No one can convince me that the timing of "spending billions on anti-misinformation" 6 months before GPT goes consumer was not a long coordinated collaboration.
Good observation to think about. Censoring data by hands of GPT trainers is impossible for such data volumes. So they throw that job to social networks owners to get more or less clean trained model.
If that the case, social networks should had been cleaned not only from usual anti-narrative, but also from some much less noticeable and known things that could heavily ruin model intended results if being interlaced with other training data.
I think they tried to do that, but don't think they was able to do that thoroughly, so first - there should be some unexpected topics banned and censored along with usual non-coronahoax, non-liberal, non-tolerant things, and second - current models could still be harmful for creators.
Good direction to look at, really. What else was wiped out from social networks during censoship campaign?
You are looking in the wrong direction -- why clean the data when you can make it obsolete and timed out.
Ayh have seen digital professionals go from searching and exploring the internet for answers to a problem to asking a single chatbot and accepting the answer.
They are conditioning a population to accept one answer results bcz "its sounds like a human" and you will see more and more chatbot isolation for most applications and searches.
There is no coincidence that mere moments.. MOMENTS after GPT went consumer that all the search engines deployed it.. as if it was in the chamber ready to blow. It was always about turning peasants tunnel vision.
Phones will come with it. Computers are coming with it. Windows 12 will now be 85% chatbot.
Good observation to think about. Censoring data by hands of GPT trainers is impossible for such data volumes. So they throw that job to social networks owners to get more or less clean trained model.
If that the case, social networks should had been cleaned not only from usual anti-narrative, but also from some much less noticeable and known things that could heavily ruin model intended results if being interlaced with other training data.
I think they tried to do that, but don't think they was able to do that thoroughly, so first - there should be some unexpected topics banned and censored along with usual non-coronahoax, non-liberal, non-tolerant things, and second - current models could still be harmful for creators.
Good direction to look at, really. What else was wiped out from social networks during censoship campaign?
You are looking in the wrong direction -- why clean the data when you can make it obsolete and timed out.
Ayh have seen digital professionals go from searching and exploring the internet for answers to a problem to asking a single chatbot and accepting the answer.
They are conditioning a population to accept one answer results bcz "its sounds like a human" and you will see more and more chatbot isolation for most applications and searches.
There is no coincidence that mere moments.. MOMENTS after GPT went consumer that all the search engines deployed it.. as if it was in the chamber ready to blow. It was always about turning peasants tunnel vision.
Phones will come with it. Computers are coming with it. Windows 12 will now be 85% chatbot.