Anastassia Shaitarova, PhD candidate at our department, gave an interview to SwissGlobal Language Services AG about her PhD research. When she started exploring the impact of GenAI on natural language four years ago, the topic felt vague and inconsequential. Today, it is attracting growing interest across academia and industry alike.
Here’s an extended TL;DR of the interview:
- Machine-generated vs. human-written text – What sets them apart? Quite a bit, statistically speaking, with trivial aspects like word length and punctuation topping the significance list. We all know that ChatGPT has a tendency toward sesquipedalianism.
- Lexical diversity vs. simplification – The fear of tech-induced linguistic standardization and impoverishment is valid, and it’s definitely observable. However, large models also expose us to an impressive array of new vocabulary - an emergent benefit.
- Priming effects – Could exposure to GenAI reshape human linguistic practices? This may be impossible to answer, but the potential for a “feedback loop” is real and worth investigating at a larger scale.
Read the entire interview
