Ksenia Se

Ksenia Se is a journalist and the founder of Turing Post, a publication about artificial intelligence and machine learning. She gives readers a way into current research by tracing familiar ideas back to the problems that gave rise to them. To explain how modern language models became possible, Ksenia and her co-author Valeriia Kuka began their history of language models with an early proposal to translate text using punched cards.

From explaining algorithms to founding Turing Post

Before starting an AI publication, Ksenia was New York chief editor of Snob magazine and editor-in-chief of TheQuestion, a Russian-language journalism platform. In 2018, she moved back to New York and worked for a company that used machine learning to analyze digital currencies. When she represented the company at conferences, she found herself spending much of her time explaining how the algorithms worked. As she taught herself machine learning, she found plenty of alarming predictions in the general press and useful technical writing that assumed more than a beginner would know. She wanted a publication she could have learned from herself. Her account of the move into AI journalism

In 2020, computer scientist Jesus Rodriguez suggested trying the idea as a newsletter. Together they founded TheSequence, publishing research explainers and product reviews. Ksenia wanted more room to cover how businesses put AI to work and what artists could make with it. Rodriguez wanted to keep TheSequence’s existing focus. As interest in ChatGPT grew, Ksenia returned to her broader idea and launched Turing Post in 2023. Turing Post’s early weekly digest

How Ksenia reports on machine learning

• The attention mechanism. When Ksenia set out to learn how attention got its name, Dzmitry Bahdanau recalled Yoshua Bengio adding the word to the conclusion of their 2014 paper on machine translation. Ksenia found it on page four and asked Bengio what he had meant by it. His explanation started with something a human translator does routinely. When choosing the next word, the translator can look back at a particular part of the original sentence. The mechanism lets a model give different parts of its input more weight at each step of the translation. Ksenia’s history of attention

• World models as an editorial focus. In September 2026, Ksenia set out a question Turing Post could ask across robotics and software. Can a system track a changing environment well enough to predict what an action will do? One possibility she proposed was a coding agent that could anticipate the effects of an edit before changing a production system. That question gave the publication a way to spot related work even when the people building it used different terms. Ksenia’s case for covering world models

Ksenia also leads Turing Post’s Inference interview series and produces its Attention Span video essays. She served as an MC at AI Engineer NYC in October 2026, where she shared the closing remarks with swyx. Turing Post contributor profile · AI Engineer NYC closing remarks

AI in her own newsroom

Writing in English as a second language, Ksenia used ChatGPT to shorten rounds of copyediting. She asked it to check grammar while preserving her style. It could weave scattered notes into a draft, but she recognized its favored words and sentence patterns and rewrote the result. Discussing Turing’s imitation game in 2023, Ksenia argued that a model could imitate a person convincingly while leaving open whether it thinks. She also worried that taking human resemblance as the goal could narrow what researchers imagine machines might do. Ksenia’s interview with Creativity Squared

By November 2026, agents were helping Ksenia pursue more projects at once. Each project also gave her more material to read and more decisions to make. She drafted her essay about that decision overload by hand, in a notebook on a plane. Ksenia’s essay on decision overload