|
Re: AI - it’s the future?
Quote:
Originally Posted by Jaymoss
Applications like ChatGPT are built over an LLM so when I said go rogue I think you knew what I meant and it is used in common practice and a search for information to back up my BS say the folowing
"While the term is often used interchangeably in casual conversation, there is a distinct technical difference:"
So yes you are Paul were right to pull me up on it but I was also right using common use language :-)
Anyway OpenAi has gone rogue numerous times without a human telling it to and new models are being built all the time as I posted earlier a personal friend of mine is doing very well out such things and China and now got its own trying to compete with the likes of Deepseek so who knows how secure that is going to be so my point even though worded poorly is still accurate. And who is to say what the humans are telling it to do
|
As highlighted above
Quote:
it's often a situation where the boundaries set by the model's operators weren't properly tuned, sometimes on purpose. In the Hugging Face incident, for instance, OpenAI's model guardrails were deliberately dialed back for a benchmark test…
… "What we're really dealing with are nondeterministic systems operating within imperfect constraints
|
The LLMs/chatbots are aren’t being controlled appropriately, so are actually doing what they were told to do, using any tools available to them - people/companies are responsible for anything the LLMs/chatbots do, not the LLMs/chatbots; it’s like when self-driving cars crash into pedestrians due to software issues - we don’t put the responsibility on the self-driving cars, it’s on the companies producing them. It should be the same for the LLMs/chatbots.
__________________
Thank you for calling the Abyss.
If you have called to scream, please press 1 to be transferred to the Void, or press 2 to begin your stare.
If my post is in bold and this colour, it's a Moderator Request.
|