Why is this here? Because the AI can't read pictures, yet
(media.greatawakening.win)
You're viewing a single comment thread. View all comments, or full comment thread.
Comments (23)
sorted by:
Not sure why you think AI cannot read pictures. Look in to OCR tech been around for a long time. We had software that could read text off of a picture about 20 years ago in consumer tech. It did a crap job then but is decent now.
Not only can AI read pictures, it was one of the VERY FIRST things AI could do. Developed in the 90s to scan and read checks.
It can read handwriting VERY well.
Especially plain text like that, some can even determine which font is being used. To block it from reading would need to obfuscate the text in some way that remains human readable.
What AI can't do reliably is read memes, because it might recognize the picture and the text, but there are layers of meaning that are beyond what an AI can grasp. That takes a human to classify the meme as blocked or not, but even then, by adding a partially transparent overlay, most of the time it will treat the image as entirely new and distinct.
Google Translate camera app does it really well, reading text.
For memes, AI is not far off from both understanding and producing.
Once that happens, you basically have mind control on demand.
I can't really say you're wrong because AI advances really quickly, and with the improvements in AI-generated art, I would lean that you're at least not far off.
The challenge with understanding is that while it might just be photo + text, take some of the LOTR memes, you can add text to almost any scene, the text might relate to the image, the scene the image is drawn from or drawn on the emotional sentiment of the scene, etc.
So, there would be plenty of opportunity for error in understanding because there are layers of understanding that might need to have an understanding of the film.
Producing memes would be much easier from that perspective.