ChatGPT feels like a helpful friend but comes with risks, says Danish translator

A Danish translator says ChatGPT can speed up his work but warns the AI tool makes unpredictable errors and requires constant fact-checking, DR reports.

Jamie Klarskov, who translates product descriptions and blog posts from Danish to English for toy company Legeakademiet in Hørsholm, has used AI professionally for over three years. While the tool saves time on routine tasks, he describes its mistakes as “more random” than human errors.

“There are days when I’m really happy using ChatGPT. But other days, it behaves strangely—it has a mind of its own,” Klarskov told DR. He recalled an instance where the AI inserted an Armenian word into a Danish-to-English translation. “It does these odd things sometimes. It’s a bit annoying.”

Despite never attending formal AI training—a common scenario among Danish workers, according to a recent Copenhagen Business School study—Klarskov relies on “common sense and critical thinking” to mitigate risks. His employer has provided general cybersecurity guidance but no specialised AI courses.

“Using ChatGPT can feel like chatting with a helpful friend who wants to assist with your work. But it’s not,” he cautioned. His personal rules include never sharing sensitive workplace details, always verifying the AI’s output, and cross-checking doubtful information with colleagues or search engines.

While unions and business organisations warn that untrained employees risk errors and security breaches, Klarskov remains unfazed: “I don’t see it as a problem. A big part of using AI is just applying common sense.”

Beyond efficiency gains, he notes the tool frees up time for creative writing: “I have more energy to refine my work and make texts more fun and engaging.”