AI Translation Market Reaches $1.4 Billion Monthly, Human Oversight Decreases

Growth of Machine Translation

Google reported that its Translate service now processes one trillion words per month. This figure does not include other major AI translation platforms such as DeepL, Microsoft, Claude, ChatGPT, and Gemini, nor custom corporate systems. The direct sales market for machine translation was estimated at $1.4 billion last year.

AI Accuracy and Human Role

François Yvon, a research director at the French National Centre for Scientific Research (CNRS), notes that well-trained AI systems can handle idiomatic expressions by finding equivalent phrases. John Zimmerman, founder of OKA, a non-profit translating Wikipedia articles, states that for many use cases, raw AI translation is sufficient, particularly for internal documents or product listings where the cost of eliminating all errors outweighs the benefit.

Decreasing Need for Human Intervention

OKA's process involves AI producing a draft, which human translators then check. Zimmerman observes that the quality of machine output has improved to the point where human intervention is less frequently required. Two to three years ago, translators would rewrite significant portions of drafts, but now, the AI-generated prose often needs no correction, making fully automated translation a serious consideration across the industry.

Importance of Human Review for Critical Content

Despite advancements, human oversight remains crucial for high-stakes content. For encyclopedias like Wikipedia, which relies on trust and accuracy, human review is essential to prevent factual errors from being disseminated. OKA, which has translated over 10,000 articles, could not have achieved this scale without AI, but human translators still verify every sentence to maintain accuracy.