What happened
Researchers combined two different AI techniques — Transformers (good at finding broad patterns) and Boltzmann machines (good at finding specific, discrete connections) — to make DNA sequence analysis both more accurate and more explainable. Instead of a black box that says 'yes, this sequence matches,' the system can now show which specific DNA regions interact with each other and why, making the predictions useful for actual biology research instead of just prediction.
Why it matters
This is a laboratory demonstration that AI models can be redesigned to show their reasoning when it matters — in this case, revealing how genes actually regulate each other rather than hiding that logic inside statistical weights that only a specialist can interpret.