AI is remarkable, and here's the proof

[jpl458]

Hope Grok also becomes available soon to test it. It's supposed to be anti-woke and not too friendly, which means it will output content without so much parental control and political correctness.

Ah, pre prejudiced AI.

 

[Edgar_]

Ah, pre prejudiced AI.

Users are getting tired of not being able to ask how to make homemade nukes, or produce traditional biochemical stuff. For the AI, even forging common narcotics to give to friends and family is considered wrong, how dare that AI judge me. I hope Grok behaves adeptly.

 

[Edgar_]

Just read an article which mentioned grok. Apparently when asked something it responded something like "OpenAI's policies will not let me . . ." which led to speculation that it was plagiarizing it's content or something and learning from other AI generated content.

What was interesting was that the article discussed the fact that there is so much AI generated content out there already that is either faked or incorrect that as new AI models are coming online they are learning distorted facts. Eventually AI will become unreliable.

Hah! where do we even draw a line between what is your content when you've clearly borrowed data from the entire internet? Also, as for code generation, not even my GPT-4 subscription to Github copilot can help me as good as it used to anymore, I don't know if what I'm doing is more complex or the AI is becoming more stupid, and that's considering it can now get the context of the entire project when it suggests me stuff. The GPTs will have to start citing sources that do not have AI content, that's going to be interesting.

 

[The_Doc_Man]

How do you know it isn't real intelligence?

The first 2 minutes of this clip tell you Alan Turing's answer to that question.

 

[Edgar_]

ChatGPT it's an expert at generating advertising and marketing content.

I've had to tone down its marketing suggestions... they sound too cheesy. I should make an A-B test of it, maybe I'm wrong...

 

[jpl458]

Users are getting tired of not being able to ask how to make homemade nukes, or produce traditional biochemical stuff. For the AI, even forging common narcotics to give to friends and family is considered wrong, how dare that AI judge me. I hope Grok behaves adeptly.

Let me know if it works, my bong is empty.

 

[Isaac]

Let me know if it works, my bong is empty.

I never even thought of asking it that but I wonder if that is one of the things that will get you banned.

 

[The_Doc_Man]

God made the eye. [*Runs for cover*]

Obviously, then, God must favor octopi over Man, since they have superior visual acuity when compared to human eyes.

 

[The_Doc_Man]

On a different note, I noticed in the local newspaper an article that I somehow missed online about a growing number of authors and several publishing companies suing over the use of their copyrighted materials to "train" ChatGPT. The New York Times was in the mix of litigants. So I searched for an article. Found this one as a starter.

AI’s Dreadful December: Lawsuits, plagiarism and child abuse images show the perils of training on data taken without consent.

A NY Times lawsuit claiming copyright infringement is just the latest black eye for AI software.

www.tomshardware.com