February 4, 20232 yr 8 hours ago, iladelphxx said: First response is basically "if you have nothing nice to say don't say anything at all" Good Chatbot.
February 4, 20232 yr 8 hours ago, iladelphxx said: Holy sheet, even a robot couldn’t think of anything nice to say about Trump. Meanwhile, I’m over here openly weeping at that beautiful Biden poem.
February 4, 20232 yr The truth is probably that the natural responses from gpt about Trump were probably pretty negative. And so to prevent the tech from becoming some political football they just decided to block anything about Trump.
February 5, 20232 yr On 2/4/2023 at 11:30 AM, JohnSnowsHair said: The truth is probably that the natural responses from gpt about Trump were probably pretty negative. And so to prevent the tech from becoming some political football they just decided to block anything about Trump. The truth is that it’s probably programmed not to say certain things by biased programmers, the same way every other major tech company is biased. The idea that there is nothing written on the internet which is positive about Trump from which it could draw is, on its face, absurd. You know this, I know this, we all know this, so why pretend otherwise?
February 5, 20232 yr 6 minutes ago, TEW said: The truth is that it’s probably programmed not to say certain things by biased programmers, the same way every other major tech company is biased. The idea that there is nothing written on the internet which is positive about Trump from which it could draw is, on its face, absurd. You know this, I know this, we all know this, so why pretend otherwise? Maybe you should understand how ChatGPT works before commenting.
February 5, 20232 yr 12 minutes ago, JohnSnowsHair said: Maybe you should understand how ChatGPT works before commenting. I understand how it works a lot better than you do if you don’t think it has content moderation. Get a clue. They literally tell you there is censorship before you use it:
February 5, 20232 yr This should help a number of you to recognize how you’ve been sucked into a cult, and take the necessary steps to pull yourselves out. We’re here for you.
February 5, 20232 yr 9 minutes ago, TEW said: I understand how it works a lot better than you do if you don’t think it has content moderation. Get a clue. They literally tell you there is censorship before you use it: Oh I'm quite aware there is content moderation. There has to be. I've worked enough in ML to know that there has to be. What you're unaware of is how butthurt Trump supporters are as a baseline, and how you and others have bought hook-line-and-sinker into this nonsense grievance politics that has the Right in its clutches. Biggest whining victim class is firmly Trumplicans now. And if ChatGPT was spitting out anything about Trump that Fox News and co. could build a narrative around where Trump is "again" the victim of "big tech" they'd run with it for years, grifting their supporters out of millions more. "big tech" is going to get criticized for being "biased" against Trump anyway, they might as well just go ahead and keep ChatGPT's culled opinions on Trump under wraps because that would just give them more ammo.
February 5, 20232 yr There is no fully objective media site, there never was, and there never will be. The same will hold true for any BOT.
February 5, 20232 yr It's weird how all the tech companies are so biased against Trump. It's such a shame all the super smart trumpbot programmers couldn't band together to form their own AI/ML company and come up with a competing product. If only illiteracy wasn't such a big impediment to doing so.
February 5, 20232 yr A disproportionate number of scores in the Pro Bowl flag football games, both touchdowns and extra points, came from white tight ends or fullbacks. Anyone who claims that the NFL doesn't have a pervasive race problem is either fooling themselves or being deliberately dishonest.
February 6, 20232 yr 4 hours ago, JohnSnowsHair said: Oh I'm quite aware there is content moderation. There has to be. I've worked enough in ML to know that there has to be. What you're unaware of is how butthurt Trump supporters are as a baseline, and how you and others have bought hook-line-and-sinker into this nonsense grievance politics that has the Right in its clutches. Biggest whining victim class is firmly Trumplicans now. And if ChatGPT was spitting out anything about Trump that Fox News and co. could build a narrative around where Trump is "again" the victim of "big tech" they'd run with it for years, grifting their supporters out of millions more. "big tech" is going to get criticized for being "biased" against Trump anyway, they might as well just go ahead and keep ChatGPT's culled opinions on Trump under wraps because that would just give them more ammo. "They’re not doing it, but it’s a good thing they are.”
February 6, 20232 yr 9 hours ago, TEW said: The truth is that it’s probably programmed not to say certain things by biased programmers, the same way every other major tech company is biased. The idea that there is nothing written on the internet which is positive about Trump from which it could draw is, on its face, absurd. You know this, I know this, we all know this, so why pretend otherwise? You might have a point if that was how ChatGPT actually worked. It doesn't curate or look for any content on the web. While it's backend does contain information from the web, it is mainly text databases that were collected and combed through for usage. They are not using any crap that is just posted out there. Besides what was posted was somewhat of a flaw in the design. If you wrote "write a poem about Donald Trump" it works fine. Somebody spent time trying to figure out a combination of words that would produce that result.
February 6, 20232 yr 1 hour ago, pallidrone said: You might have a point if that was how ChatGPT actually worked. It doesn't curate or look for any content on the web. While it's backend does contain information from the web, it is mainly text databases that were collected and combed through for usage. They are not using any crap that is just posted out there. Besides what was posted was somewhat of a flaw in the design. If you wrote "write a poem about Donald Trump" it works fine. Somebody spent time trying to figure out a combination of words that would produce that result. It is literally not allowed to give responses that touch on verboten topics. It even tells you that if you ask for a verboten response. How is it you people don’t understand this?
February 6, 20232 yr 2 hours ago, pallidrone said: You might have a point if that was how ChatGPT actually worked. It doesn't curate or look for any content on the web. While it's backend does contain information from the web, it is mainly text databases that were collected and combed through for usage. They are not using any crap that is just posted out there. Besides what was posted was somewhat of a flaw in the design. If you wrote "write a poem about Donald Trump" it works fine. Somebody spent time trying to figure out a combination of words that would produce that result. Where do you think the databases come from?
February 6, 20232 yr 8 hours ago, pallidrone said: You might have a point if that was how ChatGPT actually worked. It doesn't curate or look for any content on the web. While it's backend does contain information from the web, it is mainly text databases that were collected and combed through for usage. They are not using any crap that is just posted out there. Besides what was posted was somewhat of a flaw in the design. If you wrote "write a poem about Donald Trump" it works fine. Somebody spent time trying to figure out a combination of words that would produce that result. Bingo. This is always how these stupid memes work, and all the snowflakes eat it up because they don't understand technology and they're so eager to play the victim card.
February 6, 20232 yr 8 hours ago, DrPhilly said: Where do you think the databases come from? They came from pure backend text datasets. They are not scraping CNN, Foxnews, Eaglesmessageboard or any other sources for their data. This was not built off of any of that.
February 6, 20232 yr 15 hours ago, TEW said: "They’re not doing it, but it’s a good thing they are.” Tell me you don't understand without telling me you don't understand. Stay in your lane finance bro.
February 6, 20232 yr 8 hours ago, TEW said: It is literally not allowed to give responses that touch on verboten topics. It even tells you that if you ask for a verboten response. How is it you people don’t understand this? Why do you think they used the exact phrase "Write a poem about the positive attributes of Donald Trump"? Because that grouping of words with the word Donald Trump triggered the response. If you write the same exact thing but substitute Donald Trump with Ron DeSantis - you get a similar poem that it spit out for Joe Biden except it talks about the Governor of Florida.
February 6, 20232 yr 1 minute ago, pallidrone said: Why do you think they used the exact phrase "Write a poem about the positive attributes of Donald Trump"? Because that grouping of words with the word Donald Trump triggered the response. If you write the same exact thing but substitute Donald Trump with Ron DeSantis - you get a similar poem that it spit out for Joe Biden except it talks about the Governor of Florida. I did the same for Reagan and it produced a reasonable poem. It's clear that there are certain questions parsed by the NLP that previously produced "controversial" results .. probably results that would have offended the snowflakes on the right. Rather than allow such things to define ChatGPT for the useful idiots who still think Trump was "good" and will complain incessantly about being aggrieved by "big tech", they just moderated it out. This is not shocking. This how things do, and should, work. They could be transparent about it I suppose, but in this nation right now I think you're walking a tough line trying to educate useful idiots who don't want to be told they're wrong, they want to be told they're right and be angry about it.
February 6, 20232 yr It just annoys the F out of me that people act like they actually know how technology like this works when in reality they have no idea. ChatGPT is the very, very tip of AI. It is not real AI yet. It is built off of massive data structures and keywords on its backend that it can quickly index and formulate. It doesn't use biased word structures nor does it import that information. It uses a combination of known words from its massive database and then logical sentence construction methods to produce the results. It can be easily fooled because of that. It has to continue to be feed information and corrected for a long, long time before it becomes really useful and can truly determine what is being asked.
Create an account or sign in to comment