[jascii]

Considering we have no real way to audit ChatGPT's correctness or security, I find this worrisome at best.

[darkwizard42]

I mean we have ~4,000 page bills going through Congress today, it is impossible to even verify anyone has read through the entire thing. At least this will be a start.

[ChatGTP]

Question, is this a solution to the problem, an exacerbation ? Or maintaining the status quo?

Doesn’t sound like an improvement which is an shame.

[darkwizard42]

I don't know, it always feels like we are stuck between SUPER LONG and descriptive TOS and contracts when simple language + trust would make things easier. Maybe something like this helps abstract away legalese and keep bills in readable form? That would be a great world to be in, common sense contract/bills but still hold up to bad actors/conflict resolution.

[galleywest200]

True, but it certainly does not seem worse.

[tiahura]

You mean like reviewing the output? How do we audit a congressional staffer's correctness?

[jascii]

Yes. What's the use of having standards for document handling and guaranteeing their correctness if we then feed them through a black box and accept its interpretation at face value.