What Explore searches

Explore searches one library: Inquiring Lines. That means about 2,700 papers on AI and language models, more than 2,700 synthesis notes written from them, and more than 11,000 inquiring lines, the questions the library has worked through. It doesn't search the web. It can only show what the library holds.

A language model reads your post and lists the questions it raises: its claims, the assumptions under them, ideas it mentions in passing, and the question it puts to readers. Each question is rewritten four or five ways in the words researchers use, including once from an opposing or neighboring idea. Every version is searched across the whole library. For each question, the page shows where the research stands and one to three studies to start from.

What the results show

• Every item is linked. Each item is a paper or synthesis note in the library. A sentence that names a number or a name its source doesn't contain is removed before the page shows.
• It splits the post into lines of inquiry. You see the questions the post raises but doesn't pursue, not one answer to the whole thing.
• It looks for evidence against the post. The searches are written to find studies that complicate the post's assumptions, not only ones that back it.
• It reaches into other fields. Research from economics, psychology or philosophy shows up when it bears on the question.
• It gives you lines to follow. Each inquiring line opens onto the research behind it, so you can keep going on your own time.
• It says when the research is thin. Some questions have little direct research yet, and the page says so.

It doesn't judge whether the post is right. It shows you where the research stands so you can judge for yourself.