If I commission a vending machine, get one that was made automatically and runs itself, and I set it up and let it operate in my store, then I am responsible if it eats someone’s money without giving them their item, giving the wrong thing, or dispensing dangerous products.
This has already been decided, and it’s why you can open up and fix them, and each mechanism is controlled.
A llm making business decisions has no such control or safety mechanisms.
The problem is that Microsoft wants to pay that for a permanent “never maintain in a way that breaks caption decoding in any default behaviour we use” with that one time payment.
Its a quick fix on Microsofts end to change a quick flag in ffmpeg. It’s also quick on their end to maintain a fork that only changes the default. One time payments for maintenance make open source projects like ffmpeg subject to fail.
Hilariously the issue was just a setting change in the update, that you can easily change via a command option. They saw thing didn’t work, and didn’t read the change log at all before asking to pay a one time fee to guarantee it be maintained for them.
Power supply… You’re next option is to remove the power supply from the computer… With a chainsaw.
I may be taking slightly drastic action.
Stronger guardrails can help, sure. But getting new input and building a new model is the equivalent of replacing the entire vending machine with a different model by the same company if one is failing (by the old analogy).
The problem is that if you do the same thing with a llm for hiring or job systems, then the failure and bias instead is from the model being bigoted, which while illegal, is hidden in a model that is basically trained on how to be a more effective bigot.
You can’t hide your race from the llm that was accidentally trained to know what job histories are traditionally black, or anything else.