Post AsFb3PWXnPpXSIW9AG by feliks@chaos.social
(DIR) More posts by feliks@chaos.social
(DIR) Post #AsFSvXDwU6gnO1YAyG by feliks@chaos.social
2025-03-20T14:21:01Z
0 likes, 0 repeats
(DIR) Post #AsFWKhSTrzM8EL2eG0 by feliks@chaos.social
2025-03-20T14:59:08Z
0 likes, 0 repeats
@ticho well... you can derive some limited value from it. i guess it's worth a look. just watch out for overly enthusiastic people in that space. helps managing expectations ;)
(DIR) Post #AsFWkDPKDkpQYfjSyW by Zeugs@social.cologne
2025-03-20T15:03:36Z
0 likes, 0 repeats
@feliks you have to Apply Agents to it! so just the same model with different prompts reading each other's output... The results are the same but the costs are much higher. So that one of the evil tech companies will get more money for you wasting your lifetime.... But maybe it will work and you can showoff to other AI Brows(but no one else will give a shit)
(DIR) Post #AsFYpGOEl90pAFl4b2 by feliks@chaos.social
2025-03-20T15:27:06Z
0 likes, 0 repeats
@ticho haha okay. i guess i have found some utility in them already. i think of them more as glue or utility scripts. and some things are much more tedious without them. image recognition works very well across modalities (ocr, image to json that fits a schema, etc)
(DIR) Post #AsFb3PWXnPpXSIW9AG by feliks@chaos.social
2025-03-20T15:52:03Z
0 likes, 0 repeats
@ticho that is true but can be linked, e.g. conversion of an image to a json object. discriminative and generative models sometimes work togetherofc this technology can be used in bad ways like any other. spam is one of them. while i believe this is easy to get lost in i try to focus on beneficial applications while still being aware of these misusesif these toys bring people personal pleasure i don't see them doing that much harm individually. at scale that's another thing though