Can robots recreate Da Vinci's aura? It's already happening

For just a couple of dollars and a few selfies, Lensa spent the last month turning everyday users into magical elves and cinematic warriors. It was a massive viral hit, but for the millions of people who trade human artistic skill for a living, it felt less like a fun filter and more like an existential alarm bell.

AI image generators like DALL-E, Midjourney, and Stable Diffusion took a massive leap forward this year in quality, speed, and cost. While white-collar creatives once assumed their intuition and aesthetic taste were safe from automation, entry-level commercial art has essentially been democratized overnight.

Copying living legends and taking commercial stages

The displacement is not theoretical. This winter, the San Francisco Ballet tapped Midjourney to design the visual campaign for its production of The Nutcracker. While a representative noted that nearly 30 human creatives, designers, and producers were still hired, the machine had already made its way into top-tier cultural production.

Even more troubling is the prompt-based mimicry. Fantasy illustrator Greg Rutkowski recently discovered his name was a more popular prompt inside Stable Diffusion than Pablo Picasso or Leonardo da Vinci. As Rutkowski pointed out, the only way to avoid feeding the model is to stop posting work online, an impossible tradeoff for any working commercial artist.

Meanwhile, copyright law remains completely ill-equipped. In the EU, lawyers are still debating whether scraping copyrighted imagery for machine learning is legal, while the UK has proposed legislation offering carte blanche commercial training to attract AI companies. Making matters murkier, current copyright statutes do not protect artistic styles.

Datasets under fire

These massive foundation models come with glaring technical and social flaws. Because training data relies on human-labeled images, algorithms reflect existing societal biases. Users quickly found Lensa generating overly sexualized female avatars and exaggerating racial phenotypes.

In response, artists Mat Dryhurst and Holly Herndon founded Spawning. The team built a search tool allowing anyone to see if their art lives inside LAION, a 150-terabyte dataset powering major generators. Spawning is also developing opt-out mechanisms and licensing frameworks for creative likenesses.

Surprisingly, Stability AI and LAION agreed to honor opt-out requests for the next Stable Diffusion training cycle, and a recent update notably stripped the ability to prompt living artists by name.

Why it matters

The baseline value of raw execution just plummeted to zero. If you are an art director or marketer needing quick mockups, these tools save countless hours and thousands of dollars. But for working commercial illustrators, clinging to client-driven production without integrating AI looks increasingly unsustainable.

In our view, the backlash reveals both the vulnerability and the ceiling of current generative models. As Dryhurst noted, AI can only churn out pale imitations of what humans did years ago; it cannot invent the next aesthetic movement. When every low-effort social media feed starts looking identical, custom curation becomes the real differentiator.

Artists who treat AI as a concepting assistant, like illustrator Michelle Thompson, or pioneers like Anna Ridler and Mario Klingemann who train custom datasets, show where the puck is going. The technology is here to stay, and the future will not belong to who ignores AI, but to who can take its generic outputs and bend them into genuinely new human vision.