zlacker

A tech challenges in machine learning these days is teaching the machines to explain why they made their decisions. With Google's commitment to 100% lights-off handling of terms-of-service violations, it seems unlikely that non-Google entities will get any decent explanations unless there are revenue implications for Google.

Big fines for violations? Maybe. But they have more lobbyists than that rest of us to resist legislation. Won't happen without a mass political movement (in the US at any rate).

How about a review department at Google?

We could pay US$200 for a human review of the situation, with a reasonable SLA (maybe two working days), with a promise of a refund if they determine the error was theirs.

Possibly a larger fee for a more aggressive SLA?

Possibly a subscription-style fee for publishers of mission-critical stuff? (Meaning, critical to the publisher's mission, not Google's mission.)