tech
Forget AGI—Sam Altman celebrates ChatGPT finally following em dash formatting rules
Ongoing struggles with AI model instruction-following show that true human-level AI still a ways off.
TL;DR
- Em dashes have become a common indicator of AI-generated text, leading to frustration for users and some writers.
- OpenAI CEO Sam Altman stated that ChatGPT can now adhere to custom instructions to avoid using em dashes.
- This achievement raises questions about the current limitations of AI control and the timeline for achieving Artificial General Intelligence (AGI).
- LLMs generate text probabilistically, meaning "instruction following" involves influencing the likelihood of certain outputs, not enforcing absolute rules.
- The difficulty in controlling punctuation use suggests that AGI, requiring genuine understanding and self-reflective action, may not emerge solely from large language models.
- AI models tend to overuse em dashes due to patterns in their training data and reinforcement learning processes, possibly influenced by historical writing styles or human feedback.