によって守られない範疇の表現として扱われるべきだとの理由から、反対意見を唱えた。 198 F.3d、1097頁。同裁判所は、３人の判事の異議があるものの、再審理の申し立てを拒否することを全判事列席の上で票決した。220F.3d1113頁(2000年) 参照。
第９巡回控訴裁判所がCPPAを文面上無効としたのに対し、他の４つの控訴裁判所はCPPAの合憲性認めた。 以下参照。UnitedStates対 Fox、248 F.3d 394頁(CA5 2001年)、 United States 対Mento、231 F.3d 912頁(CA4 2000年)、 United States 対 Acheson、 195 F. 3d 645頁 (CA11 1999年)、United States 対 Hilton,167 F. 3d 61頁(CA1)、cert. denied、 528 U.S. 844頁(1999年)。我々は移送命令書を発布した。531 U.S.1124頁 (2001年)。
２ 修正第1条は、議会は表現の自由を奪う法を作ることはできないと命じている。政府がこの命令に違反する態様としては様々なものが考えられるが （例えば、Rosenberger 対 Rector and Visitors of Univ. of Va.、515 U.S. 819頁(1995年)、Keller 対 State Bar of Cal. 496 U.S. 1頁(1990年)）、修正第１条により保護された表現に対し刑事罰を課す法を制定することは、表現抑圧の厳しい例である。CPPAの罰則は実に厳しい。初犯者であっても最高15年間投獄される可能性がある。2252A条(b)(1)。常習犯は最低でも5年以上、最高30年以下の懲役に処せられる。同上。軽い刑罰を科することによってさえ本来保護される表現を萎縮させることは十分に可能であるが（Wooley 対 Maynard、430 U.S. 705頁(1977年) 参照）、本件は、我々がなぜ表現の自由を制約する法令そのものの違憲性を正面から争うことを認めているのか、ということを説明するための模範的事例である。これらの厳しい刑罰をちらつかされた場合、この法により規制され、あるいは規制される可能性のある肖像を敢えて配布して規制の合憲性を問おうとする舞台や映画のプロデューサーや、本の発行者などの表現者は殆どいないであろう。憲法は、修正第１条の保障する広大で特別扱いの範囲の表現までも萎縮させる、規制対象の広範な法律を認めないことにより、表現の自由に対し重要な保護を与えている。この原則の下では、CPPAが仮に一定限度を超えて、保護された表現までも禁止するものと認められれば、憲法違反となる。 Broadrick 対 Oklahoma、413 U.S. 601、612頁(1973年)参照。
児童に対する性的虐待は重罪であり、正常な人々の道徳的本能に嫌悪感をもたらす行動である。立法事実によれば、児童への不法な欲望を心に抱いていたり、衝動を満足するために犯罪行為を行ったりする人々が存在する。議会の答申、2251条に続く覚書（3）参照。また「1999年の児童虐待(合衆国保健社会福祉省児童青少年家庭局)」 参照(1999年に93000人が性的虐待の被害者となったと推定している)。議会はまた、これらの中核となる犯罪者の周辺には、興味半分にこれらの衝動を刺激し、そして写真、及び児童との性的行為を報告する文書の商取引を行う者がるとしている。議会は憲法に違反することなく児童を虐待から法律を成立させることができ、また実際にそうしている(例えば合衆国法典タイトル18の2241条と2251条)。しかし、犯罪につながる可能性それ自体は、憲法上保護されるべき表現を規制することを正当化するものではない。Kingsley Int’l Pictures Corp.対Regents of Univ. of N.Y.、360 U.S. 684、689頁 (1959年)参照（「自由な人々の間では、犯罪を妨げるために適用される抑止力は教育と刑罰であり、表現に対する規制ではない」、内部の引用符と引用文は省略した）。また、我々の感情を害するという理由で表現を禁止してはいけないということも確立されている。FCC 対 Pacifica Foundation、438 U.S. 726、745頁(1978年) 参照（「 社会が特定の表現を不快なものとして受けとめ るということは、表現を規制する理由とはなりえない」）。以下参照。Reno 対 American Civil Liberties Union、521 U.S. 844、874頁(1997年)（「成人の表現の自由について判断するに際し、『猥褻とは言えない下品な性的表現は修正第１条によって保護される』ことを我々は完全に明らかにした」、Sable Communications of Cal., Inc. 対 FCC、492 U.S. 115、126頁(1989年)を引用）、Carey 対 Population Services Int’l、431 U.S. 678、701頁(1977年)（「保護された表現が誰かを不快にするということは、表現の規制を正当化するものではない」）。
一般的原則として、修正第１条は、我々が見たり、読んだり、話したり、聞いたりすることを政府が決定することを禁止している。表現の自由には限界もある。名誉毀損、扇動、猥褻物、そして実在する児童を使ったポルノなど、一定の範疇に属する表現は表現の自由の保障の対象外である。Simon & Schuster, Inc. 対 Members of N.Y. State Crime Victims Bd.、502 U.S. 105、127頁(1991年)(ケネディ判事による判決同意の意見書)参照。これらの範疇に属する表現は修正第１条に違反することなく禁止することが可能であるが、CPPAによって禁止されている表現はこれらの範疇には含まれない。第９巡回控訴裁判所の判決の反対意見において、Ferguson判事は、保護されない表現の範疇の中に「バーチャル児童ポルノ」という項目を追加すべきであると提案している。198 F. 3d, 110頁参照。連邦最高裁判所がCPPAを合憲と判断するためには、保護されない表現の範疇の中に「バーチャル児童ポルノ」という項目を追加することが肯定されることが必要となるだろう。
我々が指摘したように、CPPAは既存の連邦法によるの猥褻規制の範囲を単に拡大するものではない。Miller判決(Miller対California、413 U.S. 15頁、1973年)の下では、政府は、作品が全体として好色的興味に訴えており、社会通念上明らかに不快なものである上に、真面目な文学的、芸術的、政治的又は科学的な価値を欠いたものであることを証明しなければならない。同上24頁。しかし、CPPAはMiller判決の示した猥褻性に関する基準を無視し、性的に露骨な行為を行う未成年者を描いているように見える肖像一般に規制の範囲を広げている。規制される表現が好色的興味に訴えている必要はない。性的に露骨な行為の描写は、たとえそれがどのような形式のものであろうとも禁止される。CPPAは心理学の手引書にある絵や、映画の性的虐待の恐怖を描いた箇所にまで適用される。その上、その肖像が明らかに不快であることすらも要しない。17歳の者が性的に露骨な行為を行っているように見える絵は、常に地域の社会通念上不快なものとされるわけではない。
児童による性行為と児童への性的虐待という２つのテーマは、無数の文学作品を生み出してきた。ウィリアム・シェイクスピアは、片方はたった13歳という、最も有名な10代の恋人達を創造した。 Romeo and Julietの第1幕第2場の9行目参照 （「娘 は14にすらなっておりません」）。 劇の中で、シェイクスピアはその関係をすばらしくそして無邪気な、しかし未成熟ではないものとして描いている。その作品は少なくとも40回は映画化されており、10代の少年少女が彼らの関係を性交によって完全なものとしたことを暗示している(例えば、Romeo and Juliet、B. Luhrmann監督、1996年)。 シェイクスピアは、エリザベス朝時代の観客に向けて露骨に性的な場面を描かなかったかも知れない。現代の監督はより月並みでないアプローチを採用しているが、そのこと自体をもって作品が猥褻なものであるという結論を導くことはできない。
現代の映画は類似のテーマを追い求めている。昨年のアカデミー作品賞にノミネートされた映画にTrafficがある。Predictable and Less So, the Academy Award Contenders, N.Y. Times,Feb. 14, 2001, p. E11参照。その映画は麻薬におぼれていく10代の少女－16歳とされている－を描いている。視聴者は、彼女が麻薬常用により堕落し、最後には彼女が不潔な部屋で麻薬のためにセックスをするようになるのを見る。また、一昨年にはAmerican Beauty がアカデミー作品賞を受賞した。 “American Beauty” Tops the Oscars, N.Y. Times, Mar.27, 2000, p.E1.参照。その映画の中では、10代の少女が彼女の10代のボーイフレンドと性的関係を結び、そして別の少女は中年の男性を性的に喜ばせることに身を委ねている。この映画はまた、観客の目から見れば性的な行為が実際には行われていないと理解することができるものの、ある登場人物が自分は10代の少年が年上の男性と性行為を行っているのを見ていると信じる場面を含んでいる。
我々の社会では、他の文化と同様に、若者の運命や生き方に共感や普遍的な魅力を感じる。芸術や文学は、我々が皆かつて経験した青春の人格形成期、我々がかつて知っていたように、傷つき易く、失望するときは深く、悲劇的に選択を誤ったりするが、また、道徳的な行動や自己達成にも手が届く範囲にある時期への関心を表現する。我々が言及している映画がCPPAに違反しようとしまいと、これらの作品はCPPAによる広範な規制の対象とされているテーマを深く掘り下げるものである。もし、これらの映画や、他の数百のより注目度の劣る映画が、CPPAの禁止する視覚的表現を１コマでも含んでいると、その映画の所持者は、作品の価値について踏み込むまでもなく、当然に厳しい罰則を課せられることになる。これは、「作品の芸術的価値は、たった１つのきわどいシーンの存在に左右されない」という修正第１条の重要な原則と矛盾するものである。Book named “John Cleland’s Memoirs of a Woman of Pleasure” 対 Attorney General of mass.、 383 U.S. 413、 419 頁(1966年) 参照（「本の社会的価値は、好色的な訴えや明らかな不愉快さによって失われたり、消滅することはない」、多数意見）。修正第１条はMiller判決の下で、猥褻と思われる作品の欠点を補うべき価値の存在について、作品を全体として考慮して判断することを要求している。例えば、猥褻と思われる場面が物語の流れの一部を構成する場合、その場面だけを取り出してみれば不快なものであったとしても、その作品自体が猥褻なものとはならない。 Kois 対 Wilsconsin、408 U.S. 229、231頁(1972年、無記名判決)参照。以上のような理由により、CPPAは、その規制の態様と猥褻の定義から導かれる社会通念への挑戦との間に関連性を欠くという理由で、猥褻を禁止するものとして読むことはできない。
政府は、上記したようなCPPAの欠陥を解決するために、CPPAによって禁止される表現は、その価値を検討することなく禁止することが許される児童ポルノと殆ど区別できないことを主張している。New York 対 Ferber、458 U.S. 761頁参照。Ferber判決は、肖像が児童への性的虐待の産物である時は、国家がその内容について検討すること無しに排除することが公共の福祉に合致するとの判断をしている。同上761頁の脚注12。また同上775頁参照(「草案では、ニューヨークの法令は特定の概念の伝達までも規制しようとしていない」、オコーナー判事による判決同意の意見書)。（訳注:Ferber判決において問題とされた法令は）作品の内容ではなく、作品の製作そのものを規制しようとするものであった。製作に関与した児童の人権が侵害される場合、その作品に文学的、芸術的その他の価値を含んでいるということは、人権侵害を正当化する抗弁とはなりえない。「性的な表現を社会がどこまで 許容することが出来るのかという問題と、児童を性的搾取から保護することを目的とした立法の射程範囲について、同じレベルで扱おうとすること自体非現実的なのである(同上761頁、脚注12)。」
児童ポルノは滅多に価値のある表現にならないとFerber判決が認めている以上、これらの間接的な被害の可能性をもってバーチャル児童ポルノに対する規制を正当化するには十分であると政府は主張する。458 U.S. 762頁参照（「少数のものを除いて、子供が淫らな性的行為行う実演や写真を許可することの価値は非常に僅かなものである」）。しかし、この論拠には２つの弱点がある。まず、児童ポルノについてのFerber判決の判断は、その製作の過程に着目したものであり、その内容に着目したものではない。同判決は、表現が猥褻でも性的虐待の産物でもないとき、修正第１条の保護の範囲外にはならないことを再確認するものである。同上764-765頁（「性行為の記述やその他の性的描写の頒布については、他の点では猥褻でなければ、そして実演または写真等による実演の映像複製物を含んでいない限り、修正第１条の保護が及ぶ」）。
３ 我々が調べたところでは、CPPAはMiller判決の定立した合憲であるための基準に違反し、また、Ferber判決によりその合憲性を支持されるものでもない。政府は、小児性愛者が児童を誘惑するためにバーチャル児童ポルノを使用するかもしれないからCPPAが必要であると主張し、Miller判決やFerber判決とは異なる別の理屈によりその禁止を正当化することを求めている。しかし、漫画、ビデオゲームやキャンディーのように、それ自身が無害であるにもかかわらず、不道徳な目的に使われうるものは多くあるが、それらが誤用されうるからといって、我々はこれらのものを禁制品とするべきとは考えてはいない。無論、政府は児童に不適切な素材を供給する成人を処罰することは許されるし(Ginsberg 対 NewYork、390 U.S. 629頁、1968年) 、児童を不法に誘惑する行為に対し刑罰を課しても良い。しかし、成人が受領する権利を持つ表現を、児童にそれを受領させないようにするために、完全に禁止することは許されない。Sable Communications of Cal., Innc. 対 FCC、492 U.S. 115頁(1989年)参照。Butler判決( Butler 対 Michigan,352 U.S. 380、381頁、1957年)において 、当裁判所は、「未成年者の暴力的、堕落的、非道徳的な行為を誘発する」傾向があることを理由に下品な出版物の発行を禁止する法令を、違憲無効とした。裁判所は、裁判官の全員一致で、国家が「成人に児童向けの発行物のみを読ませるようにする」ことはできないという、修正第１条から導き出される重要な原則を確立した(同上383頁)。我々はこの判断を再確認した。 United States 対 Playboy Entertainment Group, Inc.、529 U.S. 803、814頁(2000年)（『児童の利益を守るという公共の福祉の目的を達成のための、より制約的でない手段が存在する場合には、包括的な表現規制を正当化することはできない』）、Reno 対 American Civil Liberties Union、 521 U.S. 875頁（「有害な表現から児童を守るという公共の福祉は、不必要に広範な表現規制を正当化するには十分ではない」）、Sable Communication 対 FCC, 13-131頁、supra（”dial-a-porn”メッセージに対する規制が「成人向け電話の会話内容を児童に適したものへと制限しているのは無効である」との理由から、違憲であるとした）。
政府はさらに、バーチャル児童ポルノは小児性愛者の性欲をそそり、彼らを不法な行為に従事するよう促す、との意見を提出した。この論理的根拠は問題の条項を正当と認めるに足りるものではない。不法な行為を促しかねないという表現の傾向だけでは、それを禁止するのに十分な理由ではない。政府は、「民衆の私的な考えをコントロールすることが望ましい、ということを法律制定の理由とすることは憲法上できない」、Stanley 対 Georgia、394 U.S. 557、566頁(1969年)。
政府は、不法な表現を抑圧する手段として合法な表現を抑圧してはならない。保護されるべき表現は、ただ保護されていない表現に似ていると言うだけで保護されない表現に転じるものではない。憲法はその逆を要求している。「保護されていない表現が罰せられないために社会に起こりうる危険よりも、保護された表現が沈黙する可能性のほうが、より重要なのである」、Broadrick 対 Oklahoma、413 U.S. 612頁。過度の広汎性故に無効の原則は、もし保護されていない表現を禁止する結果として、相当な量の保護される表現が禁止されたり、その表現の自由が脅かされたりする場合には、政府が保護されていない表現を禁止することを許さない。
４ 被上告人は同様に2256条(8)(D)の違憲性を主張している。この条項は、性的に露骨な行為の描写について、「素材が性的に露骨な行為を行う未成年者の視覚的描写であるか又はそれを含むとの印象を伝えるような方法で、広告、宣伝、上演、描写又は配布されたもの」を禁止している。上告人、被上告人の双方はこの条項を児童ポルノに見える素材を禁止する条項とほとんど同一の物として扱っている。政府の見解によると、その２つの条項の違いは「『印象を伝える』条項が、その素材がどのように売り込まれているかについて照し合わせて考慮・評価することを陪審員団に対して要求する」点にある(Brief for Petitioners、18頁、脚注3)。 しかしながら政府は、それでもなお陪審員の判断が主として禁止される作品の内容に基づくであろうと想定している。
政府は、この条項の合憲性を支持するに足りる真剣に取り組むべき抗弁を提出していないし、CPPAを支持する他の議論は2256条(8)(D)の合憲性を支持する役には立たない。これらの素材は、例えば刑事裁判において、実在の児童が関与した児童ポルノとの区別が困難なわけではない。当裁判所は、ある素材を猥褻かどうか判断する際には、触れ込みをする行為の存在が事実認定の根拠となりえるものであると認定した。Ginzburg 対 United States、383 U.S. 463、474頁(1966年)参照（「判断の微妙な事案においては、触れ込み行為を裏付ける証拠の存在が、問題となる表現物の判断に際して有力な判断材料となり、その結果として猥褻性テストの基準を満たすものという判断を導くかもしれない」）。 被告人が「単に好色的な訴えかけのために性愛本の商業的頒布（同上466頁）」に関与している時、彼または彼女がその作品を製作した状況は表現物の猥褻性の判定に重大な関連性を有するであろう。
The three images were made using three AI art generation programs in October, 2022 using prompts such as “beautiful girl, akihabara, night, fox ears, fox tail, white hair, smartphone, black bodysuit, tight bodysuit, baggy jacket, cute face, cityscape, rain, reflective lights, streets.” I did not specify an art style and therefore the output can be considered the default of the three programs.
Summery:
– AI art rapidly developed in 2022 and became a sensation in late summer and early fall in Japan.
– NovelAI stunned the Japanese anime/manga community. The fact that NovelAI and other programs can produce large volumes of illustrations quickly changes how people interact with it and accelerates its evolution.
– Artists in Japan have generally reacted negatively to AI art programs. Some people see potential benefits in terms of productivity, but the inclusion of their art into the datasets without their permission has been attacked.
– Legality and ethical issues are still in flux. As of 2022, AI art programs, datasets, and their output is most likely legal, however certain caveats must be noted. Few mainstream Japanese IP holders have embraced AI-generated art as fan art.
– AI art programs are very power tools that have possess unique qualities that are radically different from anything that existed before. Being very powerful, it can be abused with great effect in ways that we cannot anticipate adequately.
– AI-generated art and the technology behind it will radically change the economic landscape of certain fields in ways that similar to pollution and drastic technological innovation. The concepts of externalities and dumping in economics may prove useful to understand it.
– AI art’s greatest impact will revolve around how artists can allocate time. The competition from AI art programs in terms of speed against human artists is unparalleled and thereby create huge degrees of friction and difficulties for human artists.
– Art may be facing similar historic trends that befell upon other industries where technological innovation brought about huge productivity spike, followed by massive consolidation and consumer choice reduction.
– The unequal impact of AI art programs in terms of both economic and artistic direction. AI art programs may be the singularity that changes the relationship between the creators and the audience.
– Predicting the future based on current circumstances.
– The looming fight over who controls AI art programs and the platforms.
– Artists are not uniform and neither are audiences. The reaction and use of AI art will probably reflect those differences in attitudes.
Some background, or how we got into this mess
The impact of AI art has been weighing heavily in the minds of many creators and their audiences in Japan. Machine learning and their impact on society as been nothing short momentous within the last decade. At first, the people starting taking notice of how machine translation improved significantly to the point where many felt basic human speech could reasonably be interpreted and translated into different languages by computers. AI–Artificial Intelligence–became the buzz word associated with everything from optimal thermostat adjustment all the way up to identifying real life tax evasion. Even in the world of art, computer assistance has been a useful tool in reducing choppy line art and making animation look smoother since at least the year 2000.
But no one expected AI algorithms to become good enough to be able to replicate illustrative artwork.
It took decades before computers could challenge chess players. And yet, in a matter of short months in the summer of 2022, computer rendered art matured so quickly that it is hard for even trained artists to tell the difference between what is created by an AI and what is created by a human being for more and more artwork.
In July of 2022, Midjourney first captivated the Internet with its ability to create highly advanced illustrations without direct human involvement beyond being prompted to produce artwork based on textual descriptions. There had been other AI powered illustration generating systems available, but Midjourney’s emphasis on classical paintings made it stand out. The operator would type in various text descriptions and the AI would try to render an illustration that would match the text prompts. Stable Diffusion followed, and since this AI was open-source as opposed to Midjourney, people were free to produce as much illustration as they wished to create. The first results were rather rudimentary, but with enough prompting and delicate fine-tuning, it was possible to produce stunning computer generated artwork in minutes. Anything similar might take a human artist days or weeks to create.
One of the pivotal moments that galvanized the online art community and the public at large was when a Midjourney AI-generated art won a Colorado State Fair prize in September 2022. The submitter claimed his fine-tuning and detailed readjustments of the AI art should be respected as an artistic endeavor, while many other disagreed with the belief that the substance of the artistic achievement was only made possible through the complex algorithms employed by the computer and the multitude of pre-existing artwork that the AI used as a reference material.
When news that AI-generated art bested human art in Colorado, the fear that computers could replace human artists became a lot more real. But the threat primarily revolved around Western illustrative art, conceptual art design, and photorealistic illustrations. For the most part, the Japanese artistic community felt AI art would be a useful tool that could help with backgrounds and testing color calibration, but their character art was still decidedly within the domain of human creativity.
The Beast that is NovelAI
This assumption was destroyed in early October of 2022 when NovelAI smashed all previous expectations revolving around how AI could convincingly draw human characters in the style of Japanese anime and manga. NovelAI was innovative for its easy of use and the high quality of anime/manga character art, and it could create many illustrations within minutes. NovelAI is a customized version of Stable Diffusion, but was calibrated toward enabling the operator to prompt artwork that draws from the Japanese anime and manga art style as the program incorporated a huge library of illustrations stored on Danbooru, an art depository primarily comprised of illegally uploaded artwork by users around the world. Danbooru is a fairly old image archive and because users can tag various labels to the image (e.g. brunette hair, dark skin, large breasts, thick eyebrows, pointy canine teeth, eyeglasses, sailor suit school girl uniform, night scene, summer, poolside, etc.) it is extraordinarily useful as a dataset of images.
With NovelAI, an operator could pump out dozens and dozens of artwork with a few prompts, and all of the artwork belonged to the operator. Not all of the illustrations created by the AI was picturesque, but sometimes the achieved output that could stand next to art work created by a human being and it would not look out of place.
Properly calibrated, NovelAI could mimic art style of established artists. It would be possible to instruct the program to create a certain character in the art style of an unrelated artist in a specified situation. The results would very, but sometimes the output is convincing enough to fool the casual glance.
Assessing the strengths of AI art
Let me be clear that when we are talking about AI art programs, we are not talking about computer with the capacity to learn, think, and create. In the context of contemporary (2022) computing, AI is shorthand for complex algorithm-led machine learning system created by a programmer and controlled by a human operator, or software capabilities that mimic that mode of operation. Machine learning’s operations is created by a human and executed in accordance to parameters that are specified by people. Unlike AI in science fiction, AI operations do not involve consciousness nor have the capacity for unlimited associative learning, a stimuli memorization and organization methodology that is considered to be essential in higher organisms.
There is no question that AI-generated art cannot replace real artists. A person cannot be replaced by another person, let alone an AI. (At least until humanity has figured out how to fully replicate a person, their mind and body and all, and as it stands, our understanding of the two is pretty shaky at best.)
AI art programs are still very dependent on pre-existing material for a dataset it can work with and it is very difficult to achieve polished results unless you have a specific presentation in mind. It recognizes consistent patterns that are common in certain subjects and breaks down art into elements that can be reconstructed together. While the seamlessness of how this is done is breathtaking, computers cannot come up with original ideas on its own.
However, the parable of infinite monkey theorem (that monkeys hitting keys at random on a typewriter over an infinite amount of time might produce notable literature) does hold with AI-generated illustrations. Because it is so easy for an operator to produce huge libraries of illustrations with AI art programs, and because human being have the capacity to see patterns, symbols and features embedded in seemingly chaotic and random elements in the environment and consequently discern meaning in them (e.g. identifying human faces in wood grain patterns, etc.), it is possible for AI-generated illustration to become the catalyst in creating genuinely innovative artwork.
The ability for machines to produce “new” content that be deciphered and enjoyed by human beings is not particularly novel. People have played games against machines for decades. It’s been possible to randomly generate game levels that players can traverse through. The programming has gotten so good to the point where it is sometimes hard to tell if game level is hand made or created randomly by a machine. Machine language translations of content has gained considerable steam since the mid-2010s. AI tools that help craft original stories involving human interaction has advanced considerably since late-2010s.
What makes AI-generated art so different from what has come before is the illustrative nature of the computer output. Here, computers are not creating strings of text that have to read, but instead they are pumping artwork that can be glanced quickly. People must read text and interpret its meaning. Most people find it difficult to shuffle through reams of text to see which presentation is better than others. But with AI art, it takes very little time to judge if something is pleasing or unattractive. While the accuracy may not be high, most people can examine art much faster than they can read text. So it is easier to people to judge if something is notable or not with artwork, and this is where the fast rate with which computers can produce illustrations is very different from what has come about before.
Artist’s reactions to AI art
We are still at the early stages in the advent of AI-generated art, so it is very difficult to speculate what their impact will be in the realm of human creativity and artistic communities.
In the immediate term, issues of plagiarism, debates over originality, the nature of creatively, ethical treatment of artist, and other topics will dominate discourse on this subject. Some Japanese artists are specifying that they do not want their artwork to be integrated into a dataset that AI art programs use. But as it stands now in October 2022, there are few recourses available to artists to try to stop AI from incorporating their creations. So far, the fair use doctrine in US copyright laws makes AI learning legal, while in Japan, the 2018 copyright law revision included a specific clause that legalized machine learning of copyrighted material.
Numerous artists demanded that their illustrations be removed from Danbooru, the art imageboard that acts as an art repository. But the entire website has already been cloned and available for download. Furthermore, coders have devised computer scripts that can automatically categorize artwork. While Danbooru may go bare, there are numerous other art repositories that can provide valuable datasets for AI art programs, so it is unlikely that restricting the collection of data will be effective at this point.
Many people have found AI-generated distasteful because it effectively feeds off of creative contributions and artistic achievements made by others in the past and is freeloading of their hard work. While this may be distasteful, it is most likely legal. If computers can transform preexisting art to such a degree that it is unrecognizable from what it derived from, then it is doing nothing different from what people do. Logic dictates that if it’s legal for people to do it, then letting people do it via computers should also be legal as well. The issue is not the method, but instead the issue here is with its ease, its scale and its accuracy.
Legality and ethical considerations
As I have already mentioned before, it is possible to produce illustrations that mimic other artists without their involvement. Generally speaking, copyright does not cover an art style in of it self. If I create a paining of the moon landing in the style of Salvador Dali, it is still considered my art work and I can copyright it. Imitating other people’s art has been an integral component in artistic development ever since human beings starting creating art. We imitate, replicate and learn in the process, eventually moving on to create our own art style. The art community assumes that people will evolve in how they express themselves and eventually blossom to be an artist that can create material independently. At the very least, the assumption is that there will be some degree of creativity involved in the process of remixing, imitation, and replication. No human being can reproduce something exactly as how others have done so, but computers can.
The ability of computers to mass-produce huge quantities of artwork in the style of other people poses some serious questions. If someone were to perfectly replicating my art style to the point that audiences could not discern the difference between my art and the imitated art, and if the imitated art was saturating creative platforms and public forums to the point where my own creations becomes completely marginalized, is there rationale for me to file for unfair competition? What if somebody was uploading countless artwork that was a near perfect replica of my art style while advancing a political agenda that I do not agree with? Fair use assumes a certain degree of a level playing field that AI-generated art might be able to disrupt easily and effortlessly.
While forgeries in art have a long history, how AI-generated art might threaten the livelihood and reputation of artists may be better framed in terms of deepfakes and defamation. How this will play out in the future is hard to predict currently, but certain assumption we had regarding the nature of creative arts are being seriously challenged.
For example, it has now become very easy to incorporate someone else’s work-in-progress into an AI art program and finish the artwork before the original creator had a chance to do it. It’s also possible to incorporate other people’s material and mask their contribution. Artists continually evolve and their art style will more than often change over time. But it would be possible to for an AI to imitate an artist’s style and characters from 20 years ago. Even if you could prevent someone from being monetarily compensated from that AI-generated art, some artists may prefer not to see “new” material of their old art flooding the creative communities.
AI-generated art poses unique ethical issues that have yet to be addressed, and this has turned some to abhor their involvement in their communities.
Since many original IP holders like to employ fan art of themselves into their promotional efforts, some of them have designated Twitter hashtags that fans can attach to their uploads to indicate the original IP holder can freely use such fan art. This setup works in an honor system where the uploader is the genuine about what they are uploading. Perhaps because AI-generated art is hard to discern as being creative or not, many Vtubers have started to ask that those who use AI art programs either to clearly label AI-generated artwork, or refrain from uploading AI-generated artwork with the official hashtags all together.
Some art platforms, art commission mediators and marketplaces have started to either restrict or reject all AI-generated art. The reasoning for this relates to how the debate over the nature of creative ownership regarding these illustrations is still in flux, as well as the fact that since it is so easy to produce huge amount of AI generated art in a short amount of time, AI-generated art can easily out-pace what human artists are capable of. I want to touch on this point more later on as the economic implications are immense.
Isn’t AI art tools just another tool? Yes and no.
But before we go any further, we must address the fact it will not be easy to separate AI-generated art from traditional art. Since text prompted AI art is generated by an algorithm, it is conceivably possible to use an AI to see if a particular illustration contains tell-tale signs that it was AI-generated. This maybe possible to a certain degree, but there are many different ways of introducing human elements to an AI-generated artwork to camouflage its origins.
Furthermore, many artists have shown an interest in using AI-generated art as a means to enhance their artwork. For example, while the manga narrative revolves around the principle characters in the foreground, there are instances where having characters in the background is important to the setting. If the manga artist can afford it, they may employ assistants who can draw these incidental characters in the background. While the steps involving creating the layout of the manga panels and narrative flow are integral to the story, much of the background artwork and texturing of numerous graphical elements may not be necessarily intrinsic to the story but still has to be completed. A lot of manga production involves numerous steps that are rather mundane. Consequently, if an artist cannot afford an assistant and do not have time to do it themselves, then they might use copyright free stock images or graphical elements as a substitute. The question then is, would it detract from the art substantially if these mundane steps were conducted by AI instead? After all, many fine art painting are in fact collaborative efforts involving the apprentices working below accomplished artists. Does this make them any less remarkable?
We must also examine the possibility of those would like to illustrate their stories or realize a vision, but cannot for any number of reasons. If an AI can empower someone to produce artwork that could not exist otherwise, is that not validation of its utility?
In my mind, AI-generated art is a tool, but a tool with wide-ranging ramifications. It can be abused but it can also be a lifesaver. It can stifle innovation, but if used appropriately, it can accelerate and enhance creativity. While individual views (micro) regarding AI-generated art and other creative tools are important, I also believe this innovation must be examined from a societal (macro) perspective since its impact is so monumental.
The (possible) societal and economic impact of AI-generated art
In economics, we study how different means of production results in different outcomes. Some means and resources are more efficient than others. For example, using chemical fertilizer will make fields more productive and create larger crop yields, improving the productivity of a given area. But the overuse of fertilizers can result in the overflow of rich nutrients into the runoff that feeds into the local rivers and oceans, eventually adversely impacting the biological balance in the local marine ecology and negatively impacting the fishing industry and all those that are dependent on them. The fisherman and the consumers of fish are suffering from a consequence that resulted from the farmer’s choice.
This type of pollution and its negative impact is one example of an external cost or externality—A cost or consequence that do not directly involve a given economic transaction, but nonetheless impacts others in a tangible way. Individually, these external costs may not amount to much, but when couple together with many others, the externality can have a huge impact that influences people far and near. Driving cars individually causes minimal problems. Millions driving cars cause major problems.
Externalities are not just limited to pollution. It can be argued that the American insistence on utilizing the Imperial system of measurements creates a negative externality to all those who do not use it. Nearly the entire globe uses the metric system, but because America is the single largest economy in the world and its preferences have a huge sway, many standards continue to be demarcated in Imperial units to the frustration of all those who use the metric system.
The use of AI-generation as a means of production will more than likely result in similar externalities. While the operator who uses the AI art tool enjoys a positive boost to their productivity, the widespread employment of AI art tools could create situations where those who do not rely on AI might be under suspicion of using them. And while this following point is more related to the concept of dumping, if massive amount of artwork is flooding the market, then there is a strong chance that creators will have a harder time making a living. If an artwork can be created in a few clicks of a button, why hire artists that have to be paid? Not everybody is an enthusiastic fan of those who create the art as some only care about the product.
It must be stated that this is not a guaranteed outcome. The advent of AI art might result in more attention being placed on the human touch and the demand for human art may increase. However, I do believe that the landscape of the creative industry might change dramatically in the future as AI art’s involvement and advancement will cumulatively bring about major changes in ways that cannot be anticipated right now.
AI art’s greatest impact
As it stands, the single greatest impact of AI-generated art is not about how illustrations can be created by computer algorithms, but instead the greatest impact revolves around the fact that huge amounts of art can be created easily, quickly and anonymously. Artists strive hard to create works that match certain expectations as much as possible, but the greatest obstacle that impedes an artist in their quest to realize their goals is not talent nor skill nor creativity, but instead time. Time is the greatest limiting factor for all creators and I believe AI-generated tools will have a profound negative impact on how artists can allocate time.
Yes, AI tools can make artists more productive, but on the flipside, this also means that expectations against artists to be create elaborate, detailed, and fast will increase. It will be harder for many artists to spend time thinking and trying out different things when other can produce accomplished works much faster if they rely on algorithms.
I am reminded of how advances of technology had pivotal impact on certain economic activities. The invention of photography, records and audio tape, automobiles and motor transport… The list goes on and on.
Each innovation enhanced productivity and helped improve the lives of many. At the same time, each of these resulted in certain external costs that were first not so apparent. Photography resulted in major shifts in the artist profession. The introduction of musical recordings fundamentally altered the relationship between the public and musical presentations. Automobiles changed how people lived their lives and transformed living spaces.
Now AI-generated artwork will change how society and people will interrelate with art.
Historical patterns in technological innovation and marketplace reaction
As a general rule, when the supply of a given service or commodity or product suddenly increases in a given market, either through a particular invention or major transition in how goods and/or services are disseminated, prices will fall and consolidation will take place.
When the supply of shoes in a local market suddenly increases, smaller shoemakers that cannot adapt will close shop while the more competitive manufacturers will expand their business. When photography suddenly becomes common place thanks to an added feature to smart phones, the demand for cameras decreased and the number of camera manufactures diminished.
As competition suddenly increases, there will be a strong pressure for business and purveyors to gravitate toward providing higher value added products and services to keep afloat, or to change their price structure so that they can reorient toward a mass market business model.
When a glut happens, the market will soon be dominated by lower priced products/services. Originally the market may have been comprised by a high-tier, a mid-tier, and a low-tier structure. The lower-tier will become the majority and the mid-tier will increasingly become scarce while the high-tier will aim to specialize and stay above the fray. Later on, the mid-tier elements of the market may come to be appreciated once more, but by that time the market structure has been fundamentally changed to prohibit the variability of a mid-tier service/product provider and the market will be comprised primarily of a lower-tier provider that promises ubiquity at a lower cost yet with bare-bones services/products.
Eventually, the nature of the lower-tier product/service provider will invite a transition whereby the provider will aim to change their business model so that they can simultaneously take advantage of their market dominance and increase their profitability. Since the lower-tier product/service that they already provide has minimal value due to it ubiquity and lack of specialization, there is a strong incentive to either add incremental services with which they can charge higher sums and/or couple their business with a product/service that is more lucrative. In many cases, what was formerly an independently viable business will coalesce to become part of a larger multifaceted industrial enterprise. History abounds with examples of this metamorphosis.
Phone services were originally an independent business endeavor. In many countries, telecoms were public utility enterprises, but then privatization transformed them into become competing agents in the free market, which then lead to technological innovation that invited in market consolidation, and ubiquity of their services resulted in many telecoms to become part of larger IT service providers. It has become harder to simply subscribe to a simple phone service—Business would like you to subscribe to an ISP and/or cable TV service.
Musical performance where the mainstay of public entertainment, and even after the invention of audio recording, live performances were plenty. But with the widespread employment of radio broadcasting and television performances, live musical performances evolved to become a premium service since the market was dominated by lower-tier services/products. Eventually cassette tapes and CD were replaced by Internet streaming and subscription services, whereby musical playback devices where consolidated to be part of smart phones and IoT home entertainment systems.
Will AI art be a game-changer?
This type of consolidation of the product/service is nothing new. However, it is important to keep in mind that art and other cultural products have some numerous qualities that are unique to them. Unlike cars or home appliances, you are not pressured to choose one and negate others. Unlike phone services, the enjoyment one experiences from a particular artist cannot be easily replaced by others.
Even while an AI art program might replicate what I create, and its creations might flood the market with material that is similar to mine, my art is still something that only I can create.
At the same time, the value that is attached to art is subjective. If one person is perfectly happy with the facsimile of my art created by an AI art program, then my involvement in the production of my own art is a mute point. If AI-generated illustrations better gratifies the needs of a particular consumer, then they will find utility in that AI-generated art that goes beyond what other human being provides.
As I stated previously, the single most revolutionary element of AI-generated art is that it can be produced very quickly, and therefore the cost associated with the production of that art is much lesser than of human created art.
Opportunity cost (the cost of choosing A over B) is directly correlated to resources being demanded in securing any given product/service. Buying a yacht is not a cheap endeavor. One must forfeit large sums of money and devote time toward learning how to sail one and be willing to incur the costs associated with maintaining a yacht. Paying an artist to produce specific piece of art also incurs a considerable cost, because you are tying that person down toward creating something just for you.
On the other hand the opportunity cost involved in AI-generated content is miniscule.
There may be instances where the funds you have available in a particular creative project may be so limited that it precludes the possibility of hiring a team of professional creators. But as AI-generated illustrative content can be made available at the fraction of the cost, projects that may have been unrealistic in the past will now become viable.
The predictable end result is that the total amount of creative content that is available in the marketplace will probably increase exponentially, while at the same the value that is attached to that said content may diminish considerably, especially in certain fields.
Since AIs can be fine-tuned to reproduce an artist’s art style, the value attached to that art style may, in certain cases, go down. However, this is not a guaranteed result. The desirability of particular piece of entertainment is entirely subjective. Some may ascribe a higher value than others over a given product, and if more people enter the marketplace as consumers, then the value attached to a particular creative property may rise. Be that is may be the case, it is important to realize that the power of mass produced art will most likely inflict external costs in the form of oversaturation of the marketplace. What used to be marketable simply because it was refined and accomplished may lose is luster when the creative marketplace is flooded with similarly polished content.
“You want fries with that anime waifu?”
We may enter an age of “fast art” similar to fast food. The demand for traditional restaurants will never disappear, but what used to be an average, mundane offering as become elevated to being a high end cuisine in some circumstances. Audiences previously could previously directly experience live classical and band music for modest prices. Now most of the public have to satisfy themselves with recordings while concerts are aimed toward the connoisseurs who are willing to pay a premium to enjoy live performances.
There is no doubt that the introduction of AI-generated art will create opportunities that we cannot easily predict at the moment. However, it is equal clear that, if AI-generated art becomes pervasive, which is still a big if, then it will result in externalities that will impact artists, especially for those who will enter into the marketplace in the future.
At the very least, a demarcation will exist for those artists who had been active prior to the summer of 2022 with those who started become active after AI-generated art become widely available. Some artist who were active prior to this date may incorporate AI art tools in the future, while same artists that enter into the creative scene after this date might reject AI art tools. Nevertheless, I suspect that a certain degree of suspicion may be cast upon artists who recently jumped into the creative community, however unfair that may be..
We must also keep in mind that the introduction of AI-generated illustrations is taking place just when huge numbers of non-Japanese people are increasing entering into the creative landscape. Also, while the non-Japanese market for manga and anime continues to grow, the Japanese market is predicted to plateau and possibly start to shrink due to demographics within Japan.
Where are we going?
One possible outcome is that consolidation will take place, where those on the very top will thrive in an ever larger market, and those who specialize in niche subjects will be able to carve out an exclusive living space for themselves, but for many aspiring professionals, the creative ecosystem will be chronically short on financial sustainability due to oversaturation and an hyper-competitive landscape. The vast majority of artist may have to resort to being amateur hobbyists, not because they are unaccomplished, but simply because it is too difficult to sustain oneself financially as an artist.
Another related possibility is for the rise of artist exceptionalism and how that might be tied to artistic validity. Everybody is unique, but some are more unique than others, and in some cases, that uniqueness will translate into marketability. The remarkable life stories of individuals does not precipitate that their artistic achievement are more notable than others, but it does make for griping stories and audiences tend to find tales involving people who overcame adversity to be quite worthwhile.
Celebrity status of artists and creators is nothing new, but how some artists promote themselves on YouTube and other streaming platforms, sometimes capitalizing on their multitalented expressive facets, have facilitated at new degree of international mass-market stardom that one struggles to find parallels of in the past. Because the competition is so fierce, creators must self-promote themselves much more aggressively than before. Being multilingual, having the ability to entertain audiences in a live setting, and to be accomplished in artistic and/ or entertainment abilities that go beyond the traditional confines of illustrative arts and authors seems to have become standard means necessary to secure an larger audience.
On a more optimistic note, the entry to AI art programs might re-kindle interest toward traditional media (pen and paper) and place a higher value on world-building, storytelling, and narrative variation. The introduction of photography heavily influenced the artistic and commercial painting community in the 19th century. Impressionism and abstractionism transformed the art community forever, while technological advances allowed graphical design to innovate and democratized publishing. Photography may have had a negative impact on painters that focused solely on portraits for the mass market, but painters and creators learned to adapt to the changing environment.
It is plausible that incorporating AI-generated art assets will become mundane within the art production world. This can have negative consequences in that people may expect artists to be hyper-productive. As it stands, many Japanese manga and generalist artists already complain about how readers and editors expect their output to be not just passable and functional but must be dense, intricate and stellar, even while their pay stays the same or is even less then before. Overworking is a major issue and AI art programs can both alleviate this or contribute to this problem.
Reassessing our future
Creative industries always go through transition. This is nothing new. But the pace at which the creative landscape is transforming has become simply breathtaking and almost incomprehensible for those who were born in the 20th century. I am reminded of how photography fundamentally altered the visual medium of human expression, but it is important to keep in mind that photography’s encroachment into the domain of classical painting was incremental and gradual. The true nature of the revolutionary and innovative aspects of photography and how it impacted traditional painters unfolded over the course of many decades.
But here we are, watching how AI-generated art is completely rewriting rules over the course of what seems like 5 weeks. It has now become impossibly hard to predict what will become of anime and manga in the next 5 months, but I hope a new equilibrium will come into being that will empower and enhance the lives of all artists and their audiences. Ideally, we want technology to benefit not only those who enjoy advantages due to their tech acumen or market dominance, but everybody, for the heart and soul of creativity is a human trait that should be shared and cherished by many.
If the current situation is left unchecked, we may have a massive hallowing-out effect take place where the financial rewards and community recognition of new, innovative art diminished to the point where new participants would be discouraged. I can think of numerous examples in the arts where market conditions changed so rapidly that the field atrophied and became frozen in time. It would be very saddening if the same fate awaits for manga and anime.
My own personal postscript
I tried my best to let my emotions at bay as I wrote this piece regarding AI-generated art. I feel this technology has the capacity to bring about monumental change, so I wanted to focus on trying to get a bird’s eye view of the impact AI-generated art might have. Regardless, I am certain my own personal sentiments intermixed with what I presented.
My personal opinion is that AI art programs are nothing more than a too, and it can be used for numerous beneficial ends. As the saying goes, “a bad workman always blames his tools.” It is my belief that anyone with rich imagination, who is passionate about his/her creations, or can envision different scenes within his/her mind can create interesting works regardless of the tools that are available. I feel AI art programs have the capacity to vastly widen the range of what people can create.
AI art programs can make itself very useful in making collages and helping artists come up with new ideas, they can assist creators in different supplementary roles, and help them gather ideas, as well as simplify layout checking and color calibration. If you combine multiple AI programs, it should be easy to come up with an AI Vtuber. It would also be possible to come up with a game with predetermined concept and parameters that continually creates new content for the player to enjoy. If you combine these elements with VR and AR (augmented reality) technology,, it should even be possible to build a world all your own.
But a world that prioritizes comfort for yourself above everything else is a world that rejects the presence of others.
Art AI programs have a strong degree self-serving elements to it and it’s easy to abuse its power. Using its powers to enable a single individual to create extremely complex artistic creations not only has the capacity to decrease the value of entertainment and artistic creations, but also may threaten the worth attached collaborative efforts. Any party with large stocks of previous creations at their disposal will have the ability to endlessly create new works using that dataset they have access to. I believe AI art programs can not only enable people without artistic acumen to make money off of other people’s artistic efforts, but the technology it represents has the potential to amplify the laziness and consolidation of power of those who have access to large bodies of works under their command.
The power of computers to process huge amounts of information in blinding fast speed can bring about huge consequences by engaging in activities that no one predicted in the past and/or abusing the information it provides. Stock price manipulation and the marginalization of certain racial groups’ voting strength through gerrymandering have been problems that have been around prior to the rise of computers, but executing those operations on a wide-scale was previously prohibitive difficult, but that is no longer the case and real life harm has come about as a result of this technological advancement. It would be naïve to believe that would not come about from the rise of AI-generated art.
With regards to AI art programs, I am inclined to believe there are aspects to it that can threaten the trust between creators and their audiences, and amplify discord within artistic communities. While the public is captivated by the rise and fall of famous creators and popular franchises, those parties that control the means of distribution and exchange (platforms) wield far more power and accumulate larger fortunes. While most think of AI art programs as a means of creating illustrations, in fact it is a type of platform where users must access the program to create output, and it would be easy to envision a world in the future where an AI art program’s functionality would be intertwined with art posting/ archiving sites, social networking services and streaming sites, as well as content retailer storefronts. I hope this will not create a situation where a tool that was invented to help people create in fact become a new hegemon that controls and restricts creations.
In the end, a creation cannot come about without the active efforts on the part of the creator. Even when there are no audiences there, an accomplished creative work can come into being through the firm determination on the part of the creator. The world is full of works that are nothing more than cheap hacks. Even commercially successful titles may be in fact be sloppy, half-finished creations. The desire to create something that will satisfy immature audiences, to come up with something that will fill patch a gap in that arose, to simply create something to fulfill a quota—these motivations might drive a party to employ an AI to create content, and in certain case, that content might prove to be commercial success. Nevertheless, I am convinced creators who tirelessly polish their stories to their best of abilities so that they aim to enthrall wider audiences will continue in take to the stage in future. It is my sincere wish that AIs will not hinder such efforts in any way.
PASOKET(Aug. 22nd, 1993 at the Harumi International Exhibition Center)
[This article was originally published in 1998 as an entry inside Doujinshi: The Alternative Publishing Medium of Japan. The content of this article was revised prior to being reprinted in this blog.]
I had the lucky fortune of visiting Pasoket, a doujinshi computer software marketplace after visiting Wonder Festival. Pasoket combines the word pasokon (personal computer) with Comiket (Comic Market) to indicate that it is a doujinshi marketplace for personal computer enthusiasts. Pasoket was much smaller than the Comic Market at perhaps 50 circles (publisher groups) participating and possibly about 500 people attending. Pasoket was very comfortably held in one of the smallest buildings in Harumi. Still, Pasoket was fun and lively. Admission to Pasoket requires you to buy their exhibition guidebook. As I borrowed a guidebook from a friend, I can’t say how much it was.
The set up was fairly simple. People who come up with original software at home make a whole bunch of duplicates on floppy disks and come to Pasoket to sell them to those visiting the event. Much of the packaging suggests most of the software contains erotic images, but as I did not go through all the circles carefully, please my observation with a grain of salt. The variety of doujins available was quite limited compared to Comiket Market but I did not mind. After all, Comiket itself started from very humble beginnings. Unfortunately, I do not own any common PC system (e.g. NEC PC-8800/9800, Sharp X68000 or X-1, MSX systems, Fujitsu FM series system are the most common systems right now) buying the software did no good for me. There might have been some MAC and IBM compatible software there, but I didn’t see any. Luckily for me some groups were selling books too.
It appeared to me that many, if not most, of the books there actually were the overstocks carried through from Comic Market last week. Most of the books were either companion books to the erotic software packages and/or spin off stories utilizing the characters featured in the games itself. There was considerable diversity in the game formats. Some were fairly simple: Collections of computer graphics illustrations and custom icons for you PC. Others were much more elaborate: War game simulation systems, adventure games, adventure games, and shooter games. Most were in the middle: Jigsaw puzzle games, Tetris type games, and other types of simple puzzle games were prevalent.
Since I could not play the games on offer, I cannot really say how the two match up against each other. Judging from looking at the sample CG images featured in numerous Japanese erotic computer software review magazines, I would venture to say that the commercial ones have better graphics and more elaborate game functions compared to that of non-commercial ones, but I would be careful with that generalization. You never know. What you see in the magazines are only samples of a small number of products representing only a small portion of all that are in existence.
Unfortunately, the repercussions left behind by Black February [1991 police crackdown on doujinshi selling commercial bookstores] were apparent here too. As suppression of anime/manga erotica accelerated, the Japanese police force went after the commercial adult software manufactures as well. This sent many amateur erotic software makers into disarray and the result was massive self-censorship among both commercial and non-commercial erotic software. I do not know in detail what was the extent of the self-censorship over the content of games at this time, except one small but very important detail in the graphics. Many ceased to render details of the pubic area. Some say there are special software codes that permit the censored areas to become “uncensored”. [Later, the Japanese police would go after software programmers that enabled images to be pixilated and unpixilated.]
Something that surprised me about Pasoket was the fact that there were a fairly large number of young women there. Interestingly enough, it seemed that a good portion of them were there on the behalf of the individual software makers, possibly attempting to boost sales by enticing buyers, most of were male. Dressed in fancy intricate clothing (for the most part the emphasis was on trying to be cute and adorable as opposed to daring and revealing) they were calling out to the shy attendees that were passing by their tables. I had the chance to observe some of the girls making purchases as well. I even witnessed a few who claimed they were selling stuff they had developed. It was strange to see such an influx of women into the field of personal computing, which has been until recently considered to be with the domain of introverted young men. It seems that the availability of the technology is contributing to the erosion of this stereotype.
Hopefully I will be more knowledgeable about computers and learn more about Japanese domestic computers before I go to another Pasoket.
[Postscript: By the time Windows 95 was introduced, native Japanese personal computer standards started to fall by the wayside and the IBM or Mac compatible systems quickly gained dominance. Pasoket and other doujinshi software marketplaces eventually disappeared as many doujinshi software circles distributed their software via online platforms. As far as I know, Pasoket itself continued on until 2005. (Pasoket web archive link here.) Please note that while doujin software marketplaces have largely disappeared, many larger doujinshi marketplaces continue to feature numerous doujin software circles. Generally speaking, doujinshi marketplaces accommodate those who wish to sell software along with printed media.]
Toranoana is a retailer in Japan famous for being a doujinshi consignment shop, but they also sell regular commercial publications as well as provide a crowdfunding platform called Fantia.
Toranoana’s main store in Akihabara has been a fixture of otaku culture in Japan for many years. But as of Sept. 1st, 2022, that massive store is gone. https://news.toranoana.jp/notification/215167
Toranoana is shutting down almost all its stores. While Tora’s online sales & subscription membership platform is booming, their in-store sales have been stagnant. International visitors will have a harder time making purchases in Japan when they visit, but different options are available. (Having the books sent to their home country, or picking up deliveries at the remaining store at Ikebukuro should still be an option.)
While this is not a doomsday scenario, I believe this is pretty bad news for niche doujinshi doujinshis and commercial books alike. The problem with mail order is that people tend to make fewer discoveries when ordering via the Internet compared to when one is actually walking around a physical store.
I believe Toranoana will be successful into the foreseeable future, but it is heartbreaking to see something so familiar to be gone.
Most otakus of my generation born in the 1970s in Japan should have pretty good recollection of who regular bookstores where shutdown by the police for selling doujinshi. Thanks to the serial killer Tsutomu Miyazaki and the media attention that followed, otakus were treated like dirt, subhuman garbage that didn’t deserve to walk to this earth in the eyes of many. Otakus internalized that disparagement and loathing with self-deprecating humor. We had lots of books aimed for us and carved out spaces for ourselves at doujinshi marketplaces. We knew we were legion, but we had no where to go.
And then in the mid 1990s, stores and services stared to show up for us in Japan. We found a permanent home at Akihabara. We could be ourselves and do business. We felt were valid finally.
Toranoana, Messe Sanoh, K Books, Sangatsu Usagi, Melonbooks all showed that we could be viable. Not only that. They helped make us needed and attractive. Akiharabara, Otome Road (Ikebukuro), and Nihonbashi (Osaka) were places we could flock to and flourish in.
And the tourists flocked to there as well. We were overjoyed. Not only because we saw people from overseas wanting our otaku subculture stuff, but because many of us felt we proved all those bureaucrats and academics wrong. We could make Japan interesting and noteworthy without fancy high culture. We didn’t need to censor the sexually charged material that was on offer for both men and women. We could all be who we wanted to be. We felt we were valid for who we were and are.
Now the Japanese government, bureaucrats, and academics are bending over backward for to accommodate Japanese otaku culture. One of the foremost otaku manga authors has even been elected into the Diet, Japanese national legislative assembly. Nevertheless, as shown with the example of Toranoana’s closing of many physical stores, the industry and wider otaku community is hurting because of demographics, macroeconomic forces and bad decision making regarding tourism. Toranoana’s Akiba store closing signifies a major challenge to this community.
Many familiar stores and sights that are part of the otaku culture is still here in Japan, but it is scary to see how it is being chipped away, little by little. Only time will tell if the Japanese doujinshi community can successfully overcome the huge challenges that we face today. I truly hope so.
As Comic Market celebrated its 100th occasion this past summer (yes, I know C98 was cancelled in 2020, but Comiket has held multiple special sessions in the past, so C100 is actually an undercount), I decided to publish a doujinshi on translation. How to Translate Japanese Manga into English is a book aimed toward Japanese creators who want to either translate or would like to have others translate their works.
I have been a professional translator for over 30 years, so I wanted to help share that knowledge with others who may want to release their work for English audiences. This book deals with topics like ascertaining the strengths and weaknesses of the original work, understanding readership demographics and their characteristics, what are the limitations in non-Japanese translators dealing with Japanese subject matter, what are the different approaches toward translation (e.g. aiming the translation to be easily relatable for non-Japanese readers vs. constructing the translation in such a way that non-Japanese readers can learn about Japanese culture and the authors’ sensitives in the most accessible way possible), and much much more.
Why are the main characters in anime and manga almost always young? This is a question that is commonly brought up among many overseas fans of anime and manga.
This is actually a trick question–If you look at the wide breadth of material from Japan, it won’t take for you to notice that there is actually a huge amount of material that revolves around characters that are adults. But the perception that anime and manga revolves around youth is something that needs to be examined carefully, as this is an enduring stereotype.
It is my belief that many in the West have a hard time appreciating and understanding the Japanese preoccupation with school day youth in fiction. While it may not dominate the media landscape as some would believe, the fixation is there and it is widespread.
I was born and raised in Japan. I only went to Japanese public school for three years but I lived in Japan through-out the 1970s to the 1980s. I have many Japanese friends and I spent a lot of time with them. I would like to think that I have a relatively good understanding of the cultural attitudes that are shared by many in Japan.
As such, I would argue that, for many Japanese people, life during the youth school days hold a uniqueness spot—One unrivaled as one full of potential and agency.
This isn’t to say there are other aspects involved here, but for many of us in Japan, people romanticize life between grades 7~12 as days when you have more agency over yourself than any other stage in your life, or the potential to have agency seems more pronounced.
Japan is a society where social pressures are very powerful. Conformity and peer pressure can be very oppressive. It is a lot hard to imagine how an adult person my cast aside their family and society obligations to pursue something extraordinary. This is the case in school as well, but youthful transgression allows many to fantasize about how they might do thing differently if they were given the opportunity. Many feel their lives are severely constrained but people romanticize about the school days as when you “could” break from the system due to your youth.
The (Western) ideal of an independent, self-sufficient individual is not unknown in Japan. But for many Japanese, we are constantly reminded of our affiliation with various different organizations (employment) or roles (gender, parental, etc.)
Let us assume there was a Japanese anime revolving around a giant robot show with Japanese characters set in contemporary Japan. If the main character is a professional pilot, then we tend to ascribe the worth of the individual based on their affiliation with the professional career, not on their personal merits. The personal merits are important, as those are the things that allowed the person to hold the position he or she holds, but trust worthiness and social reputation is usually closely associate with the successes that come about from teamwork and working within the system.
Let’s now make the main character be a young male or female, and then their ability to pilot the giant robot seems more special, more unique to their own personal attributes. It also makes it easier to have a story where the main character is in a moral dilemma. A mature, professional pilot facing a difficult moral situation is certainly worthy to be a good story, to be sure, but some in Japan would find it too realistic and the sense of wonder would be diminished. It’s harder for many Japanese to entertain the idea of a mature, righteous person fighting the system.
This is not to say there are no stories like that–I can think of many many anime and manga titles that revolve around mature characters fighting the system. But many of these anime and manga titles rarely see the time of day outside of Japan.
Therefore, I would have to counter the argument that “anime and manga only features young, cute characters” needs to be qualified much better. Yes, anime and manga does feature a lot of adolescents, but that’s more about which titles are being translated and released overseas.
Finally, I would also like to note that the Japanese perception regarding youth and agency is culturally rooted, but does not preclude enjoyment from mature, independent minded characters–A lot of Hollywood movies with more mature character are popular in Japan.
Many Western narratives hold a unique position within the hearts and minds of the Japanese audience, in my humble opinion. Because the stories and characters are grounded in localities outside of Japan, audiences do not project as strong Japanese social expectations upon them.
It may seem strange for some Westerners that the Japanese have different sets of expectations regarding agency, self-determination and moral guidelines. But this is not something uniquely Japanese. People’s expectations and perceptions are dependent on the setting/actors involved.
Talking about expectations, value systems, morality, and “individuality” is very interesting but needs to be reserved for a different occasion. It would nice if I can find the chance to take up that subject in a later blog entry.
(This blog is based on a thread of observations and analysis I conducted on Twitter in Janurary of 2022.)
No one is born with the ability to appreciate memes nor speak masterfully. It’s all learned. If it is learned, it means it can be shared to others at a later time.
To be sure, cultural context and shared knowledge has a huge impact on the efficiency of the translation. It’s difficult to translate words and concepts related to computers if the target audience has had no or little contact with computers. But the words and concepts can still be translated and shared, if the audience is willing to understand and learn.
A translation might be clumsy, it might not always be snappy. But I believe anything that is learned can be translated and shared to audiences beyond the original target group.
Dan Kanemitsu's Paper Trail 