Transformers vs. CNNs: Who Reads Ancient Roman Coins Better?

Kari Jaaskelainen

15 Jan 2026 — 1 min read

Can AI read ancient coins?

Ancient Roman coins are tiny time capsules—but there are millions, and decoding their symbols is slow work. A new study by David Reid and Ognjen Arandjelovic tests whether modern AI can spot coin motifs (like emperors, gods, altars) more reliably.

The team compared classic convolutional neural networks (CNNs) with Vision Transformers (ViTs). Unlike CNNs, ViTs look at an image in “patches” and can also learn from unstructured text that accompanies coin images—auction notes, catalog blurbs, and more.

The result: ViTs outperformed the newly trained CNNs in accuracy when identifying these semantic elements, using fully automatic multi-modal learning (images + text).

Why it matters: faster, scalable identification can help researchers mine huge collections for historical insights—and help collectors verify what they’re buying or selling.

Transformers don’t just label coins; they help surface stories stamped in metal.

Read the preprint: https://arxiv.org/abs/2601.09433v1

Paper: https://arxiv.org/abs/2601.09433v1

Register: https://www.AiFeta.com

AI ComputerVision DeepLearning Transformers Archaeology Numismatics CulturalHeritage History

Meet GRACE: a moral governor for safer, more transparent AI

AI agents are getting powerful—so how do we make sure they do the right thing, not just the effective thing? Meet GRACE, a reason-based moral governor that keeps AI behavior aligned with human norms by separating moral reasoning from goal-driven decision-making. * Moral Module: uses deontic logic and explicit reasons

Why some fine-tuned LLMs miss phishing—and how to fix it

Not all fine-tuned LLMs spot phishing equally. A new study tests Llama 3.1 8B, Gemma 2 9B, and Mistral on high-stakes phishing detection—and uses SHAP and mechanistic interpretability to reveal why models do (or don’t) generalize. * Architecture × data diversity matters: Gemma 2 9B hits state-of-the-art performance (F1

AI that explains itself—by following the science

AI that explains itself—by following the science Black-box AI can be powerful, but it often can’t tell us why it made a decision. Concept Bottleneck Models (CBMs) try to fix that by predicting human-understandable concepts first, then the final answer. The catch: standard CBMs ignore domain-specific cause-and-effect and

Meet GenomAgent: A Team of AI Specialists for Smarter Genomics Q&A

TL;DR: Finding reliable answers in genomics is hard. GenomAgent turns one big AI into a coordinating team of specialists—and beats the current leader by 12% on a key benchmark. Genomic facts live across many databases. Standard chatbots struggle because they can’t flexibly query those sources. GeneGPT added

Can AI read ancient coins?

Read more

Meet GRACE: a moral governor for safer, more transparent AI

Why some fine-tuned LLMs miss phishing—and how to fix it

AI that explains itself—by following the science

Meet GenomAgent: A Team of AI Specialists for Smarter Genomics Q&A