More and more domestic and international companies are entrusting translation, localization and technical content entirely to artificial intelligence to reduce costs and speed up production. Automated solutions are attractive, but without professional oversight errors can surface later as significant business losses.
Research findings and market experience
Global surveys highlight measurable financial risks from unchecked AI use. According to a recent McKinsey international study, 51 percent of organizations using AI reported at least one negative consequence, and in nearly one-third of those cases the root cause was AI inaccuracy. An EY survey found that 99 percent of respondents experienced some financial loss related to AI risk, and in 64 percent of those cases the damage exceeded one million US dollars.
Research focused on translation quality also advises caution: a 2025 study of AI-based translations of healthcare documents found at least one critical inaccuracy in 16–56 percent of cases, depending on the languages examined.
Concrete market cases and common errors
Hunnect, a Hungarian language-services provider, regularly receives companies that consider themselves “AI-damaged” and request professional help to fix machine-generated content. Király Krisztián, CEO of Hunnect, stresses that this does not mean rejecting AI outright: with proper professional control it can be a very effective tool. The problem arises when business- or safety-critical tasks are delegated to AI without expert review.
One cited example involved an electronics company that switched from professional human translation to a fully machine-based solution to cut costs. Within a year they experienced a measurable drop in conversion on their foreign websites due to unnatural style and lack of fluency, and they returned to human expert review.
Typical issues reported:
- inconsistent terminology use;
- inappropriate or unnatural style;
- inaccurate handling of specialist terminology;
- parts of a localized site remaining in the original language, or functional conversion issues (for example, foreign users still seeing prices in the domestic currency).
An illustrative case: an Italian home-decor company’s Hungarian website consistently showed the product name as “kobold” instead of “gobelin” because a single typo in the source text was translated automatically without contextual understanding. For a premium-brand seller, such sloppiness immediately undermines customer trust.
Assessing the risk: when does an error matter?
Király notes that translation errors may be less noticeable for low-cost, mass-market products, but become critical for higher-value goods or transactions with legal or health implications. In those situations translation quality is not merely a communication issue but a matter of business risk management.
He also highlights companies that are technically expert in their field but lack knowledge of the target market’s language and culture. These businesses often cannot perform an adequate professional review of machine translations themselves. Even when staff have learned the language to some degree, without industry-specific linguistic expertise they may miss errors that native customers immediately notice.
Recommended solution: hybrid localization strategy
Hunnect’s experts recommend avoiding binary choices (all-human versus all-machine) and instead adopting a differentiated, hybrid approach. Key elements include:
- full human translation and proofreading for high-priority markets and business-critical content;
- machine translation combined with post-editing for less critical content or secondary markets;
- integrating professional quality-assurance processes into vendor offerings.
Király offers a practical tip for companies preparing to outsource localization: ask the vendor in advance, “How do you ensure quality, and what happens if a mistake remains in the text?” The vendor’s answer and the presence of QA processes reveal how the partner manages translation risk.
Closing thoughts
Research and market experience show that combining AI with human expertise can speed up workflows while preserving quality when proper controls are in place. A purely fast and cheap machine-only solution does not automatically provide a competitive advantage; culturally accurate, natural and professionally localized content can help win the trust of a foreign audience.
Hunnect is one of Hungary’s leading language service providers and translation agencies; the expert quoted in the article is Király Krisztián (CEO).
