Asqat Yerkimbay

Change Agent, Connector, Educator/Teacher, Journalist, Project manager, Writer/Editor

Askhat Yerkimbay teaches journalism at SDU University in Almaty and manages media development programmes at Internews. His work sits between two rooms: the classroom, where he trains the next generation of Kazakh-language reporters in data journalism, AI and news literacy, and the newsroom, where he helps regional outlets stay independent and financially alive. Long before the current wave, he translated Facebook and WordPress interfaces into Kazakh and founded Minber, an NGO training young reporters in remote regions. He builds the tools he teaches. An AI editorial assistant he developed is in daily use at small independent newsrooms in Kazakhstan. His research asks the question the global AI debate keeps skipping: what happens to journalism in languages that AI systems barely speak? He writes on linguistic bias in AI tools, dependency on Russian-owned platforms in border regions, and how Kazakhstan's new AI governance shapes media independence. He holds an MA in Communication from the University of Wyoming. Since 2019 he has served as an external assessor for the International Fact-Checking Network in Central Asia, and since 2012 he has coordinated the Kazakh language team of the TED Translators.

Kazakh-language journalism, and the unglamorous work that keeps it alive — training reporters in small regional newsrooms, translating what the world is reading, and making sure the tools of the moment work in our language too. I'm also drawn to the places where language policy, media freedom and technology overlap: who gets to be understood by a machine, and what it costs a society when its journalists have to work in someone else's language to be taken seriously. And teaching. I've been in a classroom since 2014, and I still think a good question from a student is the fastest way to find out what you don't actually understand.

A language does not survive the AI era on nostalgia or legislation. It survives if it is present in the data. Kazakh is spoken by millions, but in the material that trains the world's AI systems it is a rounding error — and so the tools that now mediate knowledge translate it poorly, summarise it badly, and quietly push its speakers toward larger languages. The same is true for hundreds of under-resourced languages. The answer is not to wait for big technology companies to notice us. It is to build the record ourselves: to translate, subtitle, publish, digitise and open up text in our own languages, deliberately and at volume. Every subtitle file is infrastructure. Every archive we open is training data for a future in which our language is understood rather than approximated.

AI in journalism, data journalism, fact-checking, journalism education, Kazakh-language media, media development, media policy, multimedia storytelling, news literacy, translation and subtitling

I started translating TED talks for a selfish reason. I was struggling in a statistics course, and I decided that if I translated a talk on the subject into Kazakh, line by line, I would finally understand it. It worked. It also taught me something I had not expected: Kazakh handled the concepts perfectly well. The language was never the obstacle. What was missing was material in it. That turned a study trick into a purpose. Translating TED became a way to show that Kazakh is a language capable of carrying science, and that any one of us can add to it. Later I came to know the curators who organise TEDx events in Almaty and Astana, and the direction of the work reversed. We began translating the other way as well — taking talks by scholars from Kazakhstan and carrying them from Kazakh into English, so that ideas formed here can reach the people who need them. That was the point at which this stopped being consumption and became contribution. It is why we kept expanding the work. The next task is the one I care about most: making sure Kazakh is treated as an equal language in the age of AI.