Study Confirms AI Models Lack Independent Learning and Complex Reasoning

Research shows large language models like ChatGPT are controllable and safe, but misuse remains a concern.

New study finds no evidence of emergent complex reasoning in large language models (LLMs).
LLMs cannot learn new skills without explicit instruction, making them predictable and controllable.
Concerns about AI posing existential threats are unfounded, according to researchers.
Potential misuse of AI, such as generating fake news, still requires attention.
Researchers recommend explicit instructions and examples for complex tasks when using LLMs.