What I’ve been able to glean, just from overhearing a conversation about it on another site, he appears to be a sort of whistle blower relating to some of the world’s wealthiest and most powerful people.
Oh god get out of here with those unfounded gullible conspiracy bullshit rumors.
“Guys, guys, THEY don’t want you to know this. It’s the global elite trying to cover up their cabal, not a funky technical glitch like the SolidGoldMagikarp token bug from a while back. Trust me bro, I heard about it on some other website I won’t link or name. Presumably, the people on there are totally trustworthy.”
It’s always funny to me how people will readily agree that monied interests are deliberately working against the general population’s interests, but the instant you actually start to name who those people are you get dog pilled with people calling you crazy.
Here’s the thing though - token glitches don’t exhibit quite the same behavior. This appears to be the model’s output filter being hard coded to cut off messages when they generate these names.
Like yeah, it’s probably just “OpenAI was legally required to prevent these names from being generated”, but still? Weird
Of course the big question is who is David Mayer and why does the utterance of his name break ChatGPT? Numerous theories have already cropped up. As the online world quickly figured out, googling “David Mayer” results in David Mayer de Rothschild, heir to the famed Rothschild banking family, who is an adventurer and environmentalist.
What I’ve been able to glean, just from overhearing a conversation about it on another site, he appears to be a sort of whistle blower relating to some of the world’s wealthiest and most powerful people.
Oh god get out of here with those unfounded gullible conspiracy bullshit rumors.
“Guys, guys, THEY don’t want you to know this. It’s the global elite trying to cover up their cabal, not a funky technical glitch like the SolidGoldMagikarp token bug from a while back. Trust me bro, I heard about it on some other website I won’t link or name. Presumably, the people on there are totally trustworthy.”
It’s always funny to me how people will readily agree that monied interests are deliberately working against the general population’s interests, but the instant you actually start to name who those people are you get dog pilled with people calling you crazy.
Here’s the thing though - token glitches don’t exhibit quite the same behavior. This appears to be the model’s output filter being hard coded to cut off messages when they generate these names.
Like yeah, it’s probably just “OpenAI was legally required to prevent these names from being generated”, but still? Weird
As per OP article…