According to ScienceAlert, in 2023, researchers from the UK and US used a custom AI algorithm to analyze Raphael’s 16th-century painting, the *Madonna della Rosa*. The AI, built on Microsoft’s ResNet50 architecture and achieving 98% accuracy in identifying Raphael’s work, was trained on the master’s brushstrokes, color palette, and shading. It analyzed the painting, created between 1518 and 1520, and confirmed Raphael painted the faces of the Madonna, Child, and St. John. However, the face of St. Joseph was flagged as “most likely not Raphael,” validating art critics’ suspicions that began in the mid-1800s. The researchers suggest the work may have been touched by Raphael’s pupil, Giulio Romano, and published their findings in the journal *Heritage Science*.
The AI Art Detective
Here’s the thing about this approach: it’s not just looking at the painting. It’s looking through it. The team, led by mathematician Hassan Ugail, used what’s called “deep feature analysis.” Basically, they fed the AI a diet of pure, authenticated Raphael so it could learn his stylistic DNA down to a microscopic level. We’re talking about the specific pressure of a brushstroke, the exact way he blended pigments, nuances the human eye literally cannot perceive. So when it looked at St. Joseph, it wasn’t just saying “this looks off.” It was saying, “the digital signature of these brushwork patterns does not match the learned model of Raphael.” That’s a powerful, data-driven clue, not just a hunch.
A Healthy Dose of Skepticism
Now, before we declare AI the ultimate art authenticator, let’s pump the brakes. The researchers themselves are keen to say this is a tool, not a replacement for experts. And they should be. Art provenance is a messy, human business involving history, chemistry, and paperwork. An algorithm can be fooled. What if Raphael was having an off day? What if he was experimenting? The AI was trained on his known works, but that’s a limited and potentially biased dataset—it only knows what we’ve already decided is true. I think the real value here is how it quantifies what experts have felt for centuries. It adds a compelling, objective layer to a very subjective debate. But it doesn’t get the final vote.
The Workshop Was a Factory
This finding actually reinforces what we know about Renaissance workshops. Masters like Raphael, Leonardo, and Michelangelo ran bustling studios. Apprentices and senior pupils would prepare canvases, paint backgrounds, and even work on less critical figures. The idea of a single genius painting every square inch of a major commission is often a myth. So the revelation that St. Joseph might be by Giulio Romano isn’t a scandal; it’s probably standard operating procedure for the time. It makes the painting more historically interesting, not less valuable. The AI isn’t uncovering a forgery, but rather illuminating the collaborative process of a 500-year-old artistic enterprise.
So What’s the Future Here?
Look, this tech is fascinating. As noted by the University of Nottingham, which was involved, this method offers a new, non-invasive way to peer into the past. Imagine applying it to dozens of contested works. But the risk is that we start over-indexing on the algorithm’s answer. Art isn’t code. The “why” behind a stylistic anomaly—whether it’s a different hand, a restoration, or a deliberate choice—still requires a human historian to unravel. This tool is like giving an art restorer a new, incredibly powerful microscope. It doesn’t do the restoration for them, but it shows them details they could never see before. And in a field where every detail counts, that’s a revolutionary assist. For more on the painting’s history, you can explore resources like Totally History.
