What's Really Behind Those AI Art Images?

It's been a while since a particular piece of technology has felt as fraught, efficient, and consequential as today's text-to-image AI art-generation tools like DALL-E 2 and Midjourney. The reason for this is twofold: The tools continue to grow in popularity because they are fairly easy to use, and they do something cool by conjuring almost any image you can dream up in your head. When a text prompt comes to life as you envisioned (or better than you envisioned), it feels a bit like magic. When technologies feel like magic, adoption rates tick up rapidly. The second (and more important) reason is that the AI art tools are evolving quickly--often faster than the moral and ethical debates around the technology. In just the last month, the company Stability AI released Stable Diffusion: a free, open-source tool trained off of three massive datasets, including more than 2 billion images.