Software developer at work and home (C++ and Python). Sometimes claims to know about security. Dabbles in electronics. Talks to vehicle CAN networks? He/him.
@Alph4betSoup Seems right. People put up with a lot of changes to the UI and course structure, and more aggressive premium tier pushing. It got people onto subs for a while, but people werent happy. Then the AI debacle plus some other things they did was the final straw for a lot of people.
@metahodos I don't think they'd typically be able to cancel the flight, because their routes are planned on the assumption that the planes will be at their destinations at the appropriate time. If they cancel that flight then they'd also have to cancel the flight out from that destination.
@SanScintilla@drideHQ Sounds about right. Mine literally set on fire in the car within a week of being installed, because of a fault they knew about but didn't inform customers. Their email support basically said "sorry that happened" and then stopped responding.
@drideHQ I backed the 4k and have been waiting for the rear camera before installing everything, but that's obviously not coming so last week I decided to just install the front cam. Today it failed and tried to set my car on fire. Can I have a contact email address to discuss?
@drideHQ As it turns out, you're unwilling to issue a refund for a product which literally set on fire within a week due to a fault which you *already knew about* and have taken no steps to inform customers about. Absolutely appalling.
@DappierAI Your two options are "give us your data in the format we want" or "we will hammer your site in complete disregard to its TOS"?
AI developer ethics as strong as always
@_ban_de AFAIK there's no difference for physical versus streaming when it comes to Fair Use. The way you're using it (e.g. educational use) also doesn't automatically allow use of the entire work. The more substantial portion of the work used the less likely it would count.
@woodtothemax That's why he said "another tool", not "an alternative tool". The usefulness of one doesn't prevent the other also being useful. You don't have to pick one.
@DrSergioCastro If the sandboxed one doesn't have access to the payload page then there is no suspicious activity and it's not useful. If it does have access to the page and you can detect the suspicious activity the page invokes then there's no need for the non-sandboxed LLM.
@ID_AA_Carmack The number is instances of "this major game is about to release and is on code freeze and we've just found out it crashes constantly under our driver, so we have to work around it in the driver" is high enough that you can just assume it happens on every launch.
@mpopv People seem to have a hard time grasping that this is how ChatGPT works. There's a scary amount of it being used as if it was a tool that goes and does some smart research to answer questions. It's a chat bot. It generates an answer that *sounds* right, not one that *is* right.
@tavantallaaja There's a lot of people in the US and UK weirdly attached to iron crosses, I suspect not many of them have been in the Finnish Defense Forces.
@peteskomoroch You still have to confirm the results it gave you. It's entirely possible it just gave you answers that *sound* like good answers, because ChatGPT's entire thing is giving plausible answers not correct answers.