EU AI Watch: When Old AI Outshines the New: A Skin Cancer Diagnosis Surprise

October 26, 2026

In a twist that has left the AI community both scratching their heads and rethinking their priorities, a 10-year-old convolutional neural network (CNN) has outperformed some of the most advanced large language models (LLMs) in diagnosing skin cancer. Yes, you read that right. While Silicon Valley is busy hyping the latest and greatest in AI, a decade-old model is proving that sometimes, the old guard still has a few tricks up its sleeve. This revelation, which recently scored a whopping 514 points on Hacker News, has significant implications, especially for the European Union, where AI regulation is becoming increasingly stringent.

The study, conducted by researchers at the Lynch Lab, pitted the old-school CNN against several state-of-the-art LLMs in a skin cancer diagnosis challenge. The results were startling: the CNN not only outperformed the LLMs but did so with greater accuracy and efficiency. This raises a crucial question: Are we putting too much faith in the newest, shiniest AI models, when older, more specialized algorithms might be better suited for specific tasks?

For the EU, this development is particularly relevant. The European Union has been at the forefront of AI regulation, with the AI Act aiming to establish a framework for trustworthy and ethical AI. One of the key components of this act is the emphasis on transparency, accountability, and the ability to explain AI decisions. The CNN’s success in this study aligns well with these principles. Unlike the black-box nature of many LLMs, CNNs are generally more interpretable, making it easier to understand how they arrive at their conclusions.

Moreover, the EU’s General Data Protection Regulation (GDPR) also comes into play. The GDPR mandates that AI systems must be transparent about how personal data is used. The CNN’s simpler architecture could make it easier to comply with these requirements, as it involves fewer data points and less complex processing compared to LLMs.

What this means for European AI companies is twofold. First, they need to reconsider their approach to AI development. While it’s tempting to chase after the latest AI trends, this study suggests that there is value in refining and optimizing existing models. This is especially true for industries like healthcare, where accuracy and interpretability are paramount. Second, European companies should take note of the regulatory landscape. The EU’s focus on transparency and accountability could make older, more explainable AI models more attractive from a compliance perspective.

The implications extend beyond just healthcare. In sectors like finance, where explainability is crucial for decision-making, the CNN’s performance could spur a shift towards more traditional AI models. This could lead to a renaissance of sorts, where older AI techniques are revisited and revitalized in light of modern advancements.

In conclusion, the surprising success of a 10-year-old CNN in diagnosing skin cancer serves as a reminder that innovation doesn’t always mean reinventing the wheel. For the EU, this could mean a more balanced approach to AI regulation, one that values both cutting-edge technology and the tried-and-true methods of the past. As the AI Act continues to take shape, European companies would do well to remember that sometimes, the best solution is the one that has stood the test of time.

In the end, the takeaway is clear: In the world of AI, newer isn’t always better.

Source: A 10-year-old CNN outperforms frontier LLMs for skin cancer diagnosis — 514 points on Hacker News