ChatGPT Isn’t Ready to Translate for Businesses
ChatGPT has exploded in popularity since its debut in late 2022. In January 2023, the platform reached 100 million active users, smashing the previous record for fastest-growing app. Its projected revenue in 2023 is $200 million, and that amount is expected to jump to $1 billion in 2024. Yet despite its prominence and projected value, businesses should be wary of relying on ChatGPT for accurate outputs, especially in the realm of translation.
Here are the main challenges that businesses and customer service teams may face when attempting to leverage ChatGPT for translations.
ChatGPT in a workplace or customer service setting
ChatGPT’s translation abilities have been compared to leading machine translation technologies, but is it more reliable than the likes of Google Translate? Generally, no. This is because ChatGPT is trained on a large language model that consists of similar data to what other engines have been trained on. As a result, its performance is on par with commercial translation products like Google Translate for European languages, which are commonly found on the internet; however, it lags significantly on low-resource or distant languages. Since ChatGPT learns from content on the internet, if a language is not often used online (such as languages spoken in areas with low internet access), the less reliable its translation will be.
Furthermore, the reliability of ChatGPT’s translation differs based on whether the text submitted includes industry jargon or slang. Since slang often uses words with multiple meanings, ChatGPT is more likely to translate to its literal meaning rather than to a colloquialism.
Real-time translation technology can be incredibly powerful and cost effective in a business setting. However, without the help of additional technology, ChatGPT cannot currently fulfill the responsibilities of accurate real-time translators.
Reasons for ChatGPT translation difficulty
There are many reasons why text may be mistranslated by ChatGPT. Misspellings are a common culprit; in the case of customer service translation, an agent attempting to translate a customer’s inquiry will have trouble getting a good result if that customer’s request includes misspellings. This is a common problem, as 71% of American adults say they often find spelling mistakes in others’ correspondence.
Top News:
The 3 Biggest Lessons Learned From the Music Industry’s Early Web3 Adopters
In addition to spelling mishaps, ChatGPT may not know commonly used colloquialisms, slang or acronyms. Since 80% of Americans admit to using slang and 22% say they use it in every conversation, slang is nearly ubiquitous in modern communication — even in professional settings, where nearly half of Americans say they use it. Recognizing and properly handling slang is crucial for translation technology.
Another pitfall a customer may encounter with ChatGPT is its inability to decipher multiple languages in one message. Even though ChatGPT supports 95 different languages and 12 coding languages, the platform can be confused by the use of more than one language in a single message.
ChatGPT alternatives for customer service translations
Although ChatGPT’s language processing capabilities are sophisticated, the platform on its own is not reliable for conversational translation. Technologies that augment language-based platforms like ChatGPT can be used to bring the engine’s translations up to snuff and generate best-in-class translations for each language pair and customer use case. These technologies add critical context through customizable, industry-specific glossaries that properly handle acronyms, misspellings, slang and industry-specific terminology. Relying on a translation service that doesn’t account for context will lead conversations down the wrong path.
Recommended: Generic AI or Tabular Data AI: Which One Gives You the Competitive Edge?
Ultimately, ChatGPT’s translation capabilities are not yet up to par for use in a business setting. If an organization chooses to use ChatGPT for customer service, it must have additional translation technology layered on top to provide accurate, useful, quality translations. Otherwise, they’re likely leaving money on the table thanks to lost customer loyalties and repetitive chat-related frustrations.
Comments are closed.