AI alignment is a subject of nice curiosity to me. We’ve made a number of progress in AI over the previous couple of years, however not so much in AI security and alignment. It’s not like there was no work on this area, however it’s minuscule in comparison with how briskly the remainder of the sphere is rising. To grasp what precisely is occurring in AI, we have to zoom out a bit of bit and have a look at the sphere from afar to grasp the underpinnings of the present developments. So, with out additional ado, let’s start from the very begin.
AI will not be a brand new subject and has been round for the reason that Nineteen Sixties, however it appeared utterly totally different again then. It is just with the discharge of ChatGPT it reached the laymen. Earlier than it was a cool factor just for a bunch of researchers and some firms. However now it’s in all places, even within the locations we don’t even take into consideration. It has really reached the complete globe and has turn out to be way more interconnected identical to the web.
Early AI was working in silos, now we’re slowly and certainly transferring in the direction of extra of an interconnected world system of intelligence. And to be sincere, this isn’t very totally different than how human intelligence developed, in a neighborhood.
Early makes an attempt at AI tried to create information programs which may be capable of ingest the complete world’s information with enough reminiscence. We labored so much on symbolic programs to create world information. However this experiment failed rapidly and we began working in the direction of…
