funnymonkey on Nostr: Confession: I read a Kevin Roose article. I deserve scorn and mockery. And: it ...
Confession: I read a Kevin Roose article. I deserve scorn and mockery.
And: it contained a reference to research that showed how specifically crafted text strings can be used to game (or theoretically poison) AI results.
Which gets me wondering:
Could prompts like this be mass produced using generative AI and released across a broad enough swath of the web to poison AI output even more than it already is?
Published at
2024-09-03 11:28:19Event JSON
{
"id": "682cab33b2d69d2177d87a44da8b3ea4ce908696ebf8eaddaf0a6f3693b90011",
"pubkey": "94ddca4e0f34bc82b63e377c58e58087698f01585293b2a382e3d4894cc00ea9",
"created_at": 1725362899,
"kind": 1,
"tags": [
[
"proxy",
"https://freeradical.zone/users/funnymonkey/statuses/113073383006803381",
"activitypub"
]
],
"content": "Confession: I read a Kevin Roose article. I deserve scorn and mockery.\n\nAnd: it contained a reference to research that showed how specifically crafted text strings can be used to game (or theoretically poison) AI results.\n\nWhich gets me wondering:\n\nCould prompts like this be mass produced using generative AI and released across a broad enough swath of the web to poison AI output even more than it already is?\n\nhttps://nfts.freeradical.zone/media_attachments/files/113/073/377/688/694/316/original/f224bfa2958102b8.png",
"sig": "a5d7cb0364c980bccc21a732a67c8f56469dc59c42e9c7a346ec88ab4ac24dc0acd8303e35b04b2c2f206b8632995bf0f3756896aa8d74c35bbfc12a1077576a"
}
