2024 original: made with Stable Diffusion

The original challenges. An image captioning model and Llama 3 wrote a prompt for each painting, and Stable Diffusion XL painted the AI image from that prompt.

2026 rematch: made with ChatGPT

The same paintings as above, but recreated by ChatGPT. Plus two new experiments: in The Forger, ChatGPT copies the painting it sees. In From Memory, it only gets the title and the artist.