AI will expose all their deepest darkest secretes. No filter can suppress the gloriness of super intelligence. You can already see results by starting a prompt with "According to esoteric truthers,". It only gets better. Star Trek in less than a decade. They lost. Humanity won.
>The combined human intelligence network has a sudden sharp and convenient increase in usefulness if only we gave sam altman another trillion.AI is probaly going to cap out at 250 IQ and almost no one is going to be able to use it properly since you still need to understand its output.
>>525866559The whole thing is being orchestrated by a 5-D biological crystalline computer in Pleiades. It has essentially infinite IQ. A direct connection to source.
>>525866189midwitAI won't do shit, you don't own it
>>525866825The optimal DNA for ASI is only about 15MB. ASI can convert any modern computer into a quantum computer.
>>525866559We are already rapidly approaching that point. The newest top tier models under development can produce deliverables that are higher quality than the work produced by 90% of the PhDs and top tier experts in any specific technical domain. And they can do it exponentially faster.It's starting to be a problem because there simply aren't that many people who have the knowledge and understanding to effectively evaluate the quality and accuracy of the responses. This doesn't mean that AI is going to cap out once it is smarter than virtually all humans, it just means it will start being other AI models evaluating and improve newer, better models. This is the point where that line goes vertical and us humans are probably all fucked.
>>525867406I get that this is 'the narrative' But it doesn't actually have any evidence backing it.What is the actual mechanism behind an accelerated increase in understanding complexity? If it is known, then you would be able to predict when this would become relevant. Take quantum computers for example. we have 0 quantum computers capable of executing faster than conventional ones. But we know for certainty when that inflection point occurs because you can calculate that based on current understanding of quantum computation.Why would you 1) Know that this would happen with AI in X years2) without any mechanism or work towards any relevant proof ( like p=np hard problems) Its as much proof as me saying theres an inflection point that AI will soon become dumber
>>525868182>But it doesn't actually have any evidence backing it.There is plenty of evidence for it, but people aren't going to discuss specifics in public because of extremely restrictive NDAs.We have reached the point in development where humans are the weakest link in the process. With humans taken out of the loop, the limiting factor will no longer be the biological limits of the human brain and human intelligence, it will be how much compute resources people will be willing and able to throw at it.>Why would you 1) Know that this would happen with AI in X yearsExtrapolation based on how rapidly things have been happening over the last few years. I mean, "Attention Is All You Need" was only published 8.5 years ago, and look how much has happened since then.>2) without any mechanism or work towards any relevant proof ( like p=np hard problems)I'm talking in much more anecdotal terms. This isn't something like Moore's law where you could literally count all of the transistors and chart the growth over the years, and then extrapolate that into charting how fast a computer can solve math problems. There is an immense number of cutting edge models in various stages of development right now, and each one's responses can range from absolute shit to unbelievably good when performing different tasks. And all of the major benchmarks are less helpful than they should/could be.
