Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
...A lot of productive use is generating new information, not rehashing existing stuff. But the malicious cases are obvious from rehashing existing material by creating fake material...
That itself is a fascinating debate! AI fundamentally is curve-fitting. It takes existing data and knowledge. It follows instructions to mix and meld that data, connecting the dots as it were. Is this genuinely new knowledge, in the sense of a discovery? No. Is it new experimentation, or asking new questions? No. But it can detect new patterns in existing data; patterns that eluded even the most expert humans. That is also a form of discovery. So, is this "discovery" a generation of new information? Thus the debate....
Another example of AI disinformation - Met Gala: BBC Verify examines fake Katy Perry and Rhianna AI-generated images.
Perry wasn't at the red carpet event in New York but thousands of fans (4.5 million views and 115,000 likes) - and even Ms Perry's own mother - fell for the picture, believing it was real.
Rhianna wasn't there either due to flu.
The BBC’s disinformation and social media correspondent, Marianna Spring, looks at the fake picture in detail, explaining why this matters beyond the red carpet event.
This is the latest instance of AI fooling people on the internet into believing celebrities and political figures said or did something they didn’t.
Image, video and audio created by AI of prominent figures, from Pope Francis to Taylor Swift, have gained loads of traction online before.
And while this picture isn't perfect, yet - there is plenty of time for improvement.
Most people are gullible enough to believe what they see on the Internet, and they are eager to spread that propaganda and fake news all over the Internet - which we often see on Political Forums.
That itself is a fascinating debate! AI fundamentally is curve-fitting. It takes existing data and knowledge. It follows instructions to mix and meld that data, connecting the dots as it were. Is this genuinely new knowledge, in the sense of a discovery? No. Is it new experimentation, or asking new questions? No. But it can detect new patterns in existing data; patterns that eluded even the most expert humans. That is also a form of discovery. So, is this "discovery" a generation of new information? Thus the debate....
AI can be quite impressive at times. If you ask it to write a story in the style of a particular author like Hemingway or whoever, it will oblige. It is just that many of those examples involve a good deal of prompt engineering and trial and error. In a real-world use case you just ask it stuff and most people aren't trained to know the best way to frame a question, how to put in some guard rails, etc. So in order to avoid "hallucinations" I think the general-purpose bots are tuned conservatively and so tend to turn out more pablum-like derivative stuff.
There are also upper limits on how good it can get -- upper limits of compute resources and energy usage, which we are already bumping up against. Progress is being made in boiling down a huge training set to something that can run locally on a smart phone for example, for both privacy and resource reasons. But I suspect AI is going to stall for awhile except in a few areas where it excels (like, happily, medical diagnostics -- some truly exciting stuff is going on there). The more general the application, the more it will stall. It's one thing to correlate chest x-rays with what people were later diagnosed with, and relate that to some characteristic or artifact in the x-ray, so that they are doing a reasonable job of detecting incipient diabetes or congestive heart failure from those images. But ask the AI to play the role of priest and it's apt to tell you it's okay to baptize someone in Gatorade (yeah, that actually happened).
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
