Interesting piece in that it names a few researchers like Choi and Lenat. Other folks are active in this field and making progress. The hard part of the task is

Interesting piece in that it names a few researchers like Choi and Lenat. Other folks are active in this field and making progress. The hard part of the task is making decisions that have answers that don’t correspond to an equation or an algorithm. So much work in this field has essentially required building answers using humans to describe things logically or just ask people for answers and then find a way to index the answers in a way that a computer program can predict how to use the index to answer the question. Google has spent a lot of money and research to find ways to use large data sets to find ways to recognize patterns without someone using either of the two methods below. Most things people write down or want to know as answers are the unusual cases or the rare cases. This means that using data collection methods or statistical approaches like GPT-3 can’t extract the answers, because the answers to easy and common questions aren’t recorded as commonly as they need to be, and being able to take a lot of exact examples and know how to generalize them well enough to use the answers to find more is an active research problem.