In Russia, Neural Networks Are Now Dreaming Up Candy Recipes—And the Chocolate’s Actually Good

At the St. Petersburg International Economic Forum, Russia’s Deputy Prime Minister Dmitry Chernyshenko was handed what looked like an ordinary chocolate bar. But it wasn’t. The twist? Its recipe was generated entirely by a neural network—GigaChat, Russia’s homegrown large language model. And it was tailored just for him.

After a taste test, Chernyshenko gave it the thumbs-up—and not just out of courtesy.

This isn’t a one-off experiment. AI is quietly transforming Russia’s confectionery industry. Major players like Krasny Oktyabr, Babaevsky, Slavyanka, and others are using machine learning to design new flavor profiles, drawing on decades of historical recipes and consumer data. GigaChat and its AI peers aren’t just inventing new tastes—they’re fine-tuning color, texture, shape, and density like digital pastry chefs with infinite patience.

But this sugar-coated digital shift goes deeper than just innovation. Neural networks are now monitoring compliance with production protocols, optimizing safety, and helping factories cut resource use and costs—which, yes, trickles down to cheaper chocolate on shelves.

In short: AI isn’t just writing code—it’s rewriting recipes, and in Russia, even your sweet tooth is getting an upgrade.