2024 Presidential Election
17h
1262
Uh oh: President Trump leads Biden 49% to 43% in a two-way race.
Ask Blinders
21h
950
Why Pronouns shit captured US ? I don’t see this anywhere else
AMA
Yesterday
3188
I have worked at TikTok US core tech for 3 years. AMA.
India
16h
3041
Why is it so G*damn difficult to move money out of India
Health & Wellness
14h
900
Issues with sleep
https://www.axios.com/2024/04/05/meta-broader-ai-labeling Honestly whenever i use IG I wonder if things are manipulated or created through AI -- I guess know we'll know. What do you think of this change? Is this the right direction for AI? I'm guessing most people will say yes but want to hear what you all think
Why does it matter . U visit ins to get entertained and you achieved your goal .
Why do people from china call IG ins?
They should label some posts also as Sponsored Escort.
Why would you need that. Thats nearly all of IG at some point if everyone has the label it doesn’t add new info
This change can be manipulative AF, what if something is real but meta just tag it as AI generated?
The opposite is much more likely, and probably more dangerous. People will infer that a lack of tag means it is real. This will only help to note known and prominent propaganda as such.
Agreed in general, but they can use ML classifiers to detect new AI content as well. It will be an arms to race: bypass vs. detect.
This will prevent your uncle from making fake news. This won’t prevent your uncle from falling for fake news. The AI generated content that needs to be blocked the most will easily pass these filters.
lol it’s adorable of you think a watermark exists that can’t be removed. I’ve read several journal papers that address AI watermarking, and they’re all complete jokes.
Knowing if something is AI or not, it's kinda hard to do man.
Don't care, as long as I like the content. Is human made content superior? Maybe
As long as it’s not porn
it's just an endless arms race. anything that is detectable, an llm (not even talking about ai yet) can learn to bypass. so, why even bother. i think the solution is much more along the lines of: eveeybody gets a personal ai helper agent that helps categorize incoming information by reliability
Ok in principle but eventually most content will be made with some AI so concerned of banner blindness as almost everything will have the label eventually