ChatGPT - AI-based chat

[Omega Man]

Some of what I do is related to "data veracity" - is the data what it says it is? I generally deal with business files, but data is data. For example, a person says they texted someone and they provide a screen shot of the data. That image proves nothing unless you have access to its metadata. On what device was this data created? Where was the device when the data was created? When was the data created? Does that align with the represented provenance of the file?


For image files like the ones that AI will generate, unless they're super sophisticated, casual assessment of the metadata associated with a file ought to reveal its origins. It's at the basis of the document management I'm involved with for litigation.

How do you do that? On a Mac, right click on a file and select "Get Info". On a PC, right click on a file and select "Properties". You'll see when the file was created, modified, etc. For photos, you will have EXIF information that includes things like exposure, device and the like. Some photos, particularly those shot on phones or other mobile, cellular devices, may bear location information, as well.

Metadata tells all.

If your lucky to see the “questionable” content on a device, all that is a great way to check validity of the content. Not sure how to check validity of a picture that may be “questionable” when viewed in/on media like TV or a manipulated picture printed in a magazine, newspaper or solicitation.

OM

 

[kbasa]

Possible to hack/change/spoof it?

It can be, but if there's a full examination going on on a device, system logs can show whether the file's metadata is true and correct.

Forensic examiners are like detectives. Most of the ones I've worked with came from three letter agencies, Law Enforcement or the military.

If you have access to the records in context, that is in their original location, it's not hard to spot faked files.

 

[kbasa]

If your lucky to see the “questionable” content on a device, all that is a great way to check validity of the content. Not sure how to check validity of a picture that may be “questionable” when viewed in/on media like TV or a manipulated picture printed in a magazine, newspaper or solicitation.

OM

Exactly. It's exactly how the Russians have attempted to influence us via social media. They create a fake identity that "looks American", then start stuffing fake articles and images on it. It's like a variation of a spear phish in some ways.

My advice is to treat all unsubstantiated data as suspect, whether it aligns with what we believe or not. It's the biggest reason I left Facebook 7 years ago after being on it since about 2006 or so when it was still little. It's a cesspool of unsubstantiated content that is driving us apart.


A thought, however, might be using Google image search to see where else that image might live. That may help see the "history" of a document or at least where it has appeared and when.

 

[rhbike]

Possible to hack/change/spoof it?

Yes. Easy peasy to do so. And when the file is copied to the internet, you lose the means to know if the tags/properties have been altered

 

[Visian]

Yes. Easy peasy to do so. And when the file is copied to the internet, you lose the means to know if the tags/properties have been altered

You can also remove all EXIF metadata from digitak photos before uploading, to be sure it can’t be used maliciously.

 

[Omega Man]

You can also remove all EXIF metadata from digitak photos before uploading, to be sure it can’t be used maliciously.

OR- as metadata can be turned off and on fairly easily, a picture without metadata……..can just be Photoshoped.
It’s unfortunate that it seems that people when shown different pictures or different pieces of information, gravitate to the information or pictures that “suits” rather what actually went on.
As this seems to relate to “Newspeak”, are we closer to 1984 and some sort of Orwellian, dystopian, future?





OM

 

[Visian]

As this seems to relate to “Newspeak”, are we closer to 1984 and some sort of Orwellian, dystopian, future?

Dude, can you not tell that we are living in that world now?

 

[Omega Man]

Dude, can you not tell that we are living in that world now?

That and an accelerated version of “The Decline of the Roman Empire”.

The “watchlist” includes- Soylent Green, Bladerunner and I Am Legend.

OM

 

[kbasa]

OR- as metadata can be turned off and on fairly easily, a picture without metadata……..can just be Photoshoped.
It’s unfortunate that it seems that people when shown different pictures or different pieces of information, gravitate to the information or pictures that “suits” rather what actually went on.
As this seems to relate to “Newspeak”, are we closer to 1984 and some sort of Orwellian, dystopian, future?





OM

Information bias is a thing. We all do it.

There’s a book titled Nonsense I read that drills into how and why we do it. As an example, the author had decks of cards made with the colors reversed. Red spades, black hearts. When he got the galley proofs he looked at them and thought it had been done incorrectly.

So he calls the printer, who tells him that it was indeed done to spec. So he took a second look, and sure enough it was right. This is a guy writing a book about how our brains will seek to make sense out of data by ignoring the stuff that doesn’t fit the “solution model” in our head.

His proposal was that as we grow up, we build models that we believe explain a situation and then unconsciously ignore things that don’t fit what we believe is happening. Like not seeing a heart plainly printed in black on a playing card. It’s a fascinating book and has helped me understand how and why we sometimes get to the conclusions we do. Sometimes they’re not based on facts or evidence because we literally can’t see what’s right in front of our face. We’re subconsciously ignoring the parts that don’t fit in our prebuilt model.

The book examines a Ducati race team trying to solve bike problems and some business scenarios where people try to hammer observations into this kind of preformed conclusion, ignoring signs and facts evident to others. It’s really an amazing look into how people get in trouble by rationalizing to “make the pieces fit” even when they don’t fit at all.

Nonsense. I forget the author. Blue cover with a big scribbled knot on the cover. On Amazon.

To tie it to AI hallucination, depending on how the AI is trained, AI models can have similar problems with properly weighting answers and ignoring relevant facts - or simply making them up. It all depends on the data used to train the LLM.

 

[kbasa]

That and an accelerated version of “The Decline of the Roman Empire”.

The “watchlist” includes- Soylent Green, Bladerunner and I Am Legend.

OM

Read Cormac McCarthy’s “The Road”.

Bladerunner 2049 is doggone great too, and the struggle toward the end of the movie with Harrison Ford and the cop is pretty symbolic of the human/AI relationship, imho.

 

[Omega Man]

Bladerunner 2049 is doggone great too, and the struggle toward the end of the movie with Harrison Ford and the cop is pretty symbolic of the human/AI relationship, imho.

I have seen it. I enjoy the accuracy of how the future will/could/did evolve……and mutate. Ridley Scott

OM

 

[kbasa]

I have seen it. I enjoy the accuracy of how the future will/could/did evolve……and mutate. Ridley Scott

OM

An astounding vision. I keep meaning to read "Do Robots Dream of Electric Sheep", the Philip K Dick novel the movie is based on. I used to live like a mile from where he lived, though he's long dead.

 

[Visian]

The age of Gemini

 

[Omega Man]

As it seems everyone has their own version of AI, what happened when they all start to integrate with each other- VICKI?

OM

 

[88bmwjeff]

An astounding vision. I keep meaning to read "Do Robots Dream of Electric Sheep", the Philip K Dick novel the movie is based on. I used to live like a mile from where he lived, though he's long dead.

I read that book in college and got me into reading his stuff. But, WHOA. Some of the stuff I read made me feel like I was going on the LSD trip as well. He clearly wrote while not sober.

FYI, you're close on the title. It's Do Androids Dream of Electric Sheep.

 

[kbasa]

I read that book in college and got me into reading his stuff. But, WHOA. Some of the stuff I read made me feel like I was going on the LSD trip as well. He clearly wrote while not sober.

FYI, you're close on the title. It's Do Androids Dream of Electric Sheep.

Thank you.

I got the same thing from just about any Carlos Castaneda book I've ever read.

 

[Visian]

Dang!

Channel 1

 

[Omega Man]

Channel 1

“All the news that fits we print”.

OM

 

[Omega Man]

Here we go-

Here we goes, “No Soup for You”……..

Eventually, the AI construct was able to correctly predict those who had died by 2020 about 78% of the time, researchers say in the report.

https://www.usatoday.com/story/tech...ted the life2vec,deaths of millions of people

Not sure if it is something that you sign up for OR, it’s just a data scrape off your Facebook postings.

OM

 

[Visian]

Produced with AI assist...

This is going to be a series on YouTube.

It's like Clockwork Orange meets A boy and His Dog meets Max Headroom meets 1984.