Anyone else following this?
Attachment 97196
Attachment 97197
https://twitter.com/i/status/1822650475399070019https://twitter.com/i/status/1822650475399070019
Printable View
Do with it as you will...
https://www.reuters.com/fact-check/f...nt-2024-08-12/
Trumps needs a filter on occasion.
Forget a teleprompter, wait until AI tells the candidate what to say in a debate. Live access to an AI system would make you a formidable competitor. Whoever masters this technology will take over the world. The Star Trek "holodeck" could literally be come reality.
https://www.wired.com/story/chatgpt-...t-impressions/
Probably the two biggest risks of AI today is simply that people will discount any image or video they innately disagree with, or alternatively, this- misattributing generated images.
Unfortunately this is relatively tame compared to the political wrangling-bullshit we'll see in four years. It's just a gigantic tool in the box building idiocrasy.
That's weird!
“Weird”….Buzzword, Check.
You can find a few anomalies in the first pic, some of which can easily be explained due to the spontaneity of the shutter click, Other than trying to figure out the phantom hand issues with the white shirt/blue hat guy.Quote:
Originally Posted by Ah Pook
But what's everyone's beef with the circled points of the lower photo? Nothing jumps out at me. No extra or missing limbs, different screen can be attributed to myriad reasons.
Certainly wouldn't put it past the D to fake it, but some of the evidence seems a bit flimsy. Maybe I just can't zoom in enough.
I'd be more concerned that one day AI decides you are a threat to national security and has you terminated with extreme prejudice. You know for "the greater good..."
Computer generated or doctored imaged are not true artificial inteligence.
It drives me nuts when any and everything computer related these days is labeled AI.
https://adsei.org/2023/03/12/theres-...20you%20squint.
A system that most of us would think of as real AI – something that can, more or less, think like us – is known in Computer Science as Generalised Artificial Intelligence, and it is nowhere on the horizon. The term Artificial Intelligence is used instead to apply to anything produced using techniques designed in the quest for real AI. It’s not intelligent. It just does some stuff that AI researchers came up with, and that might look a bit smart. In dim light. From the right angle. If you squint.
True AGI coupled with quantum computers? That is the Terminator, end of the world scenario to fear.
Until then, it's just photoshop.
I think the term is "generative AI" in that it "generates" conclusions based on data it gathers. For instance say if the Secret Service had a shot spotter sensor when Trump was shot and it was tied to an AI network. The AI network informs the counter-sniper of the source of the shot within milliseconds.
I just can't wait for AI medicine. The doctors I am dealing with are clueless.
AI medicine, that’s the future! Not sure how it would work but, I’m in.
I'm ready for the nanobots.
Drink a milkshake and a billion tiny robots swarm through your system, fixing everything that ails you, at the cellular level.
Wake up with a fresh body.
This company develops software to detect AI generated content. I have no idea how it works. I do understand that the Pentagon is considering some math discoveries to be classified by law. Especially in advanced physics relating to nuclear weapons systems. Space based lasers developed and managed by AI for example. What would really be exciting is faster than light propulsion. 10 minute flights to Mars would be a game changer.
https://bellrock.ai/
If anyone had Max Headroom as Press Secretary, I'd vote that ticket.
Important distinction: Photoshop requires human processing to create each of those people individually (for the most part) and with purpose. AI doesn't. A good example of Photoshop would be the photo with the Where's Waldo background above. This would take a 7-year-old about 15 seconds to do with free apps/filters on their 4-year-old smart phone these days. You're outlining the subject and telling it to make that the top layer and this other photograph (that already exists) the bottom layer. Child's play.Quote:
Originally Posted by eddiememphis
What takes a human hours and hours (days/weeks?) in the case of the photos in question would take AI about 15 seconds. You would literally open a LLM prompt, drop the base image in and say something like "Recreate this image to show lots of supporters at this rally" and within a matter of seconds you'd have a picture with lots of supporters. You could then say "even more supporters and make sure they have signs that say Harris Walz" and bang! seconds later, it's jam packed with randomized and individualistic (usually) human looking people in a photo. This is the problem with AI and why so many are concerned with it taking their jobs (which it is currently doing and has been for years in some segments, this is why Hollywood writers specifically have been so active the last few years). Especially those in the creative/design space which is already using AI to do the same work that would have taken paid humans days or weeks to perform, depending on the scale. It's been going on for many years now in various segments and most consumers have no idea, because they all still use humans to verify the output before taking live.
I worked for a digital marking firm starting around the time of the pandemic for a few years and saw it first hand within a small-ish local company. The writing team shrunk (forceably) to nearly zero and other departments were being instructed to use the "company's" AI (which was literally just a reskin of an older version of ChatGPT) to write copy/SEO for websites, social media engagements and ads. It basically allowed other departments to simply ask for it, give the output a quick once-over to either sentences or pages and pages of output and then simply copy/paste it into the target. Some couldn't even get that last step right to give you an idea the skill level drop-off in the chain. Initially, we'd have to run the output through an additional step to make sure it wouldn't receive Google/Bing/[insert any other search engine algo here] copywrite strikes for plagiarism which would negatively impact SEO, which it did often early on. The LLM was tweaked over time and, before long, this step wasn't even necessary. Just as an example of relatively small resource buckets and the amount of growth that was experienced in it's effectiveness in a very short period of time. That small company saved hundreds of thousands per year by shrinking it's content writing budget to a skeleton of what it formerly was. It did this by embracing the technology it's dev team "created" to perform the same tasks that humans otherwise would have. They were doing the same on the design side with branding & graphics as well although their LLM version, at the time, wasn't as good with graphical creation but would still get the job done. It would just take more attempts & human interaction to review the output.
That part gets better with time and where AI is improving daily. This was nearly 5 years ago for context. The use of AI to replace humans isn't "coming" but is already here in many capacities/roles and most people don't even realize it. They think that it's still "coming" and laugh at the stupid stuff it does that is spotlighted when it tries to tackle more advanced requests that are outside of the scope of it's (current) capabilities. Meanwhile, the stuff it already does well just passes w/o a second look because it's impossible to distinguish from what a human did. Hell, the people "helping" the process don't even realize it until they're out of a job.
Now, the downsides to using AI for tasks such as this are widely known. It doesn't always get some concepts right, things we know to be foundational like; humans have only two arms. Some have only one or (even rarer) none but never three, four or six. Arms can't go through heads and come out the other side holding a phone and recording an event. Sometimes the AI will get lazy and create the same faces in a sea of people without really double checking that it's done this. If a human catches it, they can follow up with a command like "make sure every human is different and none are copied" and then double check the work once again to try to minimize this. This is where humans are needed still is to check the AI output. Whenever you have a human involved though you have a time component and a possibility for error.
Also, AI is notoriously awful at making human hands. Like, it sucks royally. Stretched awkward fingers, four, six or seven of them... or many times all of the above. It struggles mightily with this seemingly simple tasks. You see things like blurred facial structure and "googly eyes" a lot more regularly as it doesn't understand that this trait isn't super common. It also struggles with the concept of reflections regularly. Like those that would be present in a large, flat, shiny surface similar to that of a large airplane engine case where humans would understand that if you can see a building reflection you'd also see the mob of people between that reflection and the building. Basic, fundamental concepts that humans grasp at an early age that AI hasn't (yet) mastered.
Armed with this info, look at those photos again.
To me, everything looks out of place. A rally on the tarmack? That and she isn't doing any appearances.
It's going to be an interesting few months.
The big company in AI is Palintir Techologies. They are huge with a market cap of almost $70 billion.
https://www.palantir.com/
Attachment 97211Attachment 97212
If the report/article says images instead of photo or photograph it is either written by someone who does not understand the importance/significance or someone who does and is signing their work.
You cited evidence of very advanced programs but not true intelligence.Quote:
Originally Posted by Jer
If the computer or program or whatever were truly intelligent, it would learn from it's mistakes and create realistic hands.
The fact that it still needs human reprogramming proves it is not truly intelligent... yet.
Calling everything AI diminishes the meaning and importance of the term.
Once there is actual artificial intelligence that exists outside of a lab, everyone will be VERY aware of it since it will change everything.
ChatGPT conjuring an image of Jesus on a flying carpet will seem quaint by comparison.
Like it or not, this is the definition of artificial intelligence. It's using artificial computation to mimic what a human would otherwise have to do manually and intentionally. You can yell at the clouds all day about how good it is or isn't but that doesn't change the definition.Quote:
Originally Posted by eddiememphis
Then playing against the computer in PONG is AI.
Got it.
No. Pong doesn't learn. Pong also didn't have an agreement requirement before you could play authorizing your data to be sent back for machine learning purposes the way ChatGPT does. No agreement, no access.Quote:
Originally Posted by eddiememphis