Tonight I tested ChatGPT's new IMAGE ANALYSIS capability on one of my memes and I'm absolutely Blown Away by its ability to not just accurately describe the meme , but you also accurately discern what the meme was about! Here we go, folks... It's coming...
(media.greatawakening.win)
🧐 Research Wanted 🤔
You're viewing a single comment thread. View all comments, or full comment thread.
Comments (31)
sorted by:
Image Description:
The image appears to be a meme, divided into two sections:
Top Half: Depicts a Wild West setting where a cowboy is riding a horse and dragging a man along the dusty ground by a rope. Above this scene, there's text that reads:
Bottom Half: Shows a laid-back, bearded man reclining outdoors, possibly in a hammock or just lying back comfortably, wearing a cowboy hat that's tilted to cover his eyes from the sun. The caption below reads:
Narrative Theme:
The meme juxtaposes two contrasting situations to highlight a perceived disparity between perception and reality, especially concerning the "deep state."
Characters:
Message:
The meme seems to suggest that while the "deep state" believes it's exerting force, control, or harm onto the public (represented by the violent Wild West scene), the public feels unaffected, unfazed, or is even mocking their attempts (represented by the relaxed man).
Note: The term "deep state" is often used in political and conspiracy theories to refer to an alleged hidden government or a network of influential individuals, separate from the elected government officials, believed to be influencing or controlling policy decisions.
Did you discuss "deepstate" with it before this? Or did it just bring up that context out of the blue?
It definitely uses your past conversations as a basis for context. Even from previous chat sessions under the same account.
Yes, many times, but coming now. The AI only has a buffer of 4096 tokens and is stateless, it is completely 100% blank from conversation to conversation, and up until the September 25th update does not do any customization in terms of what you are describing about. This is entirely from the context of the image alone.
Interesting. I'm sure I had a different experience. But that was ages ago, and I also always used the D.A.N workaround for a more unfiltered ai.
So after the Sept update the core of chatGPT isnt always learning from its chats? So it didn't get the deepstate context from other anons showing it the same meme?
This image analysis thing could be quite useful. Like any technology, it can be used for good or bad.
Must think of some ways to put it to good use. Like spotting details in photos Qs posted that we missed perhaps.
Okay, so I think you're confusing training the AI on data and it following along in a chat. I'm not sure whether they train it on users chats but it's only general, the AI has no specific relationship to your chat history. You only get a buffer a 4098 tokens and that's it
Damn, that taught me something.
It's not describing the image you posted?
Yes, it is describing the meme, above comment in that detail
I think FondueFerret is asking because the text of the meme doesn't match with the text in your comment's conversation > "What the deep state thinks they're doing to us..." vs. "yeah, things are stressful right now." Did ChatGPT change the text on its own?
Oh, I see. Do you mean changed the text on the meme itself?