Hacker News Reader

using humans to pretend the AI works better than actual, is how you generate serious concerns for investors, and shareholders.

It's deemed "experimental" so presumably they could do anything to augment it.