ChatGPT Fails to Accurately Replicate News Article Style
Gizmodo attempts to have ChatGPT write an article explaining large language models, but the AI struggles with factual accuracy and tone. The experiment reveals the ongoing limitations of the popular chatbot.
ChatGPT attracts over a million users with its ability to generate human-like text, prompting Gizmodo to test if the AI can write an article in the publication's signature style. The news outlet asks the bot to write a casual explainer about large language models, expecting an effortless and accurate result based on the chatbot's previous creative performances.
Instead of a swift success, the experiment turns into a lengthy and frustrating process. ChatGPT repeatedly includes incorrect information about the history of its own underlying technology, which highlights the ongoing problem of factual inaccuracy within the program. The AI also skimps on critical details and produces text that reads more like an OpenAI advertisement than a thoughtful piece of journalism.
The chatbot even forgets its own name during one attempt, denying any knowledge of ChatGPT when asked to complete the writing task. Ultimately, the AI struggles to balance factual information, story structure, and humor, proving that it is not yet capable of replacing human journalists or accurately replicating a specific editorial voice.