the fundamental issue with AI has ALWAYS BEEN that it CAN'T do precisely what you intend because prompts are lossy compression you can't control the decompression of. that is what the technology **IS** now they are rebranding that as "lack of alignment" as if it wasn't present from their inception
If you’re a normal person whose not in tech I wrote an essay explaining where I think we are at with the news of frontier labs deciding to slow down model progress unilaterally together and what it really means for regular humans. It tries to start from first principles and explain things well.
10 reposts36 likes