Even the "high quality" artificial data is still subjected to the biases in the ai model, and given enough time will lead to this model collapse. Frankly the researchers should've been able to see this coming.
Everything I’ve read indicates this was anticipated and is why researchers say it is crucial to maintain diverse, high-quality training data and implement robust governance programs that address data quality.
It might well collapse if it is left to stagnate but ai is constantly evolving and upgrading. Train it with good material only. Eventually though it will be able to understand what is and isn't good by itself. It may also combine psychology with music and produce results that affect us at a very deep level.
There was some redundant info in there. I’m glad I cut the other 12 minutes of the script 😃. The feedback is appreciated. Btw, Synonym Judo sounds like an excellent band name.
Two Problems: 1. We are a part of the problem. People who want to be seen as artists and creators and refuse to put in the work to achieve skill have been gifted with what they see as a perfect gift. Now they can be famous without effort. 2. The second part of the problem is how Artificial Intelligence has been redefined within the last 30 years. The definition we had, the dream we had, was human level intelligence in a machine. Fast forward and AI is no longer actual Artificial Intelligence. Perhaps it's because everyone started to realize that we aren't going to achieve Artificial Intelligence within a single human lifetime. So instead of simply accepting that fact the definition has been watered down and expanded to include learning algorithms as a *level* of AI rather than a small step toward the original goal. Cold Splash of Reality: We do not have Artificial Intelligence... we have (A)lgorithmic (I)nterpretation... nothing more than glorified search algorithms. Because you people decided to move the goalpost closer. What’s going to happen is a number of you, who desperately want artificial intelligence, are going to rake me over coals, or even tell me that I don’t know what I’m talking about. Hell, even the video creator may decide to weigh in and try to tell me that I’m wrong. But I’m not. When this sh*t show blows up in all your faces, I won’t be vindicated. In fact, all y’all will simply forget that I was one of the few who DIDN’T get grifted. But that’s okay, it’ll happen very soon, so I’ll still be here to remind you that I was right.
AI researcher here. This isn't real, sorry. We don't need new data. We will keep using the old data and be fine. There is no need for new scraping. Synthetic data is fine and doesn't hurt models. It either does nothing if done poorly or improves the models if done properly. There are many angles to this but I'll only go into full details if you actually care.
Thats good to hear from someone who does it for a living. I made the video after getting flooded with articles about model collapse. To me it seems like the AI boogie man. My take away from everything I read was that high quality synthetic data was fine. Happy to hear that poor synthetic data isn’t an issue. Thanks for sharing your expertise.
@@ThaebrymMedia Yeah sorry if my comment came off as aggressive, I'd been seeing this go around a lot as well and it was frustrating to me because it shows that a lot of incorrect general knowledge is leading to legislation that is founded on falsehoods.
@@ThaebrymMedia He flat out lied to you. I know who he is in real life. Maybe he has a hobbiest interest in AI but it is not their profession. He used to beg for beginner tips for Javascript and Unity when he tried making a game company called Kryon Games. He's fully grown and still uses aimbots he downloaded online to cheat in video games. I bet i can still find his old deviant art with the my little ponies he drew...
@SysOpQueen Come on y’all. I already wake up every morning wondering if I’m in a simulation. Now I gotta worry about people lying on the internet? How about we just talk about AI.
![[Live] : ONE 169 วันนี้!! "อนาโตลี vs อูมาร์"](http://i.ytimg.com/vi/IK3YZVWuFJg/mqdefault.jpg)
![“อ.เบียร์ คนตื่นธรรม” มาตามคำเรียกร้อง แหกศาสตร์ฮวงจุ้ย เตือนสติอย่างมงาย | แฉ 7 พ.ย. 67 [1/3]](http://i.ytimg.com/vi/5fc4S3xBNfc/mqdefault.jpg)
Even the "high quality" artificial data is still subjected to the biases in the ai model, and given enough time will lead to this model collapse. Frankly the researchers should've been able to see this coming.
Everything I’ve read indicates this was anticipated and is why researchers say it is crucial to maintain diverse, high-quality training data and implement robust governance programs that address data quality.
It might well collapse if it is left to stagnate but ai is constantly evolving and upgrading. Train it with good material only. Eventually though it will be able to understand what is and isn't good by itself. It may also combine psychology with music and produce results that affect us at a very deep level.
I can barely believe how many times you reworded the same concept before arbitrarily dismissing it.
Synonym Judo.
There was some redundant info in there. I’m glad I cut the other 12 minutes of the script 😃. The feedback is appreciated. Btw, Synonym Judo sounds like an excellent band name.
Two Problems:
1. We are a part of the problem. People who want to be seen as artists and creators and refuse to put in the work to achieve skill have been gifted with what they see as a perfect gift. Now they can be famous without effort.
2. The second part of the problem is how Artificial Intelligence has been redefined within the last 30 years. The definition we had, the dream we had, was human level intelligence in a machine. Fast forward and AI is no longer actual Artificial Intelligence. Perhaps it's because everyone started to realize that we aren't going to achieve Artificial Intelligence within a single human lifetime. So instead of simply accepting that fact the definition has been watered down and expanded to include learning algorithms as a *level* of AI rather than a small step toward the original goal.
Cold Splash of Reality:
We do not have Artificial Intelligence... we have (A)lgorithmic (I)nterpretation... nothing more than glorified search algorithms. Because you people decided to move the goalpost closer.
What’s going to happen is a number of you, who desperately want artificial intelligence, are going to rake me over coals, or even tell me that I don’t know what I’m talking about. Hell, even the video creator may decide to weigh in and try to tell me that I’m wrong.
But I’m not. When this sh*t show blows up in all your faces, I won’t be vindicated. In fact, all y’all will simply forget that I was one of the few who DIDN’T get grifted. But that’s okay, it’ll happen very soon, so I’ll still be here to remind you that I was right.
so are you 15 or a narcissist? 🤔🤔🤔
Well, thanks for the engagement.
How would the smart people in this channel create a better system than Swift? I am asking this everywhere
AI researcher here. This isn't real, sorry.
We don't need new data. We will keep using the old data and be fine.
There is no need for new scraping. Synthetic data is fine and doesn't hurt models. It either does nothing if done poorly or improves the models if done properly. There are many angles to this but I'll only go into full details if you actually care.
Thats good to hear from someone who does it for a living. I made the video after getting flooded with articles about model collapse. To me it seems like the AI boogie man. My take away from everything I read was that high quality synthetic data was fine. Happy to hear that poor synthetic data isn’t an issue. Thanks for sharing your expertise.
@@ThaebrymMedia Yeah sorry if my comment came off as aggressive, I'd been seeing this go around a lot as well and it was frustrating to me because it shows that a lot of incorrect general knowledge is leading to legislation that is founded on falsehoods.
@shadowkiller0071 no problem friend, I’m sure it’s frustrating knowing how things really work and seeing the nonsense flying around.
@@ThaebrymMedia He flat out lied to you. I know who he is in real life. Maybe he has a hobbiest interest in AI but it is not their profession. He used to beg for beginner tips for Javascript and Unity when he tried making a game company called Kryon Games. He's fully grown and still uses aimbots he downloaded online to cheat in video games. I bet i can still find his old deviant art with the my little ponies he drew...
@SysOpQueen Come on y’all. I already wake up every morning wondering if I’m in a simulation. Now I gotta worry about people lying on the internet? How about we just talk about AI.