Rabat – While AI image generators are no longer news, a London-based startup is taking the internet by a storm for introducing a potentially game-changing text-to-image AI generator.
Dubbed “Stable Diffusion,” the tool relies on Machine Learning – – feeding algorithms old data and enabling them to generate new input without any programing involved – – and open-source image synthesis, according to converging reports.
Being a deep learning tool, Stable Diffusion would allow users to virtually generate creative images using two-word key phrases or more.
The basic premise behind the technology was common before “Stable Diffusion.” However, the British startup is now making the technology available to anyone online by offering it as an open-source tool.
Debuting only two weeks ago, the tool is becoming increasingly popular, even more so than its predecessors, with some commentators claiming it “brings implications as big as the invention of the camera.”
Prior to “Stable Diffusion,” the breakthrough of AI image synthesis first emerged in 2014.
Earlier this year, artificial intelligence laboratory OpenAI announced it would put the first text-to-image tool “DALL-E 2” on the market. The tool generates a wide range of visual content ranging from realistic images to sci-fi-inspired art from written text.
Soon after OpenAI released its model, Google, and Facebook – – now Meta – – announced launching their own text-to-image generators.
Much like any new technology that is available in open source, Stable Diffusion raises a multitude of ethical questions.
Under the original code, the tool is programmed not to generate any harmful content including propaganda, violent scenes, or pornography. However, given that it’s open source code, it would be possible to override these restraints.
Read Also: ‘AI Mouvement’: Artificial Intelligence At Center of UM6P’ Training Program
