In early days of computerization, companies tried to dodge liability due to "computer errors". That didn't work, and I hope the "It was the AI, not us" never gets allowed either.
Not yet. The Inquiry is still taking evidence. They haven't taken evidence from the big-hitters yet, that begins in April.
In fact the Inquiry doesn't "hold humans accountable"; but they can compel witnesses, who testify under oath.
Gerald Barnes is the Fujitsu engineer who stated in the prosecution cases that Horizon was reliable ("robust"). His testimony to the Inquiry has been delayed, because on the morning he was supposed to testify, the Post Office "discovered" a million or so emails that they'd failed to disclose. So he'll be on the stand in April, along with the senior execs.
The Inquiry videos make quite pleasant watching; the lawyers and the judge are immaculately polite, there are no trick questions, and it's all about finding out what happened. I'm looking forward to seeing the senior execs on the stand.
The police are following the Inquiry; nobody's been charged, and my guess is they'll hold off on charging people until the Inquiry is over (that's part of the purpose of statutory inquiries). So the succession of Post Office Ministers that have overseen this disgrace will all be out of office by then.
> With the British Post Office, the issue is whether or not a software system is inscrutable
Not sure what "inscrutable" means in that context. Is it supposed to mean it can't be scrutinized?
A law was passed some years ago that says evidence obtained from a computer system should be accepted as true, unless evidence is provided that opens it to question. That means, in the Post Office case, that postmasters couldn't demand that the Post Office prove that Horizon was working correctly. They had to prove that it was defective, which was difficult; they were kicked out of their shops, and denied access to their own records, including the Horizon terminals they had been using.
Of course if the chatbot cannot what is the point of them - I'll have to get something else to verify anything the chatbot says. Sure the chatbot can say "hello" in 10,000 words or whatever, but it can't do anything useful.
It worked in the British Post Office Scandal: https://en.m.wikipedia.org/wiki/British_Post_Office_scandal