Unless there is a very good reason (happens occasionally but usually not,) I, as a totally blind guy, won't boost a post with video or pictures that don't have description.
I make it a point to describe any audio I post here.
Sometimes, there are pictures described by LLM's. One could argue that is better than no description at all, but as has been pointed out in a few places, sometimes, that description is either totally wrong, completely misses the context in such a way that the meaning changes, or some other combination thereof.
Yes, I'd rather have a human-described image, even in another language, because translation tools are built-in. It's pretty easy to tell when an image description has been generated by an LLM. They have a very defined style about them. I always take those with many grains of salt, because I have no way to verify, without asking someone, whether or not what the LLM comes up with has any relationship to reality.
On the surface, most of these descriptions look pretty reasonable, but these things have built-in biases that quite often show themselves.
I'm still not sure what "the right answer" is here, other than, whenever you can, please describe images yourself, or crowd-source it if you can't, for whatever reason.
If you're stuck for time, you can always go back and add descriptions later.
This all having been said, I really appreciate the fact that so many people do take the time to describe things here. IT's quite a different vibe from any other social media I've ever been a part of.
Borris
npub1k8…lf37f
2025-04-15 00:29:09 UTC
Author Public Key
npub1k87zy57zh2hakx76h09v9jdga0a4w6jdlqmvxdxypyx5lu0tvwwqjlf37fSeen on
wss://relay.ditto.pubPublished at
2025-04-15 00:29:09 UTCEvent JSON
{
"id": "1c7fb453b1289b6deb73c805abdc3bc3f37c011ca6e149d437b01597bc19cd88",
"pubkey": "b1fc2253c2baafdb1bdabbcac2c9a8ebfb576a4df836c334c4090d4ff1eb639c",
"created_at": 1744676949,
"kind": 1,
"tags": [
[
"proxy",
"https://fwoof.space/users/BorrisInABox/statuses/114339148575095669",
"activitypub"
],
[
"client",
"Mostr",
"31990:6be38f8c63df7dbf84db7ec4a6e6fbbd8d19dca3b980efad18585c46f04b26f9:mostr",
"wss://relay.mostr.pub"
]
],
"content": "Unless there is a very good reason (happens occasionally but usually not,) I, as a totally blind guy, won't boost a post with video or pictures that don't have description.\nI make it a point to describe any audio I post here.\n\nSometimes, there are pictures described by LLM's. One could argue that is better than no description at all, but as has been pointed out in a few places, sometimes, that description is either totally wrong, completely misses the context in such a way that the meaning changes, or some other combination thereof.\n\nYes, I'd rather have a human-described image, even in another language, because translation tools are built-in. It's pretty easy to tell when an image description has been generated by an LLM. They have a very defined style about them. I always take those with many grains of salt, because I have no way to verify, without asking someone, whether or not what the LLM comes up with has any relationship to reality.\n\nOn the surface, most of these descriptions look pretty reasonable, but these things have built-in biases that quite often show themselves.\n\nI'm still not sure what \"the right answer\" is here, other than, whenever you can, please describe images yourself, or crowd-source it if you can't, for whatever reason.\nIf you're stuck for time, you can always go back and add descriptions later.\n\nThis all having been said, I really appreciate the fact that so many people do take the time to describe things here. IT's quite a different vibe from any other social media I've ever been a part of.",
"sig": "29f63f509ea907e7ee371d36d20c6348fb5d383c22a52be86513e03604765a9301085117b6ca32ae607fd22a8660c3c39187a964382ca821c3983cc7aecf2ce4"
}