ChatGPT, and what it means for translators
Is AI a fitting substitute for the brightest and best minds? That’s a startling claim that was made when ChatGPT recently sat and actually passed an MBA exam, albeit with a somewhat underwhelming B- grade. While this is a respectable yet middling score, it’s not exactly outperforming human students yet, but you can see how the potential could be there, and why people are worried about it.
So what is ChatGPT, and how much should it concern us?
ChatGPT, or Chat Generative Pre-trained Transformer, is an AI-powered chatbot tool developed by OpenAI and released to much fanfare in late 2022. At its most simple, ChatGPT is a means of synthesising and presenting information, but some of its more imaginative applications are said to include writing and debugging computer problems, creating poetry and composing songs. In addition to providing users with rapid information, there have also been some high-profile cases of journalists asking the technology searching questions – sometimes with somewhat unsettling results.
Yet some of its applications are very interesting. As a simple means of gathering information, ChatGPT could help us out in all kinds of areas other than answering questions. A recent article in the Wall Street Journal describes efforts to use OpenAI to create articles and quizzes on Buzzfeed. And if we can harness the power of technology to generate easily-consumable reading materials at scale, then what else could we create?
Imagine a scenario where we could train artificial intelligence on the work of Cervantes or Shakespeare and then emulate the work of these literary geniuses. And in some ways, we could argue that anything that throws more ideas and content out into the world is a good thing, bearing in mind obvious issues surrounding interpretation and factual accuracy that will be discussed later on in this article.
Yet perhaps another reason for the current fascination with ChatGPT is that it strikes at the heart of what it means to be human. Are our thoughts and ideas unique, or just a synthesis of other people’s? If a computer programme can sift through and evaluate data at a far greater speed than the human mind can, then are its insights somehow more powerful than a person’s are, even if they lack that much-vaunted ‘human touch’, as Nick Cave famously argued when ChatGPT tried to emulate his musical style?
All of this brings us to the potential darker sides of chatbots, one of which is vulnerability to human bias when training the technology. This is something which comes up time and again in any field related to machine learning, and is one that is indeed very relevant to our lives as translators, given some of the controversies that have risen around gendered language in machine translation and potential machine translation toxicity, among other issues common to AI data trained by humans.
Further to this is the argument that ChatGPT output is only as good as the person handling it. There have been cases where ChatGPT has produced extremely fluent output that has actually turned out to be factually inaccurate, so it needs to be handled with caution and some degree of critical thinking. And while this is arguably true of any tool, technology-based or otherwise, in this case the potential to create plausible-sounding fake news or even to interfere with democratic processes could certainly be troubling.
ChatGPT is still prone to errors, so it seems that it will be a while before (if) it completely takes over human processes. That said, as with any emerging technology, it is improving all the time. There’s always a tendency to dramatise new developments and to describe any innovative process as being either an amazing watershed that could save us all OR, particularly in the case of anything AI-related, a terrifying new development that will gain a life of its own and render humans obsolete. Yet just as many other tools have changed the way we work without completely doing away with it, perhaps the most optimistic view is that tools like ChatGPT could help further our understanding while not entirely substituting our own original thoughts – at least for now.
As translators, have you been following ChatGPT in the news? What’s your view about the future of AI? Do you see latest-generation chatbots as being an opportunity or a threat? We’d be interested to hear more of your thoughts on this topic, so please feel free to leave comments and let us know what you think.
Bloomberg: ChatGPT Gets an MBA
Digital Trends: ‘I want to be human.’ My intense, unnerving chat with Microsoft’s AI chatbot
Guardian: ‘This song sucks’: Nick Cave responds to ChatGPT song written in style of Nick Cave
Science Direct: How to measure gender bias in machine translation: Real-world oriented machine translators, multiple reference points
Slator: Toxicity in MT: Causes, Challenges, and Solutions
Wall Street Journal: BuzzFeed to Use ChatGPT Creator OpenAI to Help Create Quizzes and Other Content