Everyone seems to be talking about ChatGPT. During the past few weeks, it has been a hot topic around college campuses, in the news, and in the professional workspace. The conversation includes a large variety of opinions and theories. Teachers bemoan the development of new ways to cheat on assignments, while companies rejoice about the possibilities of spending less time and money on writing articles. So what exactly is ChatGPT, what does it do, and how could this affect the translation industry?
The three letter acronym of “GPT” stands for “Generative Pre-training Transformer”, the name of the model used to create the program. ChatGPT is an AI-based system that uses natural language processing to create answers. A user can ask ChatGPT any question. The AI responds with relevant answers or continues the conversation in a way that mimics ordinary human speech. Much like Machine Translation systems and Translation Memory Databases, ChatGPT is able to produce these responses because it analyzes and copies huge amounts of data that were written by people. This allows it to imitate how real humans converse. The AI can remember things that were said in earlier conversations, explain logic, and even apologize if its answers are incorrect.
The beta version of ChatGPT has already reaped benefits for companies. It helps brainstorm ideas when given keywords or topics. It can instantly create marketing content such as social media posts. It even generates responses for chatbots or virtual assistants on websites, allowing faster troubleshooting. ChatGPT also claims to be capable of translating text from one language to another.
However, translators are uncertain that ChatGPT will be the new machine-translation tool that some claim it will become. Currently, ChatGPT’s knowledge database is limited to the information given by its creator company. It lacks the ability to search for information outside of what it has already learned. Because of this, it also lacks understanding of context or nuance in language, and is subject to the biases of its creators. Machine Translation programs specifically meant for translating are individually trained by translators to suit their personal tastes and learn alongside the linguist. As a shared use AI, ChatGPT is unable to reach the same level of personalized accuracy.
Additionally, ChatGPT cannot guarantee privacy. Information typed into the chatbot may be utilized to continue training the model. This means that while your personal chat with the AI may not be shared with anyone else, the AI could use similar phrasing or terminology as you in other conversations that accidentally reveals your information. In companies where data security and privacy are highly valued, ChatGPT may not be the best translation engine to consider.
Despite these caveats, ChatGPT still has the potential to benefit the translation industry. The development of linguistics-focused AI will result in more attention and research. This may result in findings that will directly and positively impact the translation industry in the future. ChatGPT will also create faster and cheaper content. The need for linguists to edit, verify, and translate this content will increase consequently.
While ChatGPT isn’t quite ready to push the translation industry to the next level, it has caused higher interest in translation-capable machine programs. It still has a long way to go before it is able to cause significant change, but until then, translators will continue to reap the benefits from the influx of machine-produced texts.