Guo studied Japanese in university and entered the workforce doing what she loved — translating subtitles for short-form videos being exported to Japanese-speaking markets. It was a niche but growing field, riding the wave of Chinese short video platforms going global.
From day one, she worked alongside AI translation tools. The arrangement was collaborative: AI handled the heavy lifting of initial translation, while Guo and her colleagues served as the critical human layer — checking accuracy, fixing awkward phrasing, and injecting cultural nuance that machines couldn't capture.
"AI tools are efficient," she acknowledged, "but accuracy only reaches about 80%. The text feels mechanical — it still needs us to proofread and smooth it out." For years, that 20% gap was enough to justify her role. Her expertise made the difference between machine-readable and human-readable.
Then, a few weeks ago, her company's HR department called her in for a conversation. The message was straightforward: to save costs, the company would hand over all subtitle translation and review work to AI. Her position was being eliminated.
"If AI employees are more useful, can the company just fire me like that?" she asked a reporter afterward, her voice carrying both frustration and disbelief. "As employees, we hope to get more support."
After negotiation, Guo secured a one-month buffer period for job searching, plus standard N+1 severance pay. She's now working while looking for new opportunities — but the landscape has shifted dramatically beneath her feet.
Her case highlights a cruel irony at the heart of AI adoption in language work. Translators like Guo didn't just coexist with AI — they improved it. Every correction she made, every nuance she added, contributed to better machine translation. She was, in effect, training her own replacement.
The translation industry has been hit particularly hard. Translation rates on freelance platforms dropped 35–40% as AI tools improved. Filipino translators on Upwork saw bid prices fall from USD 0.08 per word in 2023 to USD 0.05 in early 2026. A 2024 Society of Authors survey found more than a third of translators had already lost work to generative AI.
For Guo, the question isn't abstract. She has a Japanese degree, years of specialized experience, and a skill set that the market is rapidly devaluing. The company she worked for didn't fail — it simply decided that 80% accuracy was good enough, and the humans who bridged that final 20% were expendable.
She's now one of millions of workers discovering that the line between "AI-assisted" and "AI-replaced" can vanish overnight.