Raw LLM Responses

10:45 The first assumption is based on an inherently wrong understanding of what the thing we currently label artificial intelligence is. The Thing ™ is trained on data sets so impossibly large that no human could filter how much of that data is generated by smart people vs stupid. As such Tt (The thing/it but funny) is fed on data (including wrong data) that perfectly represents an IQ of 100. Because Tt is basically putting thousands of people in a room and have them consult on a task but make it without one problem of human communication and therefore near instant. And if you ever worked on a group project of any significant size, you know the smart people don't win. They are never the majority. So that's the first big logical fallacy of Tt: it is trained on the average of humanity. The second big fallacy of Tt is what happens if you stop training it on human data. This is something I think may have already started happening. First off, why is training Tt on data generated by Tt bad. Second, why is it probably happening anyway. First: if you train Tt on Tt data, then hallucinations become sources and lies the truth. It will never learn how to draw hands and only get worse at it as it feeds on it's own bad examples. This is again caused by the top much data to be controlled/proofread issue as above. Second: people lie. They lie about using Tt so it becomes impossible to perfectly filter data generated by Tt out of the training pool even if you prohibit training Tt on Tt. Plus human data generation goes down simultaneously as generating with Tt becomes the norm. It doesn't matter how small the leak is initially because it will widen itself by changing the assessment standard in small increments - that is assuming that you try to have Tt filter Tt generated stuff out of it's own training data cuz no one else can even hope to. Again, enshittyfication happens. The only way to avoid that is to have the deciding factor be that all training data precedes the introduction of Tt to the world. That puts a definitive cap to how much it can learn and therefore how much it can grow. It will put a hard stop to innovation preventing ANY new tech from getting used. With all that being said tho, that does NOT mean that the scenario of the jobmarket that you described will not be happening. In fact it makes it even more likely. Statistic dictates that decisions will be felled by IQ100. To IQ 100, Tt being IQ100 but faster LOOKS like it is smarter than them. The buy the lie. Just like they bought into calling it AI and then thought it capable of things that it isn't because of the name they themselves have given it. Anyway thx for reading this adhd rant to the end.