▲ 795 ▼ We must resist (discuss.tchncs.de) submitted 2 years ago by chemicalwonka@discuss.tchncs.de to c/memes@lemmy.ml 74 comments fedilink hide all child comments
[+] cm0002@lemmy.world 52 points 2 years ago (9 children) [removed by mod] permalink fedilink source hideshow 9 child comments replies: [–] webghost0101@sopuli.xyz 36 points 2 years ago (6 children) Absolutely human, i even removed the context text and it didn’t even flinch. permalink fedilink source parent hideshow 6 child comments replies: [–] Bluetooth@feddit.dk 18 points 2 years ago (3 children) "Similar color" yep seems right, "...and texture" wait what? permalink fedilink source parent hideshow 3 child comments replies: [–] MissJinx@lemmy.world 10 points 2 years ago* idk what kind of muffin they are feeding this AI but sure is a hairy one permalink fedilink source parent [–] EvolvedTurtle@lemmy.world 8 points 2 years ago Cromch permalink fedilink source parent [–] Raz@lemm.ee 1 point 2 years ago You don't know until you try. permalink fedilink source parent [–] 31337@sh.itjust.works 7 points 2 years ago (1 child) Wow, that's a little too impressive. I'm guessing that image was probably in its training set (or each individual image). There are open training sets with adversarial images, and these images may have come from them. Every time I've tried to use ChatGPT with images it has kinda failed (on electronic schematics, plant classification, images of complex math equations, etc). I'm kind of surprised OpenAI doesn't just offload some tasks to purpose-built models (an OCR or a classification model like inaturalist's would've performed better in some of my tests). permalink fedilink source parent hideshow 1 child comment replies: [–] Mirodir@discuss.tchncs.de 4 points 2 years ago This exact image (without the caption-header of course) was on one of the slides for one of the machine-learning related courses at my college, so I assume it's definitely out there somewhere and also was likely part of the training sets used by OpenAI. Also, the image in those slides has a different watermark at the bottom left, so it's fair to assume it's made its rounds. Contradictory to this post, it was used as an example for a problem that machine learning can solve far better than any algorithms humans would come up with. permalink fedilink source parent [–] balderdash9@lemmy.zip 16 points 2 years ago Just take a bite out of each. You know you got the right one when its starts yelping. permalink fedilink source parent [–] photonic_sorcerer@lemmy.dbzer0.com 3 points 2 years ago Did people make you? If so, yes! permalink fedilink source parent
[–] webghost0101@sopuli.xyz 36 points 2 years ago (6 children) Absolutely human, i even removed the context text and it didn’t even flinch. permalink fedilink source parent hideshow 6 child comments replies: [–] Bluetooth@feddit.dk 18 points 2 years ago (3 children) "Similar color" yep seems right, "...and texture" wait what? permalink fedilink source parent hideshow 3 child comments replies: [–] MissJinx@lemmy.world 10 points 2 years ago* idk what kind of muffin they are feeding this AI but sure is a hairy one permalink fedilink source parent [–] EvolvedTurtle@lemmy.world 8 points 2 years ago Cromch permalink fedilink source parent [–] Raz@lemm.ee 1 point 2 years ago You don't know until you try. permalink fedilink source parent [–] 31337@sh.itjust.works 7 points 2 years ago (1 child) Wow, that's a little too impressive. I'm guessing that image was probably in its training set (or each individual image). There are open training sets with adversarial images, and these images may have come from them. Every time I've tried to use ChatGPT with images it has kinda failed (on electronic schematics, plant classification, images of complex math equations, etc). I'm kind of surprised OpenAI doesn't just offload some tasks to purpose-built models (an OCR or a classification model like inaturalist's would've performed better in some of my tests). permalink fedilink source parent hideshow 1 child comment replies: [–] Mirodir@discuss.tchncs.de 4 points 2 years ago This exact image (without the caption-header of course) was on one of the slides for one of the machine-learning related courses at my college, so I assume it's definitely out there somewhere and also was likely part of the training sets used by OpenAI. Also, the image in those slides has a different watermark at the bottom left, so it's fair to assume it's made its rounds. Contradictory to this post, it was used as an example for a problem that machine learning can solve far better than any algorithms humans would come up with. permalink fedilink source parent
[–] Bluetooth@feddit.dk 18 points 2 years ago (3 children) "Similar color" yep seems right, "...and texture" wait what? permalink fedilink source parent hideshow 3 child comments replies: [–] MissJinx@lemmy.world 10 points 2 years ago* idk what kind of muffin they are feeding this AI but sure is a hairy one permalink fedilink source parent [–] EvolvedTurtle@lemmy.world 8 points 2 years ago Cromch permalink fedilink source parent [–] Raz@lemm.ee 1 point 2 years ago You don't know until you try. permalink fedilink source parent
[–] MissJinx@lemmy.world 10 points 2 years ago* idk what kind of muffin they are feeding this AI but sure is a hairy one permalink fedilink source parent
[–] 31337@sh.itjust.works 7 points 2 years ago (1 child) Wow, that's a little too impressive. I'm guessing that image was probably in its training set (or each individual image). There are open training sets with adversarial images, and these images may have come from them. Every time I've tried to use ChatGPT with images it has kinda failed (on electronic schematics, plant classification, images of complex math equations, etc). I'm kind of surprised OpenAI doesn't just offload some tasks to purpose-built models (an OCR or a classification model like inaturalist's would've performed better in some of my tests). permalink fedilink source parent hideshow 1 child comment replies: [–] Mirodir@discuss.tchncs.de 4 points 2 years ago This exact image (without the caption-header of course) was on one of the slides for one of the machine-learning related courses at my college, so I assume it's definitely out there somewhere and also was likely part of the training sets used by OpenAI. Also, the image in those slides has a different watermark at the bottom left, so it's fair to assume it's made its rounds. Contradictory to this post, it was used as an example for a problem that machine learning can solve far better than any algorithms humans would come up with. permalink fedilink source parent
[–] Mirodir@discuss.tchncs.de 4 points 2 years ago This exact image (without the caption-header of course) was on one of the slides for one of the machine-learning related courses at my college, so I assume it's definitely out there somewhere and also was likely part of the training sets used by OpenAI. Also, the image in those slides has a different watermark at the bottom left, so it's fair to assume it's made its rounds. Contradictory to this post, it was used as an example for a problem that machine learning can solve far better than any algorithms humans would come up with. permalink fedilink source parent
[–] balderdash9@lemmy.zip 16 points 2 years ago Just take a bite out of each. You know you got the right one when its starts yelping. permalink fedilink source parent
[–] photonic_sorcerer@lemmy.dbzer0.com 3 points 2 years ago Did people make you? If so, yes! permalink fedilink source parent
