ChatGPT’s first birthday is November 30th. What has been the impact of this form of AI and how has the business world taken advantage if its offerings?
Olga Beregovaya, VP of AI and Machine Translation atSmartling, tells Digital Journal about her thoughts on the impact of ChatGPT and LLMs on the translation industry and beyond.
Beregovaya begins by marking the approximate year since the AI service was launched to the public: “As ChatGPT celebrates its birthday, we reflect on its transformative impact on technology and the translation industry. The evolution of Large Language Models, exemplified by the OpenAI GPT family, has been a dynamic journey marked by a nuanced trade-off.”
Beregovaya also notes the incremental improvements for the functionality that have taken place this year: “With additional training parameters and data, these models can perform a myriad of tasks, expanding their capabilities.”
In terms of the current iteration, Beregovaya observes: “In translation, GPT4 demonstrates improvements in achieving relevant results through fine-tuning, zero-shot and few-shot examples.”
Not everything has worked out as planned and the service has experience ‘teething’ troubles, which Beregovaya identifies as: “However, challenges persist in addressing foreign language vocabulary, grammar and cultural nuances, highlighting the ongoing need for additional model training techniques and multilingual data.”
In terms of the implementation at hr own company, the AI expert details: “At Smartling, deploying successive GPT models has revealed that tasks like translation and language smoothing require specialized skills in prompt engineering and model fine-tuning, and prompt performance validation by native speakers of the target languages.”
In summary of her own experiences Beregovaya states: “ChatGPT’s journey embodies the inherent challenges of pushing the boundaries of AI.”
Such is the ubiquity of Chat GPT that politicians have taken notice. Here, from the U.S. perspective, Beregovaya considers: “The recent White House pledge involving major tech firms, including OpenAI, signifies a positive step toward addressing the risks associated with deploying AI-based applications and publishing AI-generated content.”
Of such measures, which are paralleled in Europe, Beregovaya is largely supportive: “Allowing independent testing and commitment to data governance and transparency demonstrates a collective effort to instil confidence in the reliability and security of AI technologies.”
Beregovay ends on an optimistic note, seeing industry as the driver (albeit it within the framework of the aforementioned regulation: “While acknowledging the current uncertainties, the industry’s commitment to addressing trust, safety, and performance concerns ensures a foundation for continued growth and innovation in Large Language Models”.
