By 2025 end, will it be generally agreed upon that LLM produced text/code > human text/code for training LLMs?

By 2029 end, will it be generally agreed upon that LLM produced text/code > human text/code for training LLMs?

Will LLMs' loss function achieve the level of entropy of human text by the end of 2030?

Will any widely used LLM be pre-trained with abstract synthetic data before 2030?

Will there be any simple text-based task that most humans can solve, but top LLMs can't? By the end of 2026

Will LLMs be better than typical white-collar workers on all computer tasks before 2026?

Will we have a popular LLM fine-tuned on people's personal texts by June 1, 2026?

Will LLMs be used for academic peer review by 2030?

Will LLMs become a ubiquitous part of everyday life by June 2026?

Will an LLM improve its own ability along some important metric well beyond the best trained LLMs before 2026?