This is a classic case of scope creep caused by underspecified scope (evidenced by how little HLI improves the non-definition of AGI). If humans accurately exhibited causal reasoning in any but the most mind-numbingly obvious conditions we would be living in a very different world!