Cassidy James :eos: :gg: :fh: on Nostr: Personally, I’m still pretty hesitant to touch anything related to image generation ...
Personally, I’m still pretty hesitant to touch anything related to image generation and LLMs for a few reasons: the non-consensual training on copyrighted works, the climate impact of training the models, and the tendency for them to hallucinate/spew bullshit.
However! I’m still really interested in things like ML-powered camera quality improvements, object detection/background separation, and speech-to-text.
Published at
2024-10-02 17:52:31Event JSON
{
"id": "4f22444e4550fd32facf084b9a228ad216869bbbf913bcb212f20cebc3e32dc4",
"pubkey": "155c7af498274a045ba66b5bc073ed110382fed7401a11e2334ce3b749dc8629",
"created_at": 1727891551,
"kind": 1,
"tags": [
[
"e",
"d6d36e1b92723bbcfc201d6f2d04798b7a721073867e7ae4a0de9645f7156956",
"wss://relay.mostr.pub",
"reply"
],
[
"content-warning",
"Nuanced question about “AI”—ignore if you’re unable/unwilling to have a calm discussion"
],
[
"proxy",
"https://mastodon.blaede.family/users/cassidy/statuses/113239100689631498",
"activitypub"
]
],
"content": "Personally, I’m still pretty hesitant to touch anything related to image generation and LLMs for a few reasons: the non-consensual training on copyrighted works, the climate impact of training the models, and the tendency for them to hallucinate/spew bullshit.\n\nHowever! I’m still really interested in things like ML-powered camera quality improvements, object detection/background separation, and speech-to-text.",
"sig": "ea2d76bd08b7467a2d9fb959d1bce89b31b68e266853aff3af41f6ccb9d979f9ecfe66c21ad5e5f2272f3631e21c8feb7077e16100cfeba68e41a8956269cdb9"
}
