Language learning is risk diversification
Part of AI Era Collection
AI isn't magic—it's a tool. What matters is how you use it.
Over the past two weeks, while watching Frieren and Jujutsu Kaisen on Netflix, I noticed the Traditional Chinese subtitles were absolutely terrible. Character names weren’t even consistent within the same episode. Frieren would be called “Mrs. Fraulein” in one scene and “Lady Frielen” in the next. The Jujutsu Kaisen characters were supposedly practicing “Brazilian jiu-jitsu” and attending “Jiu-jitsu High School.” Hilarious. It was clearly raw Google Translate output. I’m thinking the subtitles are so bad that they might not even be translated directly from Japanese, but double-translated from the English subtitles. Even a basic AI translation with some context about these shows would do better.
This reminded me of platform “enshittification.” Public companies don’t care about you. They want money. Netflix’s service quality is destined to keep declining.
It also reminded me of something I wrote before about how supporting multiple languages is thankless work. I ended up removing Greek from my own app.
Some people suggest switching to fan-subbed versions. But there’s another solution: switch to English subtitles. Maybe Traditional Chinese subtitles got abandoned because the market is too small, but Netflix would never pull this on English, the world’s most spoken language. An even better option is to turn off subtitles entirely or use Japanese subtitles.
I can do both (yay!). Thank you, past me, for putting in the effort.
The subtitles were so bad that we initially switched to Japanese subtitles, and I translated live for my wife. Later, feeling too lazy to switch platforms on our TV, we just used English subtitles. Worked great.
I always thought learning languages was about communicating with people. Turns out it’s about diversifying against terrible subtitle risk and not being held hostage by platforms.