Technology

58303 readers

9 users here now

This is a most excellent place for technology news and articles.

Our Rules

Follow the lemmy.world rules.
Only tech related content.
Be excellent to each another!
Mod approved content bots can post up to 10 articles per day.
Threads asking for personal tech support may be deleted.
Politics threads may be removed.
No memes allowed as posts, OK to post as comments.
Only approved bots from the list below, to ask if your bot can be added please contact us.
Check for duplicates before posting, duplicates may be removed

Approved Bots

founded 1 year ago

MODERATORS

[email protected]

Illuminarty | AI generated Content Detection (lemmy.world)

submitted 1 year ago* (last edited 1 year ago) by [email protected] to c/[email protected]

9 comments fedilink hide all child comments

On illuminarty.ai i find this tool that detect AI generated images, synthetic, tampered images and Deepfake. There's a Free version and some subscription plan. After a couple of test it actually showed somewhat accurated results:

.> 70% Chance on actual AI drawing tested

<20% On (human) edited images

<5% On original handmade drawings

EDIT - Not so great with realistic AI Photo. Somehow works better with AI art.

UPDATE - I find this one who works better on photo | Is It AI

vvvv Share your % results and pics to see more test vvv

top 9 comments

sorted by: hot top controversial new old

[–] [email protected] 15 points 1 year ago (2 children)

This is not useful now, nor will something like this ever be useful.

[–] [email protected] 8 points 1 year ago

It's useful one one thing: adversarial training of generative AIs.

Except for photorealistic AI, which already beat this detector.

[–] [email protected] 1 points 1 year ago (1 children)

Good for finding edited images, no?

[–] [email protected] 5 points 1 year ago (2 children)

Since this seems specifically for ai, I would disagree. We humans are pretty good at detecting weird patterns in ai generated pictures, like to many fingers or weird proportions.

Also, at some point this detection is gonna have the same problems as current chatGpt detections...

[–] [email protected] 3 points 1 year ago

I detect you don’t know the word “too”

[–] [email protected] 2 points 1 year ago

like to many fingers or weird proportions

This is no longer true, the AI nowadays makes perfectly looking images.

[–] [email protected] 10 points 1 year ago

So, this picture of a lemming putting its hand in a bean soup in a beautiful mountain scenery has an probability of being AI of 0.1%. Seems legit.

[–] [email protected] 5 points 1 year ago (1 children)

I think this is just going to lead to a AI arms race with one team making AI to detect AI and the other team making AI to beat AI

[–] ChairmanMeow 10 points 1 year ago

It's literally how adversarial training works.