How four of the online AI Image generators deal with the prompt ‘football player in a stadium in the style of Warhol’
Just over 10 years ago, three artificial intelligence researchers achieved a breakthrough that changed the field forever. The “AlexNet” system was trained on 1.2 million images taken from around the web and recognized objects as different as a container ship and a leopard with far greater accuracy than computers had managed before. In 2010, the first artwork using blockchain and AI called “Plantoid,” a robot or synthetic organism designed to look, act, and grow like a plant, was scientifically published. In 2020, the team at DerAffe (parent company of GAY45) announced the first AI curator in history, based at the time on GPT-2 and now based on GPT-3.5. By the end of 2022, ChatGPT had become the most accessed web app in history, with over 100 million users.
Over years many breakthroughs have led to the development of instruments that use AI in ways we could never have imagined just a few years ago. ZKM in Karlsruhe, the most respected media art center in the world, has started a new initiative called intelligent.museum, in which I have been invited to participate. One section will focus on Queer AI.
Generative artificial intelligence presents a philosophical and practical challenge on a scale not experienced since the start of the Enlightenment. The new technology bids to transform the human cognitive process as it has not been shaken up since the invention of printing.
ChatGPT is the latest in a line of increasingly dramatic public demonstrations. Another OpenAI system, GPT-3, electrified the tech world when unveiled in the middle of 2020. So-called large language models from other companies followed before the field branched out last year into image generation with systems such as OpenAI’s DALL-E 2, the open-source Stable Diffusion from Stability AI, and Midjourney.
“These models are going to change the way that people interact with computers,” says Eric Boyd, head of AI platforms at Microsoft. They will “understand your intent in a way that hasn’t been possible before and translate that to computer actions.” As a result, he adds, this will become a foundational technology.
Here are some details on how to use the various AI models:
ChatGPT: This language model interacts in a conversational way and can answer follow-up questions, admit mistakes, challenge incorrect premises, and reject inappropriate requests. It is an invaluable instrument for writing anything. To use it, visit https://chat.openai.com/ and log in.
The image was generated by Midjourney asking: “Photograph of a dark-skinned woman with a hawk on her shoulder, fierce, epic”
The image was generated by Midjourney asking: “Photograph of a dark-skinned woman with a hawk on her shoulder, fierce, epic”Midjourney: This model uses Discord as its delivery/creating mechanism, so you need to have or create a Discord account to try the service. It generates images from textual descriptions, similar to OpenAI’s DALL-E and Stable Diffusion. Visit https://www.midjourney.com/ to learn more.
The image was generated by DALL-E2 asking: “I would like an image of two cute boys, one black and one white, kissing against the backdrop of an apocalyptic city. The resolution should be 150 DPI, and the image should be very colourful with an almost realistic painting style.”
The image was generated by DALL-E2 asking: “I would like an image of two cute boys, one black and one white, kissing against the backdrop of an apocalyptic city. The resolution should be 150 DPI, and the image should be very colourful with an almost realistic painting style.”DALL-E 2: This deep learning model developed by OpenAI generates digital images from natural language descriptions, called “prompts.” To use it, simply write a detailed description, including resolution and size, and the image will be generated in about 20 seconds. Visit https://openai.com/dall-e-2/ to learn more.
Stable Diffusion 2.0
Stable Diffusion: This deep learning, text-to-image model released in 2022 is primarily used to generate detailed images conditioned on text descriptions, though it can also be applied to other tasks such as inpainting, outpainting, and generating image-to-image translations guided by a text prompt. Visit https://stability.ai to learn more.
This article was corrected with ChatGPT and they were pushing a longer version of the section describing themselves. Isn’t this so human?
While these AI models can generate text and images quickly and easily, it’s important to remember that you are the final arbiter of quality. And who knows, maybe one day there will be an AI boyfriend.
I asked ChatGPT to write about what is GAY45: “GAY45 is an independently published journal that focuses on LGBTQ+ perspectives, culture, and issues. They may also cover topics that mainstream media outlets may not, such as intersectionality, alternative lifestyles, and the experiences of marginalized or underrepresented groups within the LGBTQ+ community.
In terms of content, including personal essays, opinion pieces, interviews, reviews of books, movies, and music, as well as feature articles on a variety of topics. GAY45 indie queer magazine focuses on producing high-quality writing that offers nuanced, thought-provoking perspectives on LGBTQ+ issues.”
An article by Răzvan Ion
Răzvan Ion is the founder of GAY45. A professor of curatorial studies and critical thinking in Vienna, he is passionate about literature, AI, technology, comic books, the stock market, art, alternative music, movies, and curiosity.
