Probably getting bogged down by endless safety testing and paperwork by now
back
1 comments
Maybe tangential to this comment, but I had been using 3.5 to write contracts, but a couple of weeks back, even when using the exact same prompts that would've worked in the past, 3.5 started saying, to paraphrase, "for any contracts you need to get a lawyer."
Anyone else had this experience? It seems like they're actually locking down some very helpful use cases, which maybe falls into the "safety" category or, more cynically, in the "we don't want to be sued" category.
As they tack on more and more guardrails it becomes lazier and less helpful. One way it gets around admitting that it has instructed not to help you is by directing you to consult a human expert.
I also suspect there’s some dynamic laziness parameter that’s used to counteract increased load on their servers. You can ask the same prompt over and over in a new chat throughout the day and suddenly instead of writing the code you asked for or completing a task, it will do a small part of the work with “add the code to do xyz here” or explain the steps required to complete a task instead of completing it. It happens with v4 as well.
> some dynamic laziness parameter that’s used to counteract increased load
That's a brilliant risk mitigation mechanism! The AI won't recursively self-improve to superhuman levels if it just keeps getting tired of thinking.
Using it for legal matters is (understandably) explicitly against use policy. It's not meant to be used for legal advice so the response you're receiving is accurate - go get a lawyer.
It's a toy, it's not meant to be used for anything actually useful - they'll keep nerfing it case by case, because letting users do useful stuff with their models is either leaving money on the table, taking an increased PR/legal risk, or both.
It saves my org thousands of dollars a week. If that's a toy then take me to the toy store.
Add 'reducing compute use as much as possible' to the list, I'm sure. I can look at some of my saved conversations from months ago and feel wistful for what I used to be able to do. The Nerfening has not been subtle.
Ha, yeah, that's the right description of my feeling. Like, thanks for being so incredibly useful, ChatGPT, saving me hours of copying and pasting bits and pieces from countless Google searches for "sample warranty deed," only to pull the rug out when I wanted to start depending on your "greatness."
I guess, as one of the other replies here said, it was never "allowed" to give you text for a contract according to the TOS, but then it would've been best if it never replied so effectively to my earlier prompts. Taking it away just seems lame.
Edit: bad typing