119: Is That Photo Real?

Welcome to Pragmatic. Pragmatic is a show about technology and contemplating the finer details in their practical application. By exploring the real world trade-offs, we dive into how great ideas can be transformed into products and Pragmatic is entirely supported by you, our listeners, and if you do. Premium supporters have access to early release, high quality versions of episodes, as well as bonus material.

From all of our shows are not available anywhere else. Just visit engineer.network slash pragmatic to learn how you can. And today I'm joined by the

How are you doing, Clay? I'm doing splendidly well. Excellent. How are you? Uh yeah, I'm I'm I'm don't wanna steal that, but I'm doing splendidly well as well. So yeah. It's um And we're not having Splendor with that splendorly roll. Uh no, no, we're cutting out artificial sweetness'cause they've I've realized they give me a headache. So um but yes, never mind. Yeah, I know, right? Funny little factoids we talked about before the show, but there you go.

Um so it's been it's been a minute since we caught up. Uh Really? Uh the last time I think we spoke uh on a podcast was actually episode fifty five of Bubble Sort. Yeah. And that was on the seventh of September twenty twenty four. Which oddly, the edit for that only went up last week for It's like uh it's like a time capsule that uh floated up. Yeah. It's a time capsule that everyone forgot about. It's like oh wow. What happened to bubble sort? Anyway. Um so anyway, so that was

Wow. Actually seven seven I haven't done the math in my head, but that's like a a ye one and a half years. So it's been a while. And the last time we spoke on the show was even before that on this show. Yes. So Uh one of the reasons that I um I wanted to talk to you again, apart from the fact that it's always great catching up. Indeed. Uh, is I wanted to t yeah, I also wanna I want to talk about the whole thing about fake photos and AI impacting photos and what it means for us as photographers.

And um although actually you know what, before I get stuck into that as the as the main topic uh to talk about,'cause I know we both have uh some interesting uh some opinions on that, um what we haven't I haven't talked about photography much on Pragmatic recently.

Uh but I have actually got some new camera gear on this show. I did briefly talk about on bubble saw, but I think it's worth covering it here a little bit more because I've had the uh the benefit now of actually using it uh quite a bit actually uh between then and now. So as you know, I'm in the Nikon or Nikon, whichever you want to pronounce however you want to pronounce it.

Uh I'm in that ecosystem. Everything that I've got now I eventually I've I got rid of all my old Uh so I used to have a D five hundred, I sold that and I've gradually gotten sold and replaced all of my F mount lenses with Z mount. So yeah. So I've got my I still have my Nikon uh Z six two and uh all the other lenses I I've spoken about previously, but before I went on Uh a holiday. I I first of all I I went to Arizona and I did a talk at Arizona, um, which I have talked about in this show.

uh and I took this camera with me Uh and took some fantastic photos. I well I like I think they're fantastic. I should probably show you, but I don't have them in front of me. I'll send them to you after the show, but anyway.

Uh, so I got some of the Verde Valley uh Railway, which was a really cool uh train ride that was part of the um when you go to the conference they had this day in the middle of the conference'cause it was like a five day conference. So then and and on day three We we basically had a few hours in the morning there were a couple of talks and then we just went and did this excursion for the rest of the day as a bit of sort of like a mid Uh mid conference break.

Which I thought was a kind of a nice touch. Yeah, it was pretty cool. I actually really liked it. So it got some nice photos of that. And I like the train ride'cause I got lots of photos actually. So there you go. Um Never mind. You got tra wait, you got photos uh while you were on the train or or off the train? Yeah. Okay. Yeah, yeah, both. Yeah, yeah, both.

So when we when we were on the train, some of the scenery was just uh magnificent. So it's a beautiful part of the world. Uh beautiful part of Arizona. So anyway, uh so I got this uh Nikon uh Z thirty and I got the Z thirty because it was

It's essentially a pocketable camera, pretty much. Uh and it came in a kit with a DX lens, a sixteen to fifty uh VR uh DX lens, which is It's very much like a pancake lens, so it's not the skinniest lens because it a as you rotate it, uh, it will actually protrude. So it will extend. Oh, okay. It does. It doesn't extend a lot, but when it's in when it's actually locked and closed, it's relatively compact. So I bought a super slim line, skinny, small camera case for it.

Uh but if you don't mind a ridiculous bulge in your pocket, you can actually shove it in your hip pocket of your jeans. Do not recommend, but you probably could. So that's that was an interesting lens. It came with it. I still have it. I have used it. It's handy in some situations. Uh but the lens I really wanted was the super telephoto. Like it starts out at it's the Nook on Z twenty eight millimeter to four hundred millimeter. Nice. So that

With with'cause um the Z thirty is of course a crop sensor. Mm-hmm. Uh so when you go one point five X therefore on that you'll get an effective reach of six hundred miles. Which is awesome. Yeah. That was impressive. So uh the only thing is on the four hundred mil end y it's an F eight aperture. F eight, yeah. Yeah, which is not

that great. It's VR, both the camera uh the the body is not VR, but the lens is VR? Uh yeah, correct. Yes. So yeah. So the so you're absolutely right. The Z thirty does not have in body stabilization.

Uh, which is a shame. But the Z six two does. So when you put the twenty eight four hundred on the Z six two, um, it's super stable and it's fantastic. Uh and it takes great moonshots, which I think I've sent you one in the past. But Where that where the combination of the Z thirty and the twenty eight four hundred really shines is uh when you are on holidays, you can basically take photos of everything from close up portraits all the way out to like distance shots.

and like when I'm trying to bring in scenery. So what I did is I took that to Arizona and I was able to get some fantastic shots. uh of of like there's uh there were all there were caverns in the side of them of the mountain range uh and uh the valley and uh they'd been dug out by like Native Americans as as shelters and so on against the heat over the, you know, centuries and whatever else. And so like

uh you couldn't get close to them'cause they're a fair way away, but I had the reach. So I was able to get some some better photos of those than most of the other people. Mm-hmm. Right, so one of the things that I One of the things I also took this uh lens away on is uh another I did it we did a cruise uh in about a oh three or four months after that. um to New Zealand and I took it s with me to New Zealand and I also took a whole bunch of really great photos I couldn't have otherwise got.

So it's been it's been I've been really happy with it uh as a lens. The only caveats that you would expect is you do start to get um a bit of You do get a little bit of vignetting and there's a little bit of pincushion distortion even with the lens corrections that are done, uh on the on the long end. It's still on the long end, okay. Yeah. Um but you know what, it's for something that that covers twenty eight mil to four hundred mil, F four to F

eight um across that range. I mean, that's still pretty impressive. And I'm not gonna complain too much. It's not like I'm buying a prime lens that that can give me six hundred mils of reach on a full frame. That's gonna cost me ten thousand dollars probably. So I mean for something that didn't cost uh f cost a fraction of that that can cover that range, it's pretty impressive. So I'm pretty happy with that actually. Yeah. Yeah.

Um, you know, I I had people who who got into photography with me uh years ago who aren't in it anymore'cause, you know, uh they're they're always trying to strive for something extremely perfect. And it's like you know, think back about the capturing the moment. Yeah. Regardless. Right? I mean a v a vignette is I'll I'll take that if I

if I got the moment that I was uh you know, striving for. Absolutely. Yeah. Uh absolutely right. And uh it I think taking Taking photos of like scene of scenery or of so landscapes or if your subject

is a landscape or you and not a person and it's not about capturing the emotion exactly in a scene. Unless you're doing something a golden hour, I guess, but um, you know, for me, like the tr the emotion for me is like human emotion when I'm taking photos of groups of people, um, those are the ones where

Um I'm not I I absolutely agree with you. So I'm not trying to capture every pixel perfect representation of what the scene was like'cause I'm only interested in capturing the emotion of that person and And I find the best photos that I take are the ones when people don't realize I'm that I'm taking the photo. Now I don't mean that in a creepy way, but I mean that in a way that that's not staged.

You know? Right. That's those are the those are my favorites. Great. But in any case, um all right. So I also, however, did get a panorama adapter uh and a tripod leveler, which is I'll put firehuis in there. uh in the edit so people can see what I'm talking about. You probably know what I'm talking about anyway. Um so it's like the

So the tripod level is just like uh two little metal triangles and they've got little turning wheels on each of the points and you can use that with the bubbles to just make sure it's perfectly level. Mm-hmm. So you turn one left and right or whatever else and it just levels the two plates so that the

the camera's perfectly level. Uh and then the other one's a panorama adapter, which is a three axis uh panorama adapter. So you it's got an Arca Swiss adapter for uh attaching and detaching So you can just put that onto the bottom of your uh mirrorless. And you can then Uh there's like a a pin and you can move the pin around so you can do uh settings for every fifteen degrees, every thirty degrees, every forty-five degrees, and it literally has a ball bearing on a spring inside it. Mm-hmm.

uh spacing between them and if you've leveled it properly then everything will be a nice perfect series of photos for a for a panorama. And I know that my iPhone can do it. But the quality of the photos are better on my Z six two. Thank you very much. So anyway. Absolutely. Hm. Do you your that lens does it have a a tripod mount as well or or No. Okay. No. Uh the only lens that I had, and this is past tense'cause I sold it. Uh I had a two hundred to five hundred uh F

Uh five point six zero Yeah, that was their five point six constant aperture and that was a beautiful lens. I did a took a lot of beautiful photos of that, but it was an F mount. and I had an F T Z adapter on it and I sold those uh in part to fund the twenty eight to four hundred. And I was actually a little bit sad when I sold that. I don't often get upset necessarily when I sell

Like the odd lens here and there, but that particular lens I'd taken a lot of really nice photos with it. It's a beautiful lens. Mm-hmm. And uh yeah, I was a little bit sad when I sold that one'cause yeah. One of these days I might get a nice

like really good prime, you know, four hundred, five hundred mil lands maybe, but yeah. I gotta go trip over a bar of gold first, I think. But the thing is that the thing is that uh time as time goes on d that that lens might actually pop up on the US market for, you know, you might... Every now and then I pop up and I I see things in not not in the store, but like if you go to Facebook marketplace, right, you you might see someone who

is selling it for half what it's worth and you know, every now and then you think, Well, this is too good to be true and actually mm you talk to them and you find out actually this isn't stolen and this is not too good to be true. They just They just don't they they just don't want to sell it for as much as everyone else's and you know, it's a bargain. So Yeah. It's uh it's not stolen, it hasn't been dropped, and it's not full of mold and fungus.

So yay. Winning. Yeah, actually I saw a Z mount. Uh it was uh these knock on Z I think it was a six hundred mil, um one. Uh trying to remember the exact lens it was, but yeah it was selling via Ted's cameras here in in Australia anyway. They had a second hand one with a warranty and everything for seven thousand dollars and those are twelve thousand dollars new Australian dollars so mm um don't make me convert that into US dollars but still it's still a lot of money.

But anyhow. But anyhow. Yeah. So that's my gear. Um so since we last caught up on the show, what else uh what have what have you been playing with in photography land? Uh I'm I'm yeah, I'm not sure if I've had the the same uh A7R four. I I've I have other cameras that I picked up in between, but my main camera is my A7

R four. Um I do have a pocket uh Rico G R uh three that I take with me every now and then. But Main camera, if I if I want to take you know pictures that I can have a lot of control over, it's it's gonna be the the A seven R four. um two point eight or the fifty. The fifty one point eight is like it's not as good a lens. It is It is it it say if it's gonna be darker um and I don't want to bump my ISO too high, I I probably will opt for that one. But if I you know if I want to

a a a better quality lens. I the three point f the thirty five two point eight is a better lens. It's interesting that the thirty five two point eight, um so So you wouldn't go for uh do do you have a thirty five one uh one point eight or? Uh no, I I actually I've been thinking about buying one. The thing is that I don't shoot thirty five as much, especially because I have um I have the Um, I have the Rico G R which is like a twenty eight and to me it's like thirty five, twenty eight, uh they're

Similar. Yeah. So if I if I sometimes I what I do is I take the GR um as my thirty five, even though it's not a thirty five, and then I take the Sony as my fifty. Yeah, okay.'Cause yeah, I I remember us talking in the past, you're a big fan of the fifty for street photography. Yeah, yeah. Yeah. Twenty five is really nice though. Like you d but sometimes um if you're out there with a twenty five

And let's say you're at the beach. Um you do you really do have to if you wanna get the the picture of that person just, you know, doing their thing, you do have to get a little bit closer than you sh than I than I like to. Yeah, that's that's that's a good point actually. So um yeah,'cause uh I've got to admit I d I do have a fifty millimeter Z mount lens, but it's actually a macro lens.

And um it's it's great at is the one to one? The yeah. I think so. I'm trying to remember. I'm trying to remember, I think so. It's it's the M C fifty. Okay. So uh and I've got the uh slide mount adapter kit. So I've been going through some of my father's old uh slides that we've come across um as we've been um as we've been tidying up and stuff and yeah so I got the the slide mount adapter and everything for that. And uh yeah, it's it it's a great lens, but macro lenses Uh I don't know.

I'm pretty sure it's that two point eight. Damn, I'm putting myself on the spot here. I want to make sure I'm not telling you I believe I believe it is a uh I know I I know there used to be a a fifty

Yeah, and that's three point five if I'm not mistaken, but uh this one is sort of F two point eight, I believe, right? Yep, yep, is it is F two point eight. I was my first answer was the correct answer, so Uh and and it's it's a perfectly fine lens for macro photography and I've done a whole bunch of macro work but

I don't know. Maybe I shouldn't feel weird using a macro lens when I'm doing street photography, which I don't do often. Well the one thing that I I so I had I don't I have a macro ninety And I love taking that one out. The only issue is that like if I do want to take a a a a sort of more intimate portrait The person's really far away. I like to I like to I like the ability and of it being a macro just because I could turn

a street session into something like, Okay, snap, this is a beautiful coin that fell on the floor, right? And I have a macro right right then and there. That's right. So that that's for that reason I wouldn't mind getting a fifty macro actually. Okay. Interesting. All right. I hadn't thought of it like that. That's uh that's a good point. I I did actually try some street photography, uh we talked about this.

uh a while back now. Geez quite a while back actually and uh different techniques for you know doing uh street photography without looking like you're some kind of I don't know what person. But anyway. Um you know. And um anyway, so I did try a little bit of that and it was a little bit of fun, but I also found a little bit nerve wracking like everyone was watching me and I felt a bit weird but It is nerve wracking.

I mean you you you were talking about how you um you know how you conquered that beast and uh You know, it it's almost the same uh same pathway, right? Um Yeah, it's true. Yeah. Yeah, I guess the more if I did it more I'd I'd get more used to it and less less uh I'd be put off less by it and yeah, I should probably do more of that. But anyway. Hm. Cool.

We should probably move on to our main topic actually, if that's okay. Yep. And the okay, so there's there's a reason that I wanted to do an episode about this. I I listened to an episode of Dakota with Ne Lai Patel.

And'cause I do listen to N Dakota, not every episode,'cause some of it's a little bit um I'm just not interested in some of the topics, so it depends on who's on. But this particular one they were talking about what Uh, some of the challenges with flagging metadata on photos and saying if they're real or not. Because the problem is that it the world we live in right now today, and the right now today for those listening in the future is the middle of March of twenty twenty six. So

Here we are in the middle of March twenty twenty six and AI generated photography is real. It's getting better all the time. Sometimes if you know what to look for scarily scared. Yeah. It's quite scarily good. And it makes it possible to generate a an absolutely fake photograph. of something and say this is absolutely uh a cross between a zebra and a monkey and it actually is a real animal that we've genetically engineered and the photo is gonna look so real

And well, it's an absurd example, but you know. Uh so I guess what I wanted to do is let's break this down into pieces because I actually really didn't uh didn't like the decoder view of things because the problem was that th they were simply saying it's an It's an insurmountable problem. Like it can never be resolved. And and I don't think I agree with that. I actually think that there are things that we could do.

So so are they saying'cause I I didn't get to listen to that episode, are they saying it's an insurmountable problem r right now or it will become an insurmountable problem? Oh okay. They they didn't use the word insurmountable, maybe they used the word intractable, but either way you slice it, it's a problem that they believe cannot be ever solved. Like as of right now it's not solved and in the future it it cannot be solved. Okay. So my problem with that defeatist attitude is that

They're just not I just don't think they thought it through enough. So, okay, what what am I going on about? Uh so the episode was talking about uh how there's a metadata system that's designed to contain a whole bunch of metadata for different things about photos and they're gonna load it up with Um like

this is a genuine photo, this is a fake photo, whatever else. And it's like, well who gets to decide what's flagged for what? And we'll get to that. But they were simply saying, Oh, it's just not it's not possible. But I actually think it is possible because let's start with how do we know for sure that th this is the the fundamental question. How do we know for sure that a photo that you look at is what it is.

So if we're gonna start at the very basis, we've gotta start at the very beginning. And the very beginning is um the raw file, the raw image captured from the sensor. And it doesn't matter what your sensor type is, right? If it's a CMOS sensor or not, whatever sensor it is does not matter. That raw data should be unprocessed. Now, if your camera does like most cameras these days, will have a JPEG version and it'll create

its own version and you can adjust a whole bunch of stuff. You can say, right, well, I want it to look black and white. When even though it's not a black and white sensor. Whatever. You know. Uh I wanna make it vivid and I want to p the colours to be brighter and pop. I just want um maybe I just want seepier colours for some reason. I don't understand, but that's okay. It's an artistic choice. I'm not across artistic stuff.

f fully clay and I don't mean to do disrespectful to people that like sepia, but whatever. It's fine. Okay. It had its time. Quick. Yeah, I n yeah, thank yes, it did. Thank you. And that time is gone. Yes. Get over it, people. Anyway. So um my point my point, what was my point? Yes, my point is that as soon as the JPEG has done something to the RAW to create what it believes is, you know, the photo, technically that has been modified.

But it's been modified not necessarily by AI, but it's been modified by an algorithm. So to me, AI and algorithm are they are subtly different. Um, but in some ways it there's still a modification. So the JPEG that you look at in your camera has probably also been sharpened. It's been denoised probably and it's not that raw image. So I guess my I guess my first point is if you were going to certify and say this image is an ap accurate representation of reality

through that lens. Uh oh, through the lens. Yeah. Like a great podcast that that I I once heard of. Anyway. Um but yes, so that's a genuine representation. Um of and for the listeners that aren't getting the joke, obviously that that was Clay's podcast through through my lens. So it's a good it was a good podcast too. I miss that. Okay. Anyway, yeah, you should do more episodes at some point.

Um it it's it's it's yeah, maybe maybe that's just yeah, just not true anchor. Um anyway. No, because you can't. Um never mind. Alright, sorry. I'm derailing us. It's my fault. Um anyway, so if you were to send that raw file around, the the challenge that I had about the whole decoda conversation is that They're saying that it's not possible

for you to then so anyone can flag that data as anything. So therefore you it could be flagged as being AI generated when it's not, or it could be flagged as not AI even if it has been generated by AI. Mm. But if you've got a raw file that's been captured from that sensor and written directly to the to a file and you actually were to encode it and in in like

get a checksum based on that encoding. Why? How is this Like this surely should be possible to then have a m a manufacturer and say, you know what, all of the sensors and all these cameras, they're going to be accredited as being if you extract a raw file in its unmodified format, here is a checksum you can validate. It is auditable it's auditable. And we are r a reputable company that is regularly audited.

This image is reality. That is something that is absolutely possible to do. But Yeah, it it ac Yes, go ahead. Oh no you No, but I wanna hit a butt. But No one's talking about that. No one is talking about that. So like w to me that's like that is that is base one of where you start. I'm I'm I'm curious what you thought. I mean it

Yeah. I'm curious what you think about that before we get into the next level down. No, no, I I I I agree with that. And the thing is that so if we if we put standards in place And I and and I know there are a few s uh open standards I believe that are probably not being accepted uh not being adopted yet by all the companies. I think I think Sony is one of them that that adopted the um basically it's almost like a rubber stamping of Of the raw file.

Um and if something comes out of the raw file, like a JPEG, there should be a signature that that raw file has been stamped and the JPEG should have the signature of that stamp. Um so There are way you know, like like two factor authentication, right, is the thing, right? I mean, you can confirm or you can you can um verify by having m multiple signatures, right? Mm-hmm. Mul multiple identifi identification um um signals. Um something like that should be

uh and and and is being worked on right now, something like that is th it's the only way that we will get ahead of it. Um and so if they're saying that That it is a a problem that is is not going to be solved. Um I'm not so sure if I w would agree with that. If it's something that isn't solved at this moment

Um there's work being done on it. Uh we we have we don't have enough information in terms of having deep fakes out there. Um we we don't have enough data out there, right? For So like as time goes on, we're we're going to have a harder time confirming or verifying that something isn't deep fake because we we have We have um a time there's a there's a cutoff. Like if you look at a picture of Bill Clinton, there's a time where we know this picture has been on the internet at XY amount of places.

But as time goes on, we are entering a of time now where a picture of you is on the internet, but another picture could come up of you on on the internet and you're, you know, sitting in a whiskey whiskey bar and you say, Well, I've never been in a whiskey whiskey bar again since

twenty twenty five. But a picture of you now shows it There's uh is you're gonna have a hard time verifying that something like that hasn't happened without something like a digital passport or some sort of rubber stamp that that shows that this picture actually was not generated from a true camera, but

Created from AI itself. Yeah. Or something to that effect. Yeah. So I guess what I'm what I'm thinking is that if you had a system that was Every CMOS oh C I keep saying CMOS, but you know what I mean, every camera sensor that you've got has a unique number that is i that is encoded with it. Yes. You encode the file against that number. There's a manufacturer code as well. Mm-hmm. That should create a completely unique traceable thing. Uh traceable image.

Yes. So if you were to do that and you were to get all of the manufacturers that made these sensors and made these cameras to sign up to this and say, Alright, so could a computer theoretically could it create a raw image. Well of course it could create a raw image if you fit it enough and say here, make another raw image. But it's like, okay, so but you're encoding it with an actual number from an actual camera.

So the next question is, okay, so well I guess you could probably clone the manufacturer number. You could probably clone or create a another number. But then with the time and the date stamp would also need to have to line up to when a real photo was taken. Right. It's extremely difficult. And I would suggest it's That this requires

a little bit of cooperation between these companies. But then if you think about domains and domain registrars and and the chain of of trust in in domain certificate certificates, you know, like root root certificates and all that stuff down to You know, the certificates you might get through something like Let's Encrypt or whoever.

It's like there's a hierarchy of trust. You should be able to achieve something very similar using photos if you were to actually encrypt it properly at the source and get not encrypt it, get to check some at the source. That if anything was changed, you would know. So let's assume that that is better to do than to not do it, to at least have some basis in reality. So this is what was taken in reality. So the first challenge for myself against this is

What if I use a filter? You know? What if I put a red light filter, a vi ultraviolet filter? You know, obviously not a neutral density filter, but you know, maybe even if I don't know. Like if you're using filters on front of that lens Is that still reality? So I'm gonna tap out of the argument at this point because as soon as Any photographer does this, you could argue that that's not reality because people don't see that. But then the sensor sees things that the human eye doesn't always see.

Absolutely. And sometimes it sees things that we do see that it doesn't see. Uh yes, exactly. So ultimately I think that I have to draw a line there and simply say, you know what, I'm gonna ignore the possibility that someone's put Vaseline on their lens and hence that that's not reality anymore. But you know what I mean? It's like Let's let's just accept that that is a f artistic license. However, once you have it in a digital format, if it becomes a JPEG and you are in camera adjustments

There is nothing stopping you from having I I I would suggest two systems. The f like like two two mechanisms. The first mechanism is Here's the raw, it's the original. Here's the JPEG that's been created. Here are the settings that were changed. So You could flag it under a category of in camera adjustments if you want to and add that metadata and then you show the image history and then then that becomes auditable.

So a human can then look at the RAW file, they could look at the JPEG file, they can look at in camera adjustments, you know, CPA tones. For whatever reason, adjust it, denoising, you know, like whatever was applied and say, right, is this photo represent representation of reality? And this is the problem. As soon as you do that, it's like, okay, well Well what is it? qualifies as modifying an image.

What when what qualifies? It'cause if you're changing that colour to seepier tones, well you're modifying it. Mm-hmm. And maybe that's okay because that's not the deep fake AI generated photos that people are worried about. But you gotta start here and think about it like this before we get to the AI generated stuff.

So I don't think that that's too big of an ask because most cameras will take both raw you have an option to take RAW plus JK JPEG. Right. So you can you can encapsulate that in a common file and you can have Metadata doesn't matter how it's stored, metadata that says I applied the following filters, it's been denoised at this level, like here's everything that was done to the photo. That is now traceable and auditable. And so far

That's not hard. So then what? The next problem. Now we get into Photoshop. Yeah. And things start to go to the next level down. So let's just say we wanna get out the the the Band Aid tool and the and the patching tool and the uh magic eraser tool. But depending upon what software you're using it has a slightly different name. But you know, whatever you want to call it, you c you you draw a circle or a box or you click

a circle or a pattern over a part of there's a defect on the wall. There's a crack on the wall. You don't want that crack showing up in your perfect photo. So what you do is you drag this tool over it and it erases magically the crack from the phido. So that is now using an algorithm Or potentially AI driving an algorithm

potentially. Uh well every they they call everything AI right now. It's'cause it's buzz, right? But it's so irritating. I know. It's not AI anything, it's just an algorithm, but whatever. Yeah, you're you're absolutely right. It g it gets annoying too, but It does. But the the question is, as soon as you start doing that Now, if I take a crack out of a wall and the photo is of a person with the wall behind them, getting rid of that crack in the wall, that's not

m really modifying the intention of that photo. Like that photo is just of that person. The wall is a distraction. By getting rid of the crack, you're drawing the viewer to the to the person in the portrait as opposed to oh look at that crack in the wall. It's like don't look at a crack in the wall, look at the person's face. What are you doing? Mm-hmm.

Are you are you are you trying to tell a Doctor Who story here? That's not a crack in the wall. Because if there's no wall there there'd still be a crack. Anyway. Wow. Now that that that's a r in reference if you're not a Doctor Who fan, you're never gonna get it. Anyway. All right. So um yeah, so uh what was the point? The point was that if you make that kind of a modification, then you could argue that is not changing the intention of the image.

But I mean you could easily then take that tool and take that person's nose and erase their nose. Mm-hmm. It's like if the subject is like Voldemort, who has no nose For those that like watching Harry Potter, mm-hmm. He has a snake like nose. 'Cause reasons asked J.K. Rowling. Slitters. And hope They're Slytherin, yeah, Slytherin guy, whatever him. But point is that's not reality. Cause like that person in that portrait had a nose. And now you're saying they don't have a nose?

So that is substantially changing the content of the photo. The problem that I've got with tagging is there are certain modifications that are going to change the intention of that photo. And that is the argument that they were arguing about with AI. is that how do you flag it? Because it comes down to personal opinion. Mm. So you can store all the metadata in the world.

Doesn't matter. But I mean there are other things that you could flag that I think you could probably get away with and people are saying, Well, well that's fair use. That's okay. That's not open to debate, it's totally fine. Like If I'm taking an a if I'm changing the white uh the white points and or the white balance in a photo, uh if I'm tweaking my colours and my colour histogram, if I'm blurring certain things

to draw attention away from them, or if I'm sharpening them to clean them up a bit. If it's not excessive, then that's probably not going to change the intention of the photo. So I kind of feel like you can't really and I don't think we should ban image modifications in this manner. And we're gonna get to the AI stuff in a minute, but it's like i you wanna be creative.

So you shouldn't ban modifications to images. I mean good luck trying to ban it anyway. Mm-hmm. I just think that tracking the metadata of what's changed and potentially why it's changed you know, might be valuable. And if you actually store a history of each of the stages that it was modified, that becomes an auditable traceable history. And then you can track back to what is reality. Mm-hmm. So I mean this Yeah. So the thing is that okay, if you're if you're talking about m tracking

modifications that someone has done. We you might get pushback from someone who is uh, you know, uh uh you know, high selling artist, right? A ph photographer who decides, well, uh, you know, my my my photography is worth, you know, t twenty thousand dollars a frame, right? Mm. Having a record of their modifications is having a record of the of the

the the secret sauce, right? Of course. Uh so so you y the pushback might come from those in the actual practice of it. Um now That's fair. I would submit my tracing just so that people understand that well what was captured actually was captured by my sensor and modifications were done by me rather than um rather than

you know, some algorithm or AI. Um do we get to a point where you can have AI do the edit for you w by giving it instructions and saying, Hey, you know, modify this picture for me in this way uh to mimic, you know, uh uh you know, another artist look. Um do we like like do we actually want to have that be

flagged um because it it i it isn't modified by your hand. It is modified by instructions, right? It's so mm. Yeah. So I guess my okay. So There's a few things in there, so let's let's let's start with the um let's start with the Is it modified or the selection made by human versus an algorithm or by AI? Mm-hmm. Um those changes were made. It kinda to me, it doesn't matter quite so much, maybe not at this point of the conversation, but it it's it matters. But it's like

what is the intent and who did it mm I think if you flag this information, then someone can decide. Okay. I like I I Look, I absolutely see what you're saying though, because what you could say is well well I'm I I'm a p if I'm a professional, I've got these really dreamy bloody settings that make all my photos look amazing and that's proprietary. It's like fine. You don't have to submit it. You don't have to. So what I'm what I'm trying to suggest here is that is there a mechanism for us

that we could we could try as a you know, from the ph pho photography sort of a I don't know if we can call it a community but whatever. Is it possible for us to agree that there would be a methodology And the methodology would need to be is is it traceable from what you submit back to the from this is reality. Gotcha. And if it's traceable, then you can submit and you can say this is reality.

Because and and why that's important is it to me. I okay, this is why I I want to talk about this, is because I think this is crucially important. in the world that we're in right now. Because what's happening is that people are taking photos of real things that are happening. Mm-hmm. And then they're being told that's fake news. That's not real. That's been generated by an AI. There is no Absolutely. And because there is no chain of trust, because there is no authenticity of this is reality

Has it been modified? Yes, the color was adjusted. Has it been modified again? Yes, well, you know, this particular photo had a crack removed from the wall. These things are not auditable. These things are not traceable. And therefore it's easy to label them as saying, Well, that's not true. That's not real. Mm-hmm. So So if if someone were to just just so I understand, if someone were to have

the steps recorded in let's say the metadata. Yeah. Or baked into something like a D N G, right? Yeah. Which doesn't have s it's a sidecar but like it's actually built in. Um let's just say um You know, they don't want to share it, but it's there if there is a question, then they could they could submit it. Absolutely. I think that would be totally fine. And I also think that

The I think I we're sort of jumping around a little bit, but that's okay. I think that if you wanted to be l Okay, so let's say you're shut a stock.

um or five hundred PX or let's say you are a major photography website that where where photographers can submit their work and get paid for it. Mm-hmm. If people want to get paid for something, uh then those websites would need to be absolutely certain that what they flagged in there as being modified or not modified, real or not real, was traceable because their credibility is now built upon The fact that that chain of trust exists.

And it's auditable. Because if it's not so for example I can foresee a future that's not that far off. if we don't have some level of chain of trust in our images and our videos saying that this is real and this is not, um, then we're gonna reach a point where news organizations who are practically now barely trusted, or if they're not, probably shouldn't be. Uh you know. Uh

And you can't trust what people post on YouTube, you can't trust what people post on social media because you you have no idea if it's real. And so this is a massive problem because it means that We we've lost our ability to trust anything that we see or hear unless it's our own eyes. So in other words, all of the everything that we've gained in the last forty years of internet and being interconnected, news, media and all that, maybe even for even longer.

mind you, if you're believing the newspaper a hundred years ago, you're probably just as much of a mug as you are today. But never mind that. You know, it's like if there is we will have lost massive ground because now we're down to I only trust what I see smell, taste and touch myself, which is therefore limited'cause I have a limited scope of movement.

I cannot observe what is going on in other countries unless I am in them. Mm-hmm. So if I have no trust, I have no ability to assess, if I can't assess then I can't actually have a a an informed opinion. And fundamentally that is where this is all coming from, I guess. So Yeah. Yeah. So my suggestion is that It is okay to modify images. That's fine. You just need to have a traceable chain and examples of what was happening at at key points in that transformation. And I think it's probably

problematic to have a scale, but it might help with a scale because it's like, okay, what I mean by scale is let's say you're making a modification. Well, you could say, let's start let's go on a scale of one to six, just because I d it doesn't have to be one to six, but let's just for any for John's example of John's um proposed random like attempt at fixing this problem. Okay. First of all, one. A modification level of one.

No modification from original. And you know what? That should actually be a r um a m scale of zero to be honest, but whatever. Okay. Two, a minor level of uh balancing. So for example, you know, balancing out colours, histograms, that sort of thing. Um maybe three might be um removed elements. without impacting the content. So that'd be things like minor airbrushing, removing

extraneous stuff in the background. There's a chair in the background. I don't want a chair in my background. So you just delete that. Bye bye chair. Right? Good. It was distracting anyway. Didn't even realise it was there when I took the photo.

It's fine. Uh like at level of four, things are getting a bit more interesting now. So you go I've either added, I've moved, or I've modified an element. Mm-hmm. So this could be things like background changes, modifying the the appearance of an individual. uh starting to modify the content in some way that may impact that context. So then we go up to a five, which would be like things like I've merged this image or image elements. So

There's more than one source image that's created a compound image and the content of it now no longer matches the originals of either a source material. So let's say so w wh what about a panoramic if you were to have panoramics? St stitching. Panoramic stitching is Uh good point. I didn't think of that. This is why I'm talking to you about it.

Capturing it in one frame. I don't have an immediate answer for that one. I w you might need to have a separate section for stitch together drawings of the same scene. Mm. I'm'cause as a photographer for the cricket club for seven years now and I think and this probably been my last season'cause I'm yeah, I'm Y you you know you you you know you said this is your last season uh twenty seasons ago.

Oh, I know, I know. If you're if this is your subtle way of saying, John, I'm not sure I believe you you're you're probably right to be skeptical. Let's see what happens. But for the moment it's my last season. Anyway. But okay, so what happens is, you know, like Stuff happens. Kids get sick, they don't show up, their parents' car runs out of fuel or electrons or whatever the hell happens. They don't make it to photonite.

So you take a photo of the team and the team has eight players in a junior grade, but you've only got six kids there. Mm-hmm. Now I'm not gonna say, sorry, six people that were able to show up, you don't get a photo. Uh I'm not gonna say that. So I take a photo of them and I have two empty spots where I will digitally splice in the other two people, other two kids, when they show up subsequently. I'll take photos of them in those rough spots.

and I'll stitch them together and you will look at it, Clay, and say, That's terrible. the majority of people will look at it and say Oh my kid was there. They were all in the same Yeah, my kid was in the same room. And I'll be like, No photographer would look at this and think it looks fine, but that's okay. Uh anyway, but but that's that's what I had in mind when I thought about that one.

But I'll I'll have a thing about the panorama, it's a really good point. And then I think the last one is a s is is a six, which is it is a fully generated image. So it is AI created, it's entirely from a range of images ingested into a system and it does not represent reality. Mm-hmm. So you could grade a photograph roughly on that sort of a scale overall and get a sense for

I'm looking for photos that have not been modified at all. If that was embedded in the metadata of the photo, it was respected and by the the manufacturer of the camera, the manufacturer of the sensor. and by the photo website, then that should be something you can search for. So the problem of course is

the flagging and tagging and definition and of these things. Like'cause you could say, like you pointed out a perfect example, Panorama. It's an exception to this list. But I mean, going back to number four where I add m move or modify elements Mm-hmm. I want to move the chair. slightly left or slightly right. Does that change the intention of the photo? What is the intention of the photo? Because it's like, mm, okay. Um a more extreme example might be um I've got this photo of somebody

who's holding a gun to someone's head, which is, you know, like purely artistic, just not getting at anything at all. Mm-hmm. Um, and it turns out this person is actually just a r a normal person. But if someone adds an element like a badge. a formal badge of some kind of some insignia onto that person's shirt. And that person is just a normal person. They're not actually representing any kind of official body or anything.

you just completely change the content the context of that photo. So suddenly a person who should perhaps not be doing what they are doing is now doing said thing, even though they are not one of these people. Photo goes viral and it's like, but it's not real. Yeah. That's not what happened. So if is it traceable? Can it be made traceable? So Any software that you've got that can modify an image should be adding some basic set of metadata of saying, well, the following things were changed.

The repair tool was changed on a previous cloud in the sky. But the problem is Um uh and the same repair tool could be to reduce some r remove someone giving a rude gesture with their finger. Or they could add a rude gesture that they're not making. Mm-hmm. The problem is that the tool, the software tool, can just say, Well, I use the repair tool. I don't know what the who the human did. And if it's the AI making the decision, it has even less idea'cause it's an AI. Yeah. You know?

So it's like uh here I have I can flag the data that yes, a change was made, but the judgment of what on earth that change does in the context of the photo has to be a human judgment. It has to be a a a judgment based on before and after. So because you can't this is this is kind of the conclusion that I've reached. And this is one of the reasons why um Neal Eye and Friends got so

tied up in knots with it is because they were just focused on that part of it. It's like you can't automatically classify an image's content or the changes made to it. It has to be a human judgment. And then second problem is how on earth do you like police what's AI and what's not if you don't know where the image started from? So that's why I started this this this idea, this proposal.

from starting at the source. So yeah. I said I foresee a time. What I mean is let's say I want to be the world's first ever fully trustworthy, fully transparent news organization. Okay? Mm-hmm. It could happen. Maybe someday. You never know. Hey, it's possible. So I am like the world's most absolutely correct factual news organization. Suspend disbelief.

We're doing this. I only want to get photos that I know are real. So how do I do this? Right now it is not possible. I cannot definitively prove that that photo that I'm using in my news story came from the real world. at this point in time. So I would go to a reputable website like let's call it five hundred and one PX and at five hundred and one PX Like they have an entire section that they guarantee is real world imagery of real world. raw data. They've got a full chain of change.

A full cha uh chain chain of trust from the original sensor, the original camera serial number, time and date, everything. All auditable by human. You can look at before and after, see the modifications made and say what I have is a representation of reality. I can trust this photo at this point in time. Then I use that on my news story and I have absolute confidence that what I've got isn't fake. So can AI defeat this?

Can you use AI to create something going all the way back down to the bottom? I don't know. Probably, maybe, but I'm s there's there's a way you can do it. Sure. The thing is that okay, so we we had our film era. Now you know, people always said oh you know, f film photography was to capture it as it was seen. Right? You you you did your manipulations on the you know, your your your your your choice of film, um what aperture, what shutter speed, right? And

Your p your picture was captured. Now yes, there are modifications. People, you know, people pretend like we didn't modify film photography. We did. You know. Um so mar manipulation of film w has i i you know, was a part of our history. Now let's just say if you take a picture with a camera that is part of the consortium of cameras that have you know, sensors that have

a stamp that the second a picture is taken, right? So you do your manipulations before you click the shutter. Right. You you have we have the ability to to gauge you know, how bright um the actual captured frame will be. How uh how in depth the the the you know, separation from back to front.

Did di this is i if you do it this way, by the way, you you will create a s a c a class of photographers that will will become sort of like elites, right? They're gonna feel elitist. But let's just say there is uh a picture that's captured, right, and let's say five hundred and one PX has a section, a picture that's captured that has never been manipulated after it's been taken. The second it has been manipulated, that stamp Fates.

Because it's gonna be hard for you to create a a to create a file where you have gradation. Right? Because in in I mean uh you know, in a d in the digital world, right? It's on off usually. Mm-hmm. I mean maybe there's a way to do a gradation. Because then y your your fra your levels of okay, um, you know, you go to one from zero, zero is a full stamp. You go to one, it fades.

You know, it becomes green. Yeah. And then you go keep going down until it becomes red. And red means you've actually completely manipulated it. Mm-hmm. Of course. So so I you know, on off is d is easy to do. Well But that that is easy to do. I I think the other fundamental problem with my proposal and I'm Just I've gonna admit this up front. I keep saying this, it is traceable, it is audible auditable, but a judgment needs to be made by a human.

And that is the fundamental failure of what I'm basically describing is it's gonna take a lot of time and a lot of effort. to prove and audit it and say this is actually real. Mm-hmm. This is what happened. Mm-hmm. And most people are not going to invest the time and the money and the effort to make those judgments. True. But

They're not gonna challenge everyone. So that's the counterpoint of the counterpoint of the counterpoint is You know, I am not going to be constantly confirming the chain of trust is in place for every single image. Especially if I have a website like five oh one PX for example, fictitious website that has no idea. No, that's real. It's coming. It's coming. Yeah. But but if you think about it

I think this is actually inevitable. Yeah. Because people are going to be so sick of it. They're gonna be like, well I can't trust anything anymore. Because be w a we're already at that point now. Like if you I think we've passed that point. Uh yeah, we've passed that point. So if you don't believe that this is a problem, then you haven't been watching any kind of media, you haven't been watching TV, radio, listening to radio. I mean, even podcasts are being

like made with fake voices and so on. Mm. People saying things they didn't say. Right. It's like it's it's ridiculous. Yeah. So where is that chain of and I have even started thinking through the the consequences of let let's say just my voice. Mm-hmm. Like if someone clones my voice And say, Oh, John Shiji said that um the the earth is flat and it turns out that uranium two hundred thirty five is actually totally harmless and you should have three spoonfuls a day. It's like well, you know what?

Yeah, I'm just saying okay, everybody, dear listener, please do not ingest uranium two hundred and thirty five. One teaspoon is n is more than enough. Don't even have that. That's the end of this public service announcement. We're gonna return to our programme now. Thank you. Okay. So it's like. Yeah. There's like a big springy bed at the bottom or something. Anyway, thing is, right. Command. Okay. Okay. I've got to come I've got to bring it back. I'm sorry. Okay. It's just it's late.

Um yeah, so I um I think that the whole audio, how do you do audio is is a little bit more diff is a little bit of a different problem, but it's still not insurmountable. So the the the problem I have fundamentally is If we aren't thinking about ways of building a chain of trust, then why are we even bothering? You know,'cause it it's like if you and I we're taking photos for

for posterity, for capturing the moment, capturing emotion. We are in so many ways not the problem. You know, the problem is when the photos that we may take someday are used in a different context to create something that didn't happen. Right. How is it that we

And it's like I don't know if anyone would ever trace photos back to what anything I've taken and said, Oh, well, you know, John took a photo of this waterfall once. Turns out the waterfall's actually flowing upwards. I'm like, Well, really? That's kinda cool. It's no it's not. That's not real. It's all John's fault. John said that the the w the photo that John took had water flowing up into the air and I'll be like, Yeah, that's the kind of

That's some crazy stuff right there, man. I know. Surely that's not real. Anyhow, but yeah, I keep on throwing out ridiculous examples, but it's like the liability is is going to eventually be on the supplier of the image. So who supplied the image? The image was supplied by the news organization.

Where did the news organization get it from? Oh, well, you know what they did? They didn't get it from the five oh one PX website. They got it from an AI generated image because they're like they were too lazy. Mm-hmm. Because this is happening a lot in news reports. Like you scroll through a news website today

Most of the images that you see on there, like the the title cover image or whatever else. Mm-hmm The beat the the the ones that wants you to click. Not mo not most of them. Yeah, not not most of them, but uh but multiple multiple of them are AI generated. So it'll be like some generic like stock market crash. They don't show a photo of the actual um New York stock exchange tickers

and so on'cause that's been done, that's boring. So instead they have some AI generated thing that's got an overlay of a graph and a big red arrow going bouncing bouncing downwards. Mm-hmm. But I mean a lot of the stuff they do, iconography and

And artwork and images now as just AI generated because it's quick and it's easy. Just type it into your AI generated to chat GPT or whatever the heck. There's a whole bunch of different tools for this now where you can build images from text and say, give me this, and then they'll they'll look at a few and say, Yep. We're gonna use that one. Mm-hmm. You know? Yeah. So it's like so this is

This is the problem. So you start with those the the news organizations. Okay, so it's your fault you're spreading misinformation. You're spreading fake news. Where'd you get the image from? Oh we we got it from this other site called, you know, five oh three PX and they only have AI generated rubbish. We figured that was good enough.

So it's like, okay, so you've broken that chain of trust. I as an individual am now gonna vote with my feet, I'm never gonna look at your stuff again'cause I know that you would just have fake news. Absolutely. So I will go to the websites that have Photos that I can trust, images, vi v videos I can trust.

And I can trust them because there is a chain of trust that is demonstratable. I don't need to see a demonstrator, but I need to know that there is a chain of trust there. Mm-hmm. And I actually do think that this is achievable. I'm not I don't think it's gonna be easy, but I don't think that we I don't think we can't I think that we must do it. Mm-hmm. And if not, what I'm suggesting, something something like it. Because if we don't, we may as well just turn off Every

way of ingesting in images that other people take and videos other people take. So all your T V so your smartphones, everything, just turn them off because you can't trust what's shown in'em. Yeah. Great. So Uh w what if the person let's say let's say you have a committee, your news committee, um something happens, you know, uh some bombing happens, a the person who takes the picture um and they happen to die. Um the file itself

Like I just just to think hypothetically about uh about this for a second. Like so if if the file was um on a memory card but it wasn't so the way something like this would work, right, is you you'd have to have a trusted software that is That is registered to you or the camera and the file itself. But let's say somehow the handshake doesn't happen. We're hypothetical, but

you know, then something that that was captured, right? Some sort of massive event, uh it was only captured by one person. Let's just say the person happens to die. The handshake doesn't happen. Uh it's it's it's an extreme s example, but yeah. Um th we d we don't have now the submission of something to prove that it was actually uh an an actual picture taken. That's true. That's why to me, in a way, the on off It's always on.

Until it's been t you know, like the canary in in in the in the mine, right? Yep. If if it's dead, hey guess what? It's time to bail or you're already dead. then you are essentially

Yeah, you can no longer trust that that is an accurate representation of reality. Mm-hmm. If that's is that what you're kinda that's what is that That's that's kind of that's kind of where I'm going, yeah. Yeah. Yeah. And and and I absolutely see that and that's fine. And I and I agree with that. And that's why I say, well Um, that's why we need a way of encrypting it. Not encrypting it. Go, I keep saying encrypt. What I mean is getting it a checksum that has been encoded

with the serial number of the sensor, the serial number of the actual camera that took it as well as like manufacturer information as well, time and date. Yeah. So if anything changes at all then that checksum will change and that checksum is embedded in the original. So if anyone modifies it, you know. And this is this is exactly what every

So if you want to go to a um to download software, some software packages from people that are security conscious, they will give you here's the file, here's the checksum. You download the file. You take the checksum, you run the checksum check against it, and if they match, you know that that file has not been modified. Right. So this is exactly the same idea but just for for images.

Problem is as soon as you modify it, then you are changing that you are destroying that chain of trust, which is why I'm saying absolutely

it's fine to modify images'cause you need to modify them. You need to tweak the white balance, tweak your histograms, you gotta do some denoising, you gotta get rid of the crack in the wall. Yada yada yada. Mm-hmm That is fine so long as you have a traceable documented path and you can say, Right, I made the following changes, this is what it looks like now, but wrapped up in the same file is this is what it was.

at the beginning. So you any human can look at the two and see what was done and say, Yep, that's an accurate representation of reality. So this this is my proposal. But then again, um You know, l your point about film. Film is the closer to that because it's more difficult to modify.

Yes, you can Yeah. Uh what I is you had to jump through a lot of hoops, but yes. Yeah, you do. But I mean and what they call it, uh dodge and burn. I can't remember the name of the the the thing that they you can you can do with film. I'm not a film guy but I know you are, but still yeah, you can still modify those. Uh, but at the same time it's a lot more difficult and you need to have a lot more talent and skill to do it. Right. So in any case, Clay, I guess my problem is

I don't have all of the answers. I never said I did. But what I do have is a reasonable amount of concern about the path that we are on as a a civilization across the and I I mean that across the board. Everywhere in the world has this issue to varying degrees. So our world expands as we improve communication and as we are able to see more of the world happening around us from afar. We don't have to go there to see it. But as we erode the trust in what we see.

Suddenly we shrink again back to where we were. hundred years ago or more, where there were no newspapers, no radio, nothing. Because in the end, even though we have all these amazing technologies, we've now made it possible that we can't trust them. Which is hilariously sad. Yeah. Yeah, well we're building we're building silos.

I I feel like what we're doing is we're dragging ourselves back to the Stone Age to an extent. Right. And we have we have the Stone Age of different titles. So you over here in Australia you can't us you know trust trust the stories you hear from here.

and we can't trust the stories from there. It's like, wait a second, you're selling beef to our our who peop you know, the people who are s you know, again, silos. Yeah. Yeah, that's true. And and if anyone told you that you can throw this stick in Australia, we have this stick.

You throw the stick in Australia and it and it comes back to you, you'd be like, What? That's that's B no way. That sounds like AI. It's like here no I'll give I'll show you this photo, it's like, nah man, that's fake news. Uh huh. Exactly.

Cool. They are so cool. Is it cool if it hits you in the head though? Oh no no no. It hurts, really hurts if it hits you. Believe me, because that's happened to me. So the first thing you do, okay the first Uh when you're learning to throw a boomerang, okay, from personal experience. The first half an hour you are angry and you are frustrated and annoyed that it will not come back to you, no matter what you do.

And then there's this moment of elation as you see it arcing up in the air, doing a lovely little turn and coming right back at your head. And then you're like frustrated and annoyed as it clips you on the side of the head. And you're like Ow. I am now annoyed that it hit me and it came back to me. So you start with a boomerang ha hating it'cause you can't get it to come back and then you hate it because it comes back. It comes back with a vengeance. But anyway, it's fine.

It comes back.'Cause remember what you the the the energy that you use to throw it out is uh roughly the same energy that comes back. So If you're giving it a good w good good uh riff with your arm, that's the same amount of force that that thing's gonna hit your head with if you're not careful. And yes, I did actually duck so that it wouldn't get me in the damn head. And and all of the um

Um uh so various names when I was growing up we were they were called we we we we call them aboriginals. These days they prefer either the first Australians or you know, native Australians, whatever. Um they were absolutely laughing it up. They thought it was the most hilarious thing'cause I was eating dirt because the boomerang came back and almost took my own head off. They thought it was so hilarious. I didn't think it was funny.

Sorry? Yeah. Is that th their invention? Yeah, the Australian the Australian Aboriginals invented boomerangs and Oh, I didn't know that. Oh Yeah. Absolutely for real. Yeah. Um absolutely ingenious. And I and these things are Kinda dangerous. Actually there's lots of different designs for boomerangs as well. Like you get ones that have got a slightly longer end on one end that's heavier and they're designed for taking down like

larger, heavier birds, so when it's more like a club on the end, like it's just thicker and heavier, but it still comes back if you throw it properly. Wow. I know, right. Very, very cool. But never mind. Never mind. So all right. Um I just bring it back to photos and everything though. Yes. I don't have too much else I w Boomerang it back to photos. Boomerang, yes, bring that boomerang back. My boomerang will come back to photos. Uh so I don't think that all is lost.

And and don't don't misunderstand what I'm saying. I do believe that this is absolutely achievable. But if it's something that we're not working towards, it will never happen. Yes. And the easiest thing to do is just throw your hands up in the air and just say, Oh, it's all fake news which

There are people that I know right now today who are saying exactly that. Every single thing that they see on the news, oh, that's fake news. Like right now, um there's a there's a war going on in the Middle East. It's like I've had a few people say to me randomly, Oh, that's that's all fake news. It's not actually real. Wow. And I'm like, okay.

So'cause I have um a friend of a friend who is living in Dubai right now and they are sending um video of bombs exploding i in Dubai. And it's happening. And it's like fake. So um but you know, but this is where we're going. Yeah. And I do not like where this is going. I can't fix this. But uh maybe I can help people who have more connections, more knowledge as to how this could be done.

maybe if we're a little bit less defeatist about it, I think it's something that we couldn't achieve. And we need to be thinking about how we do this, uh, and work towards it. Otherwise We're gonna lose everything we gained. in the last however long we've had transmissible media, in my opinion. Agreed. No, I agreed. And then the the fact that you're actually talking about it means that you actually do understand that I think a lot of people just put their head in the sand and just

hope it will you know, w i it'll just brush over. Um it'll pass us by, but it it's here. I think we do have to address it. Yeah, we do have to address it. I do think a lot of people are just feel helpless. They just feel like well there's nothing you can do. It's like if I can flag any photo and say it's not generated by AR AI or it is generated by AI

Um, that's not good enough b because that's can be easily gamed by anybody. So it can be gamed by people in both directions. So I can say, Well, this is a genuine photo of my foot. Um, and I can and someone can flag and say, no, that's an AI generated image of my foot. And I'd be like, hang on my man, you see my foot?

There's there's some really bad tinia, there's like a couple of really bad bruises there and it's a real ugly foot. So come on, this is actually real. Anyway, uh and then on the other hand you could you know, take a photo of But get AI to generate a an image of my foot or a foot and say that it's my foot and not flag it as AI. You can flag it as AI, you can not flag it as AI in either direction. And you have no chain of traceability. You can't tell what's true. Yeah. So anyway.

Yeah. No, I I think it's I think it's uh it's absolutely crucial that we we do figure out how to do the tr you know, get the traceability down or or some set or stamp to begin with. Yeah. All right. Well if you would like to talk more about this, you can reach me on the fediverse at chige at engineer.space or the network at engnet at engineer.space.

And if you are enjoying Pragmatic and you'd like to support us and keep the show ad free, you can by becoming a premium supporter. Just visit engineer.network slash pragmatic to learn how you can help this show to continue to be made. Thank you. A big thank you to all of our supporters. A special thank you to our silver producers.

Mitch Bjorger, Shane O'Neal, Jared Roman, Katerina Will, Chad Jeering, Ian Gallagher, and Jamie Russell, and an extra special thank you to our gold producers Stephen Breidel, Kellen Frodelius Fujimoto, Steve Branham and our gold producer known only as R. If you would like to get in touch with Clay, what's the best way for people? CW. But only the real ones not the same. Big thank you to everyone for listening and as always it's always a Coming on again. Thank you so much. It was a pleasure.

And we should do this more. Yeah. This was good. Yeah. I think we should do an episode on on boomerang. Um uh now you got me intrigued. You wanna do an episode on boomerang? Um that'll be interest and look honestly, I think uh in all seriousness, boomerangs are one of the absolute coolest things that the Aboriginals um developed. But there's some other really Yeah, it is very cool. Um, but there's lots of other things that they came up with as well that's that's really fascinating. So I

One of the things that has been good in the last, you know, few decades is that there's been a lot less of the stigma and a lot more of their history is being brought to light for people like when I grew up, we just weren't taught much of it. And that's changing and that's a good thing. That's great. So it's a very it's a very deep, a very rich culture and it's something that um

that uh the whole concept of the dream time for example is unique in the world. Mm-hmm. Um mm and it's just yeah, I dunno, I just kinda like it. I I can't explain why. It's just uh yeah, it's kinda