The author suggests that language models like ChatGPT are at a stage where they've started working effectively, but the novelty is already wearing off. He points out that text generators mimicking styles aren't new and that the novelty of ChatGPT screenshots has declined. He ends by posing a question: Is the ability of language models to synthesize style on demand impressive, or unsporting?
Key takeaways:
- Novelist and essayist Walter Kirn criticizes AI chatbots like ChatGPT, arguing that their ability to generate text isn't as impressive as it's often made out to be.
- ChatGPT's ability to mimic different writing styles is compared to a robot pitcher mastering every pitch, suggesting it's more about machine brute force than genuine creativity or ingenuity.
- The article draws parallels between the development of AI in games and language models, noting that while it was initially surprising that computers could excel in these areas, the novelty and impressiveness have worn off over time.
- Despite the advancements in AI language models, the author suggests that there's a potential for the novelty and interest in these models to decline, similar to how interest in computer chess has waned over time.