Silicon Valley’s highly developed sense of self, Elizabeth A. Kessler, δημοσίευση Apollo Magazine [30/1/2023]
In July 2022, the San Francisco-based company OpenAI posted to its blog an array of headshots made by DALL-E 2, an artificial intelligence app. From a simple phrase – ‘a portrait of a woman’, ‘a portrait of a heroic firefighter’, or ‘a photo of a CEO’ – DALL-E 2 uses AI to generate several novel and impressively convincing images to match, each a plausible likeness of a woman, a firefighter, or a CEO.
Other text-to-images AI apps, such as Stable Diffusion and Midjourney, have been similarly trained to produce credible new human faces. Close scrutiny reveals a few tells: the eyes and teeth aren’t quite right, the fashions are impossible to date. But the resemblance to people is so strong, the trick of producing unique pictures so compelling, that it is easy to overlook the mismatch with the first word in the phrases used to generate them. These are not portraits of women or firefighters, which would assume they represent real people, or photos of CEOs, which would rely on the recording of variations in light intensity. Instead, they simulate a representational genre, the portrait photo. In doing so, they give AI a face (or many) and position this latest technological advance firmly within the visual culture of Silicon Valley, the region of California most associated with the high-tech industry.
AI-generated portraits belong to a well-established tradition of using portraiture to represent machines and of creating portraits with those new machines. Pictures (mostly photos) of inventors, engineers and company founders have long mediated our relationship with the engineering marvels produced in Silicon Valley. Today’s tech companies are often associated with a famous face: Elon Musk and Tesla, Mark Zuckerberg and Meta, Steve Jobs and Apple. But whatever the era, whatever the latest technological advance – vacuum tubes, semiconductors, personal computers or neural networks – portraits have promised a way to grasp innovations that are too small, too complex or too visually opaque to stand alone. In turn, a machine’s ability to craft a portrait – or to manipulate those crafted by humans – seems to testify to increasingly sophisticated technological capabilities.
Η συνέχεια εδώ.