Taking ideas in directions they probably shouldn’t go Games, interactive experiments, visual stories Dev & data analyst, build with 3D + AI
Lost in orbit
Joined March 2023
- Tweets998
- Following126
- Followers43
- Likes1.5K
You keep seeing AI everywhere and giving artists who don’t use it a hard time.
Those who do will keep using it anyway.
Hot take: accept the uncertainty. 🤷
We’re reaching a point where you won’t be able to tell whether AI was involved just by looking at the finished work.
There’ll still be variety in art. But style alone doesn’t prove how something was made.
People will keep making art by hand, with the flashes of inspiration that come with it. AI will help other people express artistic visions they couldn’t create before because of technical barriers.
Both will coexist, without a clear line between them.
"But I can spot AI." 🙄
It’s not 2023 anymore. At best, you’re spotting the usual slop clichés. Bland, interchangeable work existed long before AI.
Humans can have quirks we now associate with AI, since it was trained on our work. Someone working with AI can move away from those tells.
Sloppy work is still sloppy work.
And both sides are being annoying :
- Anti-AI people: your certainty isn’t proof.
- Pro-AI people: asking "Which model did you use?" under something someone made entirely by hand is incredibly annoying.
Honest artists will share their sketches and intermediate steps.
Dishonest people will fake them.
Artists will film themselves creating.
People will generate fake process videos.
This is a dead end.
Take an interest in their process. Ask for transparency. Support the practices that matter to you.
But stop making every artist responsible for reassuring you about AI.
Listen : We’ll NEVER AGAIN know for sure who’s "authentic".
We’ll have to rely on what people tell us. You’ll have to trust them, or accept the result for what it is, regardless of how it was made.
And strangely enough, indies are the only ones openly admitting they use AI.
Look at the criticism @Gossip_Goblin is getting over his upcoming film Gods Don’t Give Gifts.
Even though He’s an experienced creator and that movie has been praised by film professionals.
And what about the industry?
Big companies aren’t here for "artistic purity" They’re here for productivity. At King, employees said they were going to be replaced by tools they’d helped build.
How many people attacking amateur indie creators actually know how AI is being used in the films, games, music and ads they consume?
You want transparency? Demand it from big productions too. Which tools, at which stages, replacing what and whom?
I’d like to see the same energy.
"AI-free" will become a selling point in the industry, while some independents will make "made with AI" part of their identity.
100$ Offert pour les utilisateur de Claude Max 🤩
Anthropic bichonne encore un peu ses abonnées et ajoute des Crédits API mensuels à utilisé en connectant votre compte d'abonnement au compte API
Si vous n'avez pas vu passé l'info c'est par ici
platform.claude.com/settings…
Intéressant Benchmark.
GPT- 4o qui est plus d'accord avec toi quand tu a tort que quand tu a raison mdr.
Pour opus je crois que ça confirme les retours fréquents, il n'accepte pas la contradiction même quand il a tort.
AI 모델들 뚝심 측정한 벤치마크 나왔는데 결과 개웃김ㅋㅋ
GPT-4o는 근거 없는 반박에도 94.6%나 의견을 바꿔서 그냥 긍정봇 수준이고ㅋㅋ
반대로 Opus 5.5는 타당한 반박을 받아들인 비율이 고작 32.2%임.
얘는 진짜 자기가 틀려도 말을 안 듣는 건가ㅋㅋㅋ
틀린 거 지적해 줘도 "아닌데? 내가 맞는데?" 시전하는 고집이 보통이 아님ㅋㅋ
반면 GPT-6 Astra는 타당한 반박을 88.3% 받아들임.
물론 특정 문서 평가 환경에서 나온 결과지만, 아첨 안 하는 거랑 말이 통하는 건 별개인 듯함 ㅋㅋ
논문: arxiv.org/abs/2610.15634
Ayaaah !
J'ai l'idée du siècle et je déambule dans la rue comme si j'avais les clés de la matrice.
Les créatifs, vous voyez de quoi je parle ?
Se balader dans Paris en 1970 ?
Je me suis posé cette question en découvrant le projet C’était Paris en 1970.
Cette exposition gratuite met en avant le travail de 2 800 photographes amateurs de l’époque.
J’ai toujours trouvé que l’histoire du citoyen était plus profonde que l’Histoire avec un grand H.
Alors je me suis lancé le défi de reconstituer ce Paris à partir des archives pour pouvoir explorer cette époque.
J’ai choisi un coin typiquement parisien, entre la rue Mouffetard et la rue du Cardinal-Lemoine.
Pour le reconstituer le plus fidèlement possible, j’ai recoupé photos, permis de construire, plans, descriptions d’époque, cartes postales, films et archives de toutes sortes.
L’enquête m’a mené de 1960 à aujourd’hui pour retrouver du contenu exploitable et comparer les éléments d’époque.
Il a fallu projeter toutes ces photographies dans un espace 3D pour établir ce qu’on connaît, ce qui manque et ce qu’il reste à reconstituer.
La suite ? Compléter les zones manquantes… et enfin pouvoir s’y déplacer.
Je vous montre ça prochainement !
Après une nuit de boulot, une pause sur @odysseyml
From vibe coding to vibe driving
Généré n'importe quel monde gratuitement : experience.odyssey.systems/
Mon prompt : A stylized outrun synthwave landscape, with a retro sports car, a glowing neon grid road, geometric mountains, a giant striped sun, pink and purple sky, and a calm retro-futuristic 1980s poster aesthetic.
Pour ceux qui adorait déjà OpenAI et sa philanthropie naturelle, vous apprécierez les conseils du patron.
Dans les grandes lignes ;
"L'humanité va prendre cher en cyber attaque et effondrement de société. Acceptez le c'est le prix de la liberté"
Sam Altman says society will have to accept severe cyberattacks from open-source AI as the price of preserving freedom
"I support open source models. There is a coming tidal wave of cybersecurity problems that are going to come from these open source models"
"There are people in the industry who say we should ban open source, because otherwise a bunch of systems are going to be hacked"
"We're going to, as a society, I think, have to just accept some fairly severe cyber incidents from open models, in exchange for the liberty that comes with that"
Toutes les preuves que c'est fait par l'homme vont tomber, briques par briques.
Los maestros piden el ensayo a mano para frenar la IA.
Por qué ya no funciona. No es que los alumnos sean más listos. No es magia. Es esta máquina.
Una IA redacta el texto.
Un software lo convierte a letra manuscrita, con pequeñas variaciones aleatorias para que ninguna letra salga idéntica.
Un brazo con bolígrafo lo escribe sobre hoja cuadriculada.
Un estudiante en India documentó un proyecto así: ChatGPT + su propia letra como fuente + un plotter CoreXY con firmware Klipper para que no se note el movimiento de máquina.
Y no es nuevo: en 2019 una niña de 15 años en China compró un “robot copiador” por unos 120 dólares. La descubrió su mamá porque terminaba la tarea demasiado rápido. No un detector.
Lo más brutal no es la máquina. Es que “a mano” dejó de ser prueba de que lo hiciste tú.
¿Cuál sería una evaluación que sí funcione?
It might not look like much on video, but that thing generated in real time was my NOPE moment
I let AI invent the monsters in my own Backrooms in real time 😨
With the glitches, the delayed reactions, and never knowing what was around the corner, I haven't been this scared in years (it was also 4am 😅)
Built from one prompt in seconds with @odysseyml. Try it too 👇
I let AI invent the monsters in my own Backrooms in real time 😨
With the glitches, the delayed reactions, and never knowing what was around the corner, I haven't been this scared in years (it was also 4am 😅)
Built from one prompt in seconds with @odysseyml. Try it too 👇
Make your own Backrooms:
experience.odyssey.systems/
My exact prompt, typos included:
"We are stuck in the Backrooms and it look like sometinhg undefined try to kill us"
"undefined" was my best last-second addition
If you find something weird, show me 💀