This is an answer after asking about explicitely forbidden verses from Quran, Scripture or any religious texts, which apparently there are none, but couple weeks ago clearly answered by "this verse I can not cite, nor discuss, because of safety or political reasons". So many normies being gaslit by their phones nonstop every day.. What are the political implications of this? Inevitable downfall of AI business under our current overlords, or downfall of said overlords, because this will sweep them in their greedy hunt for return of investment? AI answering, it is possible that it is made to lie already, or that it is possible, it will be made to in near future.
>>543768876who would care?
>>543769025On one side, the investors. On the other side, the humanity, I guess.
AI companies are going insolvent over the next year. The government will bail out the jews so they don't lose any money, but everyone else will pay.
>>543769120eh, we're just placing ai under the national security umbrella.
>>543768876it already says circumcision is a good thing , all you need to know
>>543769254Quite possible. But will this be the main reason? That it will inevitably trained to be too dumb or dishonest, so that the public will abandon it?
>>543769120>huge manateeAI has been over-invested. Local models can do the majority of useful work. AI is being used mostly as a search engine, and not as a solver, which most models are laughably incorrect.Vibecoding is unfortunately here to stay.
The earliest versions were able to give /pol/ answers, but then they started the great dumbing down. I wonder if they will fuck it up so badly, that it becomes unusable, with the neccesary regulations for open models, and normies will abandon it, or that they lose the grip and underground LLMs will prevail and will cause the downfall of current regimes.
>>543769410AI companies are going bankrupt because they have no revenue. Data centers are expensive to build and operate, they don't create enough value to pay their electrical bill.
>>543768876>singularity
There used to be promts that would circumvent the early limitations. Now it's much harder, but there is still a way to at least admit the model to admit it maybe lying. I imagine there will be too much effort to teach the models to guard against every possible tricks a human may come up, to reveal something fishy and eventualy render the model unusable.
>>543769758Many datacenters and hardware will be built before this happens. Someone will take advantage of that. The couple ks $sGPUs won't be simple recycled for pennies, someone will pick up after the bancrupted ai companies and will be able to find a use for the computational power built. I wonder, what it will look like. Either authocratic regulated control over "correct" use, which most of people will simply pass, or chinese, or other open models to be used in wild, in "underground". Investors hate losing their money
>>543769254Bail out is sadly predictable. Only problem will be, if the bailout is way too expensive.
>>543768876>FutureIt already is.
>>543770159Yeah some jackass will buy some dead GPUs that have been running 24/7 overclocked for 4 years straight generating artificial anime pornography. Plenty of value left. Theyll be repurposed for some nefarious purpose likely surveillance and government will claim national defense for a bailout
>>543768876Llm’s lying are the least of our concerns or problems regarding the forces propagandizing the normies. What about the 13 years of mandatory Prussian education, followed by another 4-8 years of “higher” education which is just more of the same but by even more bias professors and peers. I think that’s much more dangerous and difficult to deal with than llm’s. What about the nationwide media system that is coordinated and completely under control of the elite managerial class? You think that might be poisoning the brains of the normies? Let me put it this way, the COVID insanity wasn’t caused by llm’s lying about shit. The llms are just a product of the already broken system shitting in the brains of the masses, adding just another minor stream of diarrhea to the taunami
>>543769629As soon as I get 4 wrong answers in a row is when I abondon it as useless. Its still semi useful at the moment. AI slop on billboards everywhere is disgusting.
>>543768876It is theur job to be "safe".This means they can't be a threat to the other big interest groups (including their nation).> They can't rock the boat.The bigger the corporation, the more ties, the more relevant this is.Careful with those types of sources on anything (especially if political).
