The year 2022 was when AI learned to paint. In April, OpenAI unveiled DALL-E 2, an AI system that could generate photorealistic images and artwork from text descriptions. In July, Midjourney entered open beta, making AI art generation accessible to anyone with a Discord account. In August, Stability AI released Stable Diffusion as open-source software, allowing anyone to run AI image generation on their own computer. Together, these three systems launched the AI art revolution (OpenAI, 2022; Stability AI, 2022).
The capabilities were breathtaking. Users could type "a cat astronaut riding a horse through a nebula, oil painting style" and receive a high-quality, original image in seconds. The systems could mimic virtually any artistic style, from Renaissance painting to anime, photorealism to abstract expressionism. By the end of 2022, hundreds of millions of images had been generated by AI — more artwork than humanity had produced in centuries of art history.
For some, this was creative liberation. People who had never been able to express their visual ideas — writers who couldn't draw, game designers without art budgets, educators needing custom illustrations — suddenly had a tool that could translate imagination into image. Concept artists used AI to explore ideas faster. Architects generated building visualizations. Therapists used AI-generated imagery in therapeutic contexts. The barrier to visual expression dropped to nearly zero.
For professional artists, the impact was devastating. Within months, reports emerged of illustrators losing commissions, concept artists being replaced at game studios, and stock photo revenue plummeting. The training data controversy added insult to injury: Stable Diffusion was trained on LAION-5B, a dataset of 5.85 billion image-text pairs scraped from the internet — including copyrighted artwork, personal photographs, and medical images, all used without consent. Artists discovered they could type their own names as style prompts and receive outputs mimicking their distinctive artistic voices (LAION, 2022).
The legal and ethical fallout continues. Class-action lawsuits were filed against Stability AI, Midjourney, and others by artists alleging copyright infringement. The US Copyright Office ruled that AI-generated images cannot be copyrighted. UNESCO and creative industry groups called for regulation. The AI art explosion of 2022 encapsulates the central tension of the bitter-sweet AI era: a technology that simultaneously democratizes creative expression and threatens the livelihoods of those who spent lifetimes mastering it.
Key Sources
- OpenAI (2022). DALL-E 2.
- Rombach R. et al. (2022). High-Resolution Image Synthesis with Latent Diffusion Models. CVPR.
- Schuhmann C. et al. (2022). LAION-5B: An Open Large-Scale Dataset for Training Next Generation Image-Text Models.