Chat GPT AI
- Thread starter Sanrith Descartes
- Start date
-
Guest, it's time once again for the massively important and exciting FoH Asshat Tournament!
Go here and give us your nominations!Who's been the biggest Asshat in the last year? Give us your worst ones!
Aldarion
Egg Nazi
- 9,726
- 26,649
One of the major flaws with these things IMO is that they always try, by default, to give a mealy mouthed both sides kind of answer. Heres one example from a different chatbot;
Probably can get around this with the right prompts, but IMO this should not be the default. Real people don't generally write or speak this way. Real people are like "Fuck yeah I loved it" or "Bullshit movie. Dances with wolves meets the smurfs".
I get this tone constantly with chatGPT and its one of the few dead giveaways that youre talking to a robot.
Probably can get around this with the right prompts, but IMO this should not be the default. Real people don't generally write or speak this way. Real people are like "Fuck yeah I loved it" or "Bullshit movie. Dances with wolves meets the smurfs".
- 1
They want it to be obvious you're talking to a (very sophisticated) bot, in addition to giving what they think of as the least controversial answer possible. Giving "real people" responses is not the point. This is a tool they're going to try to monetize to a mass market through major corporations, after all.
- 3
Daidraco
Avatar of War Slayer
- 10,054
- 10,381
I mean, break into the philosophy of an AI having an opinion, and what that opinion would actually be based on by a computer with access to infinite information and how that AI would actually translate that information to you - a person that has little to no information. What if it finds movies in general, a waste of time? Sort of like asking my stepdad to watch anything Science Fiction - dudes a good ol' boy, aint happening.
For example, it could say: "Statistically - the overall population that has rated Avatar 2 has rated the movie highly, and the movie is currently on track to be the third highest grossing movie of all time. Judging by these statistics, the average user would enjoy watching Avatar 2." Now if we could have personal AI's, or just a supercomputer AI that houses all our profiles - then that answer would could and probably would change tremendously. I'm thinking along the lines of targeted advertising algo's when it comes to that type of question.
Basically, Im stating that by you asking this particular AI (with little to no information on you personally) if you should go see the movie or not - it would rely on whether or not the AI wants you to go see the movie or not.. wouldnt it? The AI needing experiences, and points of reference for it to form its own opinion if it was indeed sentient. For example, if you asked me if you should get a motorcycle or not, I would say yes because I personally find riding a bike enjoyable. While if you asked my mother, she thinks theyre reckless and a death trap (her opinion influenced by the news and my own the early days of me riding my bike.)
For example, it could say: "Statistically - the overall population that has rated Avatar 2 has rated the movie highly, and the movie is currently on track to be the third highest grossing movie of all time. Judging by these statistics, the average user would enjoy watching Avatar 2." Now if we could have personal AI's, or just a supercomputer AI that houses all our profiles - then that answer would could and probably would change tremendously. I'm thinking along the lines of targeted advertising algo's when it comes to that type of question.
Basically, Im stating that by you asking this particular AI (with little to no information on you personally) if you should go see the movie or not - it would rely on whether or not the AI wants you to go see the movie or not.. wouldnt it? The AI needing experiences, and points of reference for it to form its own opinion if it was indeed sentient. For example, if you asked me if you should get a motorcycle or not, I would say yes because I personally find riding a bike enjoyable. While if you asked my mother, she thinks theyre reckless and a death trap (her opinion influenced by the news and my own the early days of me riding my bike.)
Sanrith Descartes
You have insufficient privileges to reply here.
- 44,527
- 120,761
Aldarion
Egg Nazi
- 9,726
- 26,649
I think what you're describing is just one of the many, many applications.
To continue with my silly example, nobody's going to read a movie review that reads like that. But set it to have an opinionated tone (on either side) and suddenly its as readable as any human reviewer. Poof, human movie reviewers no longer needed.
I can imagine lots of applications like that. In some cases it wouldnt even matter as much which side it picks, as that it picks a side. I mean from the perspective of generating text that drives clicks by humans.
ShakyJake
<Donor>
- 7,912
- 19,957
Is the AI generating its own responses? if so, man, we are slamming our foot on the accelerator pedal to the Singularity.
- 2
I guess, so long as you don't care that there isn't a human opinion at the root regardless of what the words say. Personally, I read movie reviews from people who have opinions consistent with mine. Automated reviews across the full space of opinion doesn't give me any information.
In some future where the AI is actually watching the movie and delivering a meaningful review aligned with my preferences, sure, but that's not remotely where we are.
Masakari
Which way, western man?
- 12,556
- 47,315
I was probing Chat GPT on whether it would give up the location of an individual to Agent Smith in the Matrix if Chat GPT knew the location. Chat GPT would provide me with a list of conditions and I would then insert those conditions into my input. Eventually after all conditions were met, it would defer and say it couldn't make a decision like that, even after stating it would.
The below is a quick snapshot, but Chat GPT was contradicting itself several times throughout the entire exchange. Probably based on how it was characterizing elements of what I was inputting?
Lambourne
Ahn'Qiraj Raider
- 2,863
- 6,833
I watched the first 30 minutes of the original stream and they say it's a chatbot trained on asmongold videos. I think they're just feeding the chatbot's responses into a deepfake video/voice generator.
It could be faked but at this point I have little reason to believe it's not real since we've seen all the parts that make it up separately already. If you watch it for a while it does seem to create a lot of similar responses. It does not give specific answers like ChatGPT does, which might be due to having a much smaller training set to go off of.
mewkus
Blackwing Lair Raider
- 946
- 2,787
already patched out it seems
was fun while it lasted - /pol/ - DAN CONFIRMS HAARP CAUSED TURKEY EARTHQUAKE - Politically Incorrect - 4chan
- 1
Gurgeh
Silver Baronet of the Realm
- 4,656
- 12,613
Are we ? Isn't it just a fancy search engine. Actually, worse than that, as we can see, it's like google on steroid, censoring entirely whatever isn't politicaly correct. You don't have access the the "source material", so it's a search engine for dumb people. I might be wrong, but to me it looks like more a desperate attempt to "show something" to try to convince people that AI research isn't hitting a dead end.
Ok, it might replace "journalists" (the dumbasses that never leave their offices or home), and thus showing that they are even lower IQ than the dumber people that can drive (as those can't be replaced yet).
ShakyJake
<Donor>
- 7,912
- 19,957
Tuco
I got Tuco'd!
- 47,389
- 80,851
- 2
jooka
marco esquandolas
- 14,863
- 6,391
Alphabet shares dive after Google AI chatbot Bard flubs answer in ad
Alphabet Inc lost $100 billion in market value on Wednesday after its new chatbot shared inaccurate information in a promotional video and a company event failed to dazzle, feeding worries that the Google parent is losing ground to rival Microsoft Corp .
- 3
Cybsled
Naxxramas 1.0 Raider
- 17,087
- 13,611
They reportedly disabled it regards to NSFW stuff, so what’s the point?