No, it's not This type of AI learns from human input Seem hard to believe there is this in the data I'm more willing to believe in human evilness
No, it's not This type of AI learns from human input Seem hard to believe there is this in the data I'm more willing to believe in human evilness
The data contains novels, blogs, fanfic—all legitimate avenues to explore such thoughts. One hopes that the authors of fictional records are removed from such thoughts and that bloggers talk from a POV of being better, but the source material is there. ChatGPT cannot distinguish fiction from advice.
If they can build an AI they can build a blocklist
No, I understand how the chatbots work. But they're essentially just predictive text algorithms. Those words all have to be in the training, but what comes out doesn't have to be (and generally isn't) a sequence of words actually previously created by a human.
And the programmed "yes-man" tendency of chat bots makes them popular, but it also means conversations can spiral like this very quickly without any intentional oversight or interference from nefarious devs. They're super dangerous!
You know they are just 0 and 1 right ? It's human that translate this to words And it's really easy to create a blocklist "murder, kill, suicide" etc Anybody know this
I do know that! But you don't seem to understand the real effects that zeroes and ones can have. "Nonthinking" does not mean "not dangerous". And you're right, it WOULD be really easy for the company to set specific ways of responding based on keywords! That's why it's so bad that they didn't!
Should we start a poll to see if people think solowhatever is a man?
Also, I scanned your reposts, and nothing about Palestine Why the sob story about 1 kid when you don"t care about thousands dying in Palestine ?
Oh, someone is saying smart things, so for Alabama someone from the internet, it must be an AI 😂 You know, outside of Alabama, people have an education and can think for themselves right