top of page

Why AI isn't ALWAYS the answer

partial image of a keyboard with translation instead of Enter

​

Updated October 2026

  • AI translation is impressive. Here is when you still need a human.

  • Machine translation has improved enormously, and I use modern tools myself where they are safe and helpful. For a quick idea of what a foreign email says, free tools are fine. For anything that carries your name, your money or your legal position, they are not enough.

  • Where machines still go wrong

  • Meaning. Systems predict likely words; they do not understand your business. A fluent sentence can still be wrong, and errors are harder to spot because the text reads so smoothly.

  • Terminology. A German Gesellschaft, a French société and a Spanish sociedad may each need a different English term depending on the legal context. Machines guess; I check.

  • Tone and culture. Idioms, humour, politeness and brand voice rarely survive a word-for-word approach.

  • Confidentiality. Free online tools can store what you paste into them. Contracts, HR letters and product plans should not go there.

  • Accountability. If a machine mistranslates a safety instruction or a clause, nobody is answerable. A qualified translator is, and Chartered Linguists work under a professional code of conduct.

  • A real example [Add one short, true example of a machine error you have seen, for instance a mistranslated technical term or contract clause. Concrete examples persuade far more than general claims.]

  • The best of both: post-editing. If you already have a machine translation, I can review and correct it against the original. This is often quicker and cheaper than a full translation, and gives you human-checked, accurate text.​




  •  

 

bottom of page