[chrisseaton]

> For the speaking part it's humans evaluating the results

That's not what I understood from the article, which says

"uses voice recognition technology to test speaking ability, with audio recordings then marked by a “scoring engine”"

"Other test providers have said they use human assessors to determine if people are competent speakers of English"

Both of those imply there aren't humans evaluating the results at this test provider.

That's the entire point of the article.