Why AI image editing isn’t “just like Photoshop”
45:50 Share

Support for the show comes from A T, N T. What does he feel like to get the new iphone sixteen pro with A N T next up anytime? It's like when you first light up the grill and think of all the mouth watering possibilities, learn how to get the new iphone sixteen pro with apple intelligence on A N T and the latest iphone every year with A N T next up anytime A N T connecting changes everything.

Apple intelligence coming fall twenty twenty four with theory and device language set to U. S. english. Some features and languages will be coming over the next year. Zero dollar offer may not be available on future iphones. Next up, anytime feature maybe discontinued at any time, subject to change additional fees, terms and restrictions apply C A T T doc com sash iphone for details.

Support for this episode comes from the current report. From data privacy to the future of TV retail media and beyond, the world of digital marketing is constantly in flux, so how can you keep up? Well, the current report is there for you each week. Marketing leaders on the cutting edge give you the latest insight. So if it's creating a buzz, they'll be talking about IT so described to the current report wherever you get your podcast.

Hello, welcome to decoder. I mean, I patel added and chief the verge and decode a is my show that big ideas, other problems we've been covering the rise of AI imagined very closely here and the coder and at the verge.

Overall, for several years now, the ability to create photo realistic images with nothing more than a chatbot prompt has the potential to completely reset our cultural relationship to photography, and in particular, how much we instinctively trust photos to reflect the truth. Every time we write about IT or talk about IT, we are loudly reminded by multiple people that the debate over image dating and the inherent truth of photos is nothing new. That debate has existed war as long as photography itself has existed.

And in particular, it's raged since digital photo editing tools have been widely developed. You've heard IT a million times. You've seen IT in our comments. You've seen IT on social media. It's when people say it's just like photoshop, with photoshop standing in for the concept of im generally.

So today we're going to die ve into this argument that responds and try to understand exactly what IT means and why our new world of A I image tools is different. And yes, in some cases the same. The reporter, just why the Better recently go into this forest.

And I asked her to join me in going through the debate and the arguments one by one, to help figure IT out, make sure, in many ways, A I imagining really is just a faster, easier version of what people have been doing in photoshop. Even photoshop itself now has adobie A I tech firefly built right into IT. But making powerful tools instantly accessible to everyone has big consequences.

And we are seeing those consequences right now. Say you wanted generate an image of Donald trump pointing a gun at come. All Harris, you can just ask, rock the AI chap up.

The elan musk has built right into x. It'll do IT with no issues because IT has very few of the same filters that have prevented competing AI products from depicting politicians or outright violence. High schools around the country are being iraq by so called notification apps that make a trivial to create deep fake nudes of female students.

The effort to do this is trivial. That is already happening, and IT is fast becoming a national crisis. You might say that these are all problems.

Photoshop and other image dator have to do all sorts of awful things with no guard rails for years. In fixed celebrity photographs have been a problem for decades. And even in the days before computers, you could create convincing fake images to mislead people.

But the difference of general AI is its scale. We're giving these sophisticated tools to everyone with a very little oversight, and that has landed us firmly in the tory. And i'd just be directly my view is that people say it's just like photoshop to diminish these new problems that A I tools are already causing to make them seem already solved or worse, not worth considering.

But I would remind you that we have hardly solved any of these problem when IT really was just photoshop. And then any proposed solution that requires everyone to just understand that every image they see is edited isn't really much of a solution at all. So what are the problems? What are the differences and what are the solutions? That's all in the AI versus photoshop debate.

Here we go.

Just won the bed. Welcome to together.

Hi.

thank you. Your article about why A I imagining is not like photoshop, I think, has some of the most comments in any piece we published last year. Let's walk through what's going on a AI image dating and why is in some way similar to the photoshop? To be, in many ways is different.

And the way I want to do this is by just going through the main arguments we hear. I assume the decoder audience is aware of what he image dating is and is aware of photoshop is, and honestly, the argument about whether images on the internet or real or faker or can be trusted, IT has indeed been region for years, but they're not quite the same thing. So let's start with the first argument, which is you can already manipulate images and photoshop. How does that relate to what we're seeing right now with .

A I am magdi if we're taking IT at at a tennis argument, it's not incorrect for shop has been able to make edits like this for very, very long time. But IT completely ignores the main issue of all of this, which is scale. I think you could probably point to that just being the the single aggregator that makes all this worse witches, if you wanted to do this kind of thing in photoshop or any editing software.

Really, there were so many skill and financial barriers stopping the general population from doing so, which usually meant that those edits had to be done with intent that could have been good or bad. And and I just meant that there was a little bit more of thought process behind that. You had to invest in photoshop, paul.

We find a free version of IT, learn how to use all of the complex tools. And IT would maybe take you about minutes, maybe an hour sometimes, to make a very photo realistic manipulation that you could use for the fairs verbis. If you want to do that, we're inclined AI kind of scrapped all of that.

It's now landed on phones and web apps. And you can just open a window, tell IT what you, anna, see, and it'll put IT there for you. IT completely changes the entire landscape of what we've been dealing with.

Let me push on the idea of scale for one second. IT is possible to go on a website and hire people to use photoshop for you for no dollars, for five dollars, and get an image that you want. You can go on to read IT today and say, I have a picture of myself with my my x partner, but the picture of usually, can someone photoshop them out of this picture, and people will do IT on, right?

That right now, for you, for free, is IT. The scale is enabled by how accessible and cheap IT is to do on a phone with an AI to all is IT. The even asking someone else to do IT for you for cheap or free requires you to say what you want, which is a deterrent to people. Where does the scale come from?

The accessibility is the main thing and there's gonna a lot of is especially the stuff that I think people are concerned about, right? Like the kind of like slot that seeing on x or twitter or whatever recording at these days. You're not gonna want to reit and go.

I want to make a very memorable picture of a politician doing something great task. You're gonna get push back. You might get banned from platform. You're gonna be bad from contacting those people in general, if you're given the means to just in act that without any barriers, which is affordably, what this tech is doing, half of the advertising for this is you can recreate anything that you can imagine.

I think that was pretty much google entire kind of ad campaign for this, to the point they've called the latest tall reimagine, like it's meant to be a creativity thing. People don't always have good imagination, intentions. And yeah, describing that to a natural human being is gonna comfortable or potentially illegal.

In some cases, those barriers are just completely removed. A I again has has no calms about whether you should be asking you to make something. It's just going to do IT.

That was always the case, right? For a while there I felt like A I chap pots and image genres would be restricted that they simply wouldn't generate images of real people, especially not politicians like trump or biden. Some like goods german.

I only very recently started letting paid users generate images, real people. And that still has a lot of restrictions around IT, but he has come along way. Now it's pretty easy to get your hands on an eye tool. Maybe it's open source, maybe it's built right into x like the AI chat bott rock that will do pretty much whatever you want with very few filters. The other thing I was think about is the notion of the people who want to run scams or tell lies, often quite lazy.

And so photoshop has existed for a long time, and some of the fake images we've seen have been so obviously fake, and they still have the effect people with bad intentions want, which is to confuse people or pretend there are really work sharks floating around the city and a flutter, whatever IT is. Maybe the increased quality of the fake images from AI isn't the point, like the technology, to be reasonably convincing about a lie. His art existed. Even the photoshop is someone hard to use. Do you think that the scale combined with the increased visual fidelity is actually meaningfully different?

I think so. I think that the barrier for who was being convinced by the stuff before was there was still a good chunk people that would seek something within a split second automatically assume that yeah, it's actually but actually looking at the fact that they've got, I don't know, eight fingers on each hand or something even in the early days of image generation. But the improvement of the general technology is definitely exacerbated IT.

And I think there is also an aero of people will just believe something with the narrative of IT works for them anyway. They don't actually need that much substantive evidence towards that. If it's somewhere aligned on opposition that they already have and they can use that as an illustrative guide, they're just going to run with IT. So I I do think it's making things considerably worse now that is convincing the people that used to actually try and keep an eye out for obvious fakes. But yeah, it's definitely exactly ating the issue considerably.

One of the other earns to be constantly here is that people adapted to photoshop. People understood the photoshop existed. There were some controversies around fashion magazines using photoshop to beautify people.

In the trouble, we stopped using IT. We've developed a sort of cultural vocabulary, understanding the photoshop and how IT can be used and should be used, and that will adapt to this as well. Does that feel convincing to you?

No, not at all. I think the photoshop argument itself almost feels outdated already because it's becomes anonymous with the active image dating in general. But the actual software itself was a barrier that hasn't really existed for a little while now since we ve started having filling apps and face two in and stuff on a mobile phones.

What IT did was create a real societal problem. But as since we had photos of introduced that became this idea that we need to be chasing perfection in everything that we do, and that was pushed in marketing images, and I was, like actively impacting body image stuff. And IT was also this idea that a picture should be perfect and you should maybe feel bad if it's not or that you're less if you're not able to take stunning images like that.

And ever since we have had these filtering applications, which were incredibly limited, you could meet you self, have smooths the skin, maybe slimmer raw line. They weren't massive changes. I think it's become this kind of thing that if we are given a tool now that can effect vely run with the limits of our imagination, people are gona do so.

We need to take a quick break. We will be right back.

Think scaling AI is hard. Think again with whats and x, you can deploy A I across any environment above the clouds, helping pilots navigate flights and on lots of clouds, helping employees automate tasks on prem, so designers can access proprietary data and on the edge, so remote bank tellers can assist customers. What's the next works anywhere so you can scale AI everywhere, learn more IBM dot com flash watch mix IBM let's create support for .

the show comes from the refinery a domino location and atmosphere are key when deciding on a home for your business. And the refinery can be that home if you're a business leader, specifically one in new york, the refinery, a domino is an opportunity to claim a defining part of the new york cy skyline.

The refiner atomy o is located in Williams s burberry lin, and that offers all the perks animatedly of a brand new building while being a landmark address the dates back to the mid nineteen century, its fifteen floors of classic modern office environment house with in the original urban art effect, making get a unique experience for inhabitants as well as the wider community. The building is outfit with immersive interior gardens, a glass stone to paint house lunch and a world class of space. The building is also home to a state of the art equal ox with a pool and world and restaurants and exceptional retail as new yorkers return to the office, the refinery, a domino, can be more than a place to work.

IT can be the magnetic hub fit to inspire your team's best ideas. This is the refinery that N. Y, C. For a tour support .

for this podcast comes from stripe payment management software isn't something your customer sink about that often they see your product, they wants to buy IT and then they buy IT. That's about as complex as a gets. But under the hood of that process, there are a lot of really complicated things happening that have to go right in order for that sale to go through.

Stripe handed les the complexity of financial infrastructure, offering a seamless experience for business owners and their customers. Example, strive to make sure that your customer see their currency and preferred payment method when they shop. So checking out never feels like a shore stripe is a payment and building platform supporting millions of businesses around the world, including companies like uber, bmw and door dash.

Stripe is healthy, countless starts and establish companies are like reach their growth targets, make progress on their missions and reach more customers globally. The platform offers a sweet of specialized features and tools to power businesses of all sizes like stripe billing, which makes IT easy to handle subscription, base charges, invoice and over current revenue management needs. Learn stripe helps companies of all sizes make progress, and stripe dot com that strike not calm. To learn more, stripe make progress.

We're back with verge reporter just what have been before the break. We were discussing the idea that photoshop in the I image editing exist as distinct forms of technology because one has historically required a good amount of money, time and effort, and the other now just requires you to have a smart phone that's created a problem of ease and scale. That argue makes the AI version of imagine a much harder issue for society to keep in check.

You just heard us introduce the next argument, which is that society as a whole will adapt to A I either way, we did with pra, I imagining and it'll be fine, but it's worth asking if the results of ambiguous image dating have actually been fine. After all, image dating is now a cornerstone of how we share photos online, whether it's face tune in a cell phone for a snapp chat or applying a filter to a sunset on instagram. It's also worth considering the major change with general ai, which is that where going to go from editing images to get them closer to perfect, to simply creating images of whatever we want, or in some cases, manipulating images by using A I to create a realistic component that was never actually there.

That's very different, at least with photoshop or ad filters. You have to start with something with A I. You can simply start with a prompt. How we think about the boundary between these technologies is important because it's moving quickly and not everyone can seem to agree on where IT stops being one thing and started to being another.

I think if we're talking about this specific conversation, there's definitely a line if we are saying photoshop itself, that was a revolutionary technology, right? That was what was saying, kicked all of this off on a dish aspect fater manipulation exercise for way longer than nap. But that was incredibly difficult.

That wasn't just text to know how that was physical skills. You had to know how to cut tiny little film roles and be able to manipulate them in magnified kind of situations for as photoshop enabled. Due to do that without having to have all the expensive tools, IT was still expensive software, but it's gradually become more accessible. I completely agree with you. There's a difference between almost manipulating something that had substance to begin with and just creating something that is a complete false reality IT had never existed begin with. And there's something to be said about intent there, which is why I think accessibility is one of the more significant concerns because even if you wanted to photoshop an image or something I don't know, like a lion and a playground or something, again, the ferrous, you would usually do so for a giggle or you're really going na have that much effort to learn on necessary skills, pick up the tech again, free or paid, which is expensive, if you want to go down that route, go through all that effort for a joke, for something that you think it's going to be funny, where's now there isn't any effort required. So the idea that you're changing an image with the fact that these two things are similar isn't a bad argument, but IT does completely ignore the fact that the accessibility and the scale of these things is the issue at hand, not what's actually happening.

One major element of adaptation is just that everyone starts to do that, right? So IT becomes Normalized. But one of the results of the instagram era, in particular, is that there are now arguing multiple generations of Young women and increasingly Young men who have body image and other mental health issues, because what they see on social media, they see filtered photos, edited images, people's faces looking smoother and sharper than there's ever could everyone just constantly having a question with what they're seeing is in fact, genuine or manipulated? Is that a form of adaptation that eventually enough people will learn to distrust what they see online IT they stop trying to measure themselves against IT because their mental health issues that teens are now facing doesn't seem ignorable and IT feels like A I is only going to make IT worse.

Is such a wash answer to give, but they really existing context, right? Girls and boy's men and women, like with all the body image shoes that they're wanting to have themselves, perceived on social media to, they are enclosed audience as being in some way the ideal version of themselves. And I don't think that they are bad for that from being exposed to all these list of images over the last decade or so.

Evidence cy other of social media has happened, especially instagram, which is really, really bad for this. But IT is the same argument effectively as to what we're doing right now. It's how much of this are we willing to accept before IT becomes problematic because the more this landslides into a situation of how much of this is actually reality is we can't trust the image is in front of visit this point. So I think it's lined IT was almost symptomatic in a way.

Do you think that addiction has happened though, that people understand? We should not believe what we see in social media because that's part of the argument here that at some point people will stop believing what they see on the internet and that will be the adaptation. And then this won't be a problem.

I think it's definitely happens like it's firmly happened already. If any celebrity about photo, you can go through the comment, something like I know the court actions, you can go through all of that stuff and you will have people corona alizon. The background of the image is trying to find any kind of distortion to see whether theyve made their way slipping or their bomb bigger or whatever.

So people are already very, very heavily scrutinising the images at the same in front of them. But I don't think they understand quite on a big scale yet the level of changes that can be applied because at this point, we've come to accept the body imaging in a way or body editing in this context is just something that people do, that people feel bad about themselves. They might make their teeth Whiter, they might make their face smaller.

All of these kind of things. I don't think people are going to rush to the realization of. They're looking at a picture that the entire background might not be real or that again, like if someone is sharing a viral image of something that's meant to be i'm trying to think of something that's not going to be too controversial, but like it's an explosion in a bin or something that's going to stick local news.

They're not gonna look at that immediately, ingo. That's fate because why would you that has to be a narrated behind that. They understand the body side of that. So there is not attain of that definitely.

Do you think that, that patients just resignation? We understand that people are going to put filters on their faces and on their bonds, and we're just going resigned to IT and not believe anything in now maybe would have thick explosions and track cans.

I think so. Yeah, like the one of the most common cases that are keeps seeing through out the Solomon a lot is the cat is at the bag is already out. You can't do anything about IT and it's very defeated.

It's almost like people half understand the scale of the issue and they can't see a future. Whether gonna anything that happens about us just discounted any kind of reality in front of them from this point. So any picture that they see online from this point, this fake people are already at that point, definitely.

I don't think a lot of people are there, not in terms of especially older relatives. When we talk about people on facebook card like names or arts or stuff, daring, obviously fake images as if they were going to be real. Ed, there's the whole crap jesus means of early little Midjourney edits and things like that.

But there are enough people, I think, scrutinising the right things at this point. They're only come to the understanding that we can change our physical appearances because they understand why they would do so. They can't really understand why people would use these tools for nefarious purposes despite the fact that when you think about IT for second, it's quite easy and you wouldn't even have to be that evil about IT.

I saw a picture that a colleague criswell ship mate using the reimagine tour on the pixel nine. And I believe IT was like a roach added to a take away or something. I take out, sorry, british and and immediately I was like, that's going to probably cause some problems for small businesses, right? If I could just order some food local complaint and say, hey, look, you've added something to my food.

I want my money back. There are so many smaller level scans and bad intentions that could be fulfilled using these tools that people wouldn't have out the energy of the effort or even the scale if we're being realisti C2Carry out bef ore bec ause it' s not wor th the eff ort. But if I can do that in five seconds, it's quite tempting for some people. I imagine there's a slipper slope. The other adaptation .

that you and I have talked about a lot, that our audience talks about a lot to the industry is talked about about a lot is technical adaptation. So you make the edit and reimagine on your pixel that it's flag as being an A I edit. And when you send IT to someone, they know that they're not looking at reality but looking at.

So I added, there's a variety of systems that are attempting to do this, that the most important is called the content authenticity ity initiative. And by the standards organization, you've reported on this IT doesn't seem to be doing anything. We've talked about IT a lot. But now as you say, the cats out of the bag in the systems to verify whether images are really fake or or not ready, are they close to being ready as or something that would stop them from from shipping today alongside these imageries?

The way that i've had to explained is that the system itself is absolutely fine. The way that it's supposed to be rolling out is it's fine IT IT makes complete sense. The problem is that you need everyone on board for IT to work, which is just completely unrealistic. You would need to get people that have completely different ideologies in terms of pro or against generative, our technology, to get on board with saying we're going to make the a robust identification system.

And unless you have all of the camera makers, all of the editing software makers, all of the online platforms, not just the social media ones, but like literally everywhere where you would see an image on board with this one system, I don't think it's going to make a meaningful difference. And IT doesn't really solve the issue that they do have at the minute, which is how to provide that information that an image is A I generated or just edited using A I. In a conscious, meaningful way without giving people a wall of text which no one is going to read, right.

I'm not going to sit and read a paragraph of what is going into a picture of my friends holiday snap or something. Meta tried this when they did there, made with A I levels, and IT went pretty badly because they provided no context and photographers who d used a couple of what sounded like pretty basic tools in adobe photoshop that use a very standard version of generate very something like background removal or object select. This system was effect vely flagged that and saying, hey, you ve used geni.

Therefore, we're going to tag your entire image just being made with this stuff. And IT gives the wrong intentions, because people see that word A I and they immediately think, okay, fake. This entire thing is fake, so there's no new ones in IT.

That's a much more complicated problem to solve alongside the existing issue of already trying to get thousands of organza companies on board with this one system adopting IT. So it's not gonna be a bullet proof solution. They know that as well that the content authenticity initiative were fully transparent with the father.

They said this might help, but it's not going to solve the problem. So we're kind a bit of a mess at the minute. There's a bit of bind. There should have been a lot of things that would put into place before I got to this point. And now that we're playing catch up, we're too slow and there's not a meaningful way to speed up this process at this moment.

I just want to stay on the idea that we couldn't label these images. And I agree with labelling the images is tRicky because you have to decide when something is sufficiently A I edited to append the label, and that is totally objective. But I said we can agree on that.

Let's just set that decide where is this data store? If I just take a screen shot, does that stick with the image? IT seems like we could solve all these hard problems, and we still left with the label that may be a bad actor. Could just get rid of .

it's stored as far as where on their own independent database are trying to set up the process where you can independently check images. That's one avenue. The other venue was supposed to be that you would access that information through the online platform as we've already used damage that may or may not be fake, right, in terms of actually manipulating that information is already improved that you can do.

So there are safeguards in place. I've heard that apparently if you screen shot IT, we using certain desktop software that metadata can still Carry over. There are still systems that can recognize that what you are screen shooting using that software Carries the meta date or more Carry some other over.

But then if I were to take, I don't know, my phone out of my pocket and take a picture of something on my desk computer, yes, the quality is going to absolutely awful by that point, but none of that mediator is going to be present. And I still have a copy of that manipulation. So at that point, all of the year, the data is district town.

There's nothing you can do like physical water Marks as well. The'd explored that we can remove those Samson's own tool removes that, if I remember greatly IT water mox images that IT the text to image tall. They put on yeah something devices.

You can just use the object errest tall and remove the water mark directly in the out that made IT. So at the minute, there isn't anything more robust. It's there's a constant kind of our company that we go down where people are finding ways around IT.

Adaptation is one thing. We will adapt. Things change all the time for Better, worse, and we adopt the new Normal.

But will that adaptation be good for everyone or even neutral? That's a much more chAllenging question. We did adapt to a world of photoshop and later instagram filters in face tune and so on. But the negative side effects are still playing out in deeply affecting the mental health of people everywhere.

So should we now adapt to a world where every single teenage girl has to worry about her classmates making deep fake news of her, or a world in which you can't reasonably believe that any photo of a politician you see on social media hasn't been tampered with? A I, as a society, we decide all the time that some problems are too complex, are difficult or too politically inconvenient to solve. So we live with IT. And if something doesn't change, the effects of widespread A, I imagine, might just become one of .

those problems we need.

Second, on a quick break will be right back.

Support for this episode comes from A T, N T. What does I feel like to get the new iphone sixteen pro with A N T next up anytime? It's like when you first pick up those songs and you're now the one running the girl, it's indescribable, like something you've never felt before.

All the mouth watering anticipation of new possibilities, whether that making a perfect cheers burger or treating your family to a girls' bake potato, which you know will forever change the way they look at potato es for the A N T next to any time. You can feel this way again and again. Learn how to get the new iphone sixteen pro with apple intelligence on A T N T and the latest iphone every year with A N T next up any time A T N T connecting changes everything apple intelligence coming fall twenty twenty four with theory and device language set to U.

S. english. Some features and languages will be coming over the next year. Zero dollar offer may not be available on future iphones next up, any time feature maybe this continued at any time, subject to change additional fees, terms and restrictions apply. See A T T docs sash iphone for details.

This message is a paid partnership with apple pay. When you've got a gift list to finish, the last thing you want to do is take out your wall at a million times. Instead pay the apple way with apple pay.

You can pay with the phone you're already holding just double like, smile like face I, D, tap and your done, the people in line behind you will. Thank you. Apple pay is a service provided by apple payment services L, L, C, A subset area of apple ink, any card used and apple pay is offered by the card issue R.

Support for the show comes from alex partners. You already know artificial will be transformative. Beyond that, IT might be a little bit of a mystery as A I opens the tech industry.

Alex partners is dedicated to making sure your business knows what really matters when IT comes to artificial intelligence because disruption brings not only chAllenges but opportunities in these pivotal moments change. Alex partners is the consulting firm chief executives can rely on with clarity, direction and most importantly, implementation. Alex partners provides a steady hand for your business needs when decisive leadership is vital.

Alex partner spoke with nearly three hundred and fifty tech executives from across north amErica and europe to dig deeper and to how tech companies are responding to these changing headwinds. You can see the results and learn how you can turn digital disruption into growth by reading alex partner's latest technology industry insights available at W W W dot alex partners dot com slash box that's W W W dot A L I X partners dot flash V O X. In the face of disruption, alex partners are who businesses trust to get to the point and to get things done when IT really matters.

We're back with verge reporter, just whether I to talk about the last big argument in the A I is just like photoshop debate that legislation can solve this problem. This brings me too, I think the biggest argument, right? These are big problems.

You need a lot of people to degree on a solution to actually solve. They all have to implement the same standard to the same technology. They all have to draw the same boundaries. The way that you typically solve that problem is by passing a law, right? You don't count on the USB implementation forum to solve every problem every time, right?

Like you a standards organza can they go so far? But if you want everybody to agree to do something, usually need the state to say, okay, this is how we going to do IT. I hear this a lot, right?

The problems with bad and the government will solve some these problems. The european governance will set some standards that people can copy, which is basically what's happening with a lot of other things right now. Is that possible that the government or a government could step in and say, okay, here's the law.

No deep fakes of real people. This is the media ata standard we're going to use. And you have to display these labels with this language everywhere we go, the way that we display, I don't know. Nutrition label s on food.

You got two manage issues without at the moment, one is speed. So even though like european is pretty far ahead of this at the minute, we've already enacted laws which have been heavily scrutinised because of the second issue, which i'll get to in the moment. But if we are talking about things like the U.

S. Legal procedures, we could be waiting years for something to actually come into play that will take effect and rain in the bad actors that are using these apps for bad purposes because they're not inherently bad tools. You could be using them for something when sal and absolutely innocent, but you have to separate the bad causes from the good causes. And when you talk about things like deep fakes, you get onto the second dish, which is new.

What do you consider a deep fake at this point? If I take a picture of myself on a google pix and I use their facebook in tor, I put my face onto my friend's body and I put that on facebook, would that be counted as a deep fake for a purpose if they wanted to take me to court, or any kind of legal fallout? At that point? IT becomes so granular of an argument that is really difficult to put on paper what should be restricted because at that point you're placing effectively limitations on creativity, which is difficult to do.

And you've also got to worry about things like free speech around there and all the existing what laws that you have to unwind. So it's not an easy or fast process. And a couple of years, an hour time frame is a millennia in the development time that we've seen is happening in the air landscape.

So we have no idea what could be happening with generated area in two years. At this point. Again, two years ago, we were seeing pretty shortly mashup being done together by these things.

They weren't believable. They were interesting. They were like pretty fun to play around with, but they weren't necessarily convincing. And now work at the point where you could spit something out on grog on on a social media platform itself and yeah, mislead thousands of people, millions of people if you wanted to. That's a completely separate thing that happened.

That's how I grog for sec. Greg, is the A I system that is on x, which is on by you on mask X, A I, which is on C I outfit, is behind rock. They distributed on x.

You can just log into x. You can tell rock to do all kinds of things, including generate photos of elon mask. At some point, IT feels like elon musk and mark jara and send arpa try and such.

No, I will will see enough deep fix of themselves. Sam altman, we'll see some sort of horrible deep fake of himself and did enough is enough. Is that where we have to get to the actual previs of the technology, feel bad effects before they they real their own creations in?

Maybe, but that would to be considerably, as there are significantly worse, have things happening then whatever elon mask could be friends to be doing, right? But if if we think about how a lot of these applications are already being used, there are a lot of celebrities, particularly female celebrities, that are being deep fake, particularly in sexually graphic ways. And that was a big deal.

I can't how long ago, this is my memory loves me. But there was a big hacking incident where people actually broke into celebrities. Hard drives install new self, is that they taken of themselves.

And in that circumstance, there was a lot of virtual spute their direction at these celebrities for taking those pictures in the first place. Now that can happen completely unconcerned without them having to do so. People can just make them look new and put them into very compromising positions online.

So IT would be pretty, I think, grim of all these people to be ignoring the fact that is already happening. I'm pretty sure that has has already happened to them as well. I haven't looked because I you complete to generate, but I think there is some arguments to be made.

The illegal process is probably going to for this that it's going to cause enough problems for enough people on a smaller scale. Maybe like to take out thing that I described, that it's just going to increase the use of small scale scams and fraud cat fishing on like platforms that people are going to make enough of rockers to go. We sick of this and we sick of not trusting everything that we're seeing in front of us anymore. And you need to do something about IT because everyone is just too easily manipulated at this point.

The reason I asked why taxi is, is sometimes I think narcisse is a Better regulator than than anything and not wanting the bad thing to happen to you that you control might be faster than these other processes. And i'll bring all the way background to to photoshop fake new images of celebrities have been around since the dawn of computer, since the donor photo shot.

This is one of the the first things men the internet have done with the tools have given. And we didn't stop them, right? No one ever said you shouldn't do this.

This is illegal. When I see deep fakes, I say, well, this is bad because it's more accessible, because this is more damaging. You can do IT to more people, high schools doing to the classmates, which I think is devastating in a real crisis.

But when that happened to the celebrities last time, we didn't run around showing those websites of existence. They continue to happen and continues to happen to the day. Are we back at its just the scale that makes you different. And so this time we should actually do something about IT or am I still bedding on a bunch of artist, still stop IT before IT happens them again.

The noises thing could absolutely work in this context. I think the thing that they have to fight against is the scale thing inherently, though good, even if something deserving happens to them.

They the same boat that point that everyone else is in, which is that if you wanted to photo shot one of these images before, you could maybe if you were very skilled photo editor, you could maybe do so in twenty thirty minutes, and you would have to be very skilled to do so. If not, I knew we're just like some regular j wanting to do something again, militias, or to get money off of these weird dog web point sites or something. You would be able to do that on a much smaller scale and much more contained way.

So if you went through photo filter systems to find these images and get them removed from the internet, but there was a much easier process. Now there aren't necessarily tools from my matter or something that are training these out. Stuff definitely slips through their god rails.

But there are generated vo systems out there that have no such god roles like theyve just been built using all the existing technology that's already available. And they can spend out thousands of these images that they want to, if you want to be looking for, again, new, deep fakes of celebrities or got to bids, really illegal stuff. We're talking like children involved.

No, this kind of stuff. There are violent and gorges and all sorts of things. The sheer scale of IT makes incredibly difficult to remove, to track.

And then because everything else online at this point, no one is using their real identities, how do you find these people to prosecute them? And how can you expect a platform to reasonably police that if they've proven in some kind of way in count that they have actually used all of the tools at their current disposal, which are an adequate and we know they are an adequate to address this situation. So it's kind of almost like bridges going round in circles at this point. The problem is getting progressively Better. And the tools that we have to address IT .

or not coming through this conversation IT feels like one answer is we should just not let these tools exist. We kind of don't know how to deal the scale of the problem that they will introduce. It's already happening.

Maybe we can stop IT, maybe it's start too late. But we could just say, okay, this was a weird run. We're just going to shut this down. We're not going allow these imagination.

There's a flip side to that, which is they have to provide some value, right? Is that baLance ed real to you? right? Photoshop and imaginations in general created a lot of problems, but the value they created for so many people was real. And I think we unbaLance have said, okay, well, the arms are the harms values real and I tell you, end up with, it's just like photoshop. Are we in the same kind of dynamic with general AI imaging?

I can definitely see the prose. I'm not preparing the idea of generating eye that I think I should all be burn to the ground and forever. And I don't think that even possible anyway, just because someone will rebuild these systems, you're going to get them cropping up, right?

If photoshop disappeared tomorrow, there's a million other photo editing tools that can replace IT if you try hard enough. The bigger issue is that even the good is doing, which is definitely helping efficiency. And there are applications for that are beneficial.

I think the magic, a raised tool that was on all there is on google pixel phones, is still a fantastic thing. There are all things in the background of photos you don't want to have there, and getting rid of that isn't something that you should feel guilty about or is necessarily going to cause problems in the vast majority of situations. So there are definitely use cases where they're useful and especially for like big industries and stuff.

I personally don't like the argument works like IT, boost fiction, ency for creatives because a lot of created, I will look at these things and go, but it's taking the creativity out of what I do is doing IT for me. But if the end of the day, all you want is a marketing picture to slap on an ad and then maybe changed a couple of backgrounds or something, yeah, this is helpful. I can do that lot quicker than a human can.

absolutely. So there's financial benefits to completely creatively. I think that's where we're going to have some issues. And then that leads onto the free speech argument and everything else and IT all landslides from there. So I have trouble at the minute trying to find if there is an adequate baLance, because every intention i've seen for IT that affected me as a person has just made me want to get right of IT.

Every time I on pin and trust, I want to find A, I got a picture of haircut, examples or something, drawing references, and I look hard enough, and i'm like, a god, say, this is generous. Ri, and it's all unrealistic and have only just noticed after a looking in IT for three seconds. There's not even effective way to filter to that stuff out of the minute because we can't identify these images. So for me, it's making my online experience definitely worse. I know that the same for a lot of people that do you just like to enjoy the internet most times?

Could you do the flip? I think about the solar time. It's very hard to have the AI generated image and prove that it's fake or show what's been to have people read the wall of text explaining the manual tions have been done.

Would IT be easier to label the real photos and say these ones are real and they haven't been touched in the second you touched them. The flag that says the real is way. I think if IT .

would be a an easier processing warm, that people were less likely to tamper with the content authenticity to initiate already does this with the exact same system is using to try and identify, generate very images, right? And if we're talking about press images, that is really useful. So there are certain SONY cameras and A, I believe, a like one as well, that will automatically record that at the moon you take a picture.

And if you were then to upload up something like getty or there's some big media agencies as well, like I believe the new york times has this, they were automatically registered flags and be like, cool, we've noticed you haven't tampered with this. We're going to publish this is a documentations image to prove that something has happened in a husband tempted. The difficulty then comes that there aren't really online platforms doing that.

Everyone's focus on trying to identify which which are fake and know which ones are real. And then when you get the online platform situation, you get the general public involved. And the general public is where people are going to start messing with things because they have different goals.

They might have some strange biases and want to manipulate, or muslim people intentionally. So does, I think, an argument to be made here that the preservation of the relationship between photography and journalism is going to be really important going forward, because that is a much smaller technological bridge to keep moderated. If you are having a an establish relationship between people that are documenting these things with photos and then sending them to media agencies to be reported as news, that is a much smaller space to be policing whether something is real. But yeah, as soon you get any kind of like open sourcing involved, it's just going to go to mess.

right? So we've gone through many of the big arguments in this debate. We have talked through the new answer, and we have discuss how we might try to solve some of these problems. And I say some of things we've talked about really do echo what people have been saying since photoshop and other digital image editing tools have existed, especially as they have evolved into the kinds of filters we see on social media.

But it's the scale of A I images and image generation tools that really takes us to a whole new level in the power, in the speed of the technology, is far outpacing any of the safeguards that might conceivably stop IT. Like I said at the beginning, I think the reason people say it's just like photoshop is to diminish the very real differences in very real problems posed by A I image dating to avoid thinking about them or to make IT seen my technologies forever unstoppable, or to assume someone else will solve the problem. I don't think that's tannic, which is why we spend so much time talking about. Just thank you for coming undercoat.

Thanks for having me had a great time talking about this, and i'm very curious to see what it's all gonna go.

I don't think just whether I I for taking time to join decoder. And thank you for listening. I hope you enjoy y IT.

If you you have thoughts about the substance, what you like to hear more of, you can email us a decoder of the verge coome we really do read the mails where you can hate me directly on threads on that reckless for the most have a tik up check IT out its at coder decoder please share with your friends and subscribe over your pocket really like the show he is with that five star of you. Decoder production, the verge and part of the boxing to protest network producers are k. cox.

And next that our editor is called right. Or supervising producers, lain James. The decoder music is very big, later, sooner will see an extent.

Support for this podcast .

comes from strike. Strike is a payment and billing platform supporting millions of businesses around the world, including companies like uber, bmw and door dash. Stripe has helped countless startups and establish companies like reach their road targets, make progress on their missions and reach more customers globally. The platform offers a sweet specialized features and tools to fash track growth like strike billing, which makes IT easy to handle subscription based charges, invoicing and all occurring revenue management needs. You can learn how stripe helps companies of all sizes to make progress at stripe dot com, that stripe dot com to learn more stripe make progress.

Support for the show comes from a new york magazines. The strategist, the strategist helps people who want to shop the internet smartly. Its editors are reporters, testers and obsessives. You can think of them as your shop holic friends who Carry equally about function, value, innovation and good taste.

And their new feature that gifts out takes the best of their reporting and recommendations and uses that to surface gives for the most hard to shop for people on your list. All you have to do is typing a description of that person like you are a parent who's where they don't want anything, or your brother in love g je, or with a sweet tooth. And the gift scout was scan through all of the products they run about and come up with some relevant suggestions.

The more specific you make your request, the Better, even down to the age range. Every single product you'll receive is something they written about, so you can be confident that your gift has a strategist live approval. Visit the strategist dcom slash gifts out to try this out yourself.