Muse Collectors Forum

Notifications

Clear all

On The One Hand

Group: Registered

Joined: 2023-07-24

New Member

About Me

The Spiegel cover story of the current issue deals with the unreality created with AI image generators such as Midjourney or Dall-E and offers a good overview of the debates of the past few months synthetically generated image "The Electrician" won the "creative open" category at the prestigious Sony World Photography Awards in April. This category has always been open to experimental approaches and post-photography, but the case attracted international attention, precisely because Jason M. Allen’s AI-generated image "Théâtre D’opéra Spatial" was still fresh in his memory, with which he won a prize in an art competition and thus fired up the still young debate about the relationship between art and image synthesis. This time, however, the Spiegel story is less about the nature of art in times of stylistic interpolation and more about the effects of a technology that can be used with just a few clicks can produce realistic images of practically anything, our perception of reality and the ability for social consensus. But this debate is anything but new, as Stalin is known to have used image manipulation techniques to create alternative realities, and at the latest with the democratization of these techniques, which were formerly reserved for retouching professionals, through software such as Photoshop, most people are actually aware that images per se cannot be trusted. The problem with image generators is not simply that they can depict alternative realities, people have been able to do that for a very long time, but that they democratize these techniques and make them accessible to everyone, including bad actors. And so, of course, right-wing politicians were the first to use AI image synthesis for their election campaigns - also in this country through the AFD. On the one hand, I personally think that the problem is overestimated. We’ve long lived, especially on social media, with the awareness that images are being faked, and "this is shooped, I can tell by the pixels" has long been a common phrase on the internet. The actual influence of disinformation is still controversial, studies find no real impact of Russian disinformation campaigns or social bots, other papers show how disinformation created an atmosphere of fear and insecurity in the wake of the Covid pandemic. The last point in particular is a cause for concern in view of the possibilities of AI image synthesis, because photorealistic images from alternative realities are something different than drawn propaganda illustrations, Midjourney Porn Generator which are widely known from history. However, the Spiegel cover story does not deal with it not only with the new possibilities of AI-supported image manipulation, but also with what the philosopher Daniel Dennett called "Counterfeit People", i.e. mimetic AI systems that imitate real people. Behind this somewhat brittle term are systems that can produce non-consensual deepfake porn with the faces of real people, or imitate voices in a deceptively real way and thus open up a new playing field for scammers and criminals, or even the whole personality and way of speaking of a person in a chatbot imitate - in my opinion, this is the biggest problem of this new technology in our digitally enabled post-truth world. Because we have been living with fictitious, realistic image worlds since the invention of photography at the very latest, and even more so since the invention of the cinema. But not with fictional people who have a real impact on our psychology, since we automatically attribute human autonomy and cognition to them. A study recently found out that people who write texts in cooperation with AI systems can be influenced by them in their opinions, which at the same time makes our opinions susceptible to subtle attacks by prompt hacking, which we do not even notice in case of doubt .It is precisely this new susceptibility to hacking, not for technical systems, but for our psyche, our attitudes, our thinking, that makes me very cautious after the initial AI enthusiasm. When mimetic systems simulate human behavior in a deceptively real way, and we as humans cannot help but ascribe human-cognitive properties to these obviously synthetic systems and thus make them a real social actor in our thinking, even if they are actually only interpolations between data points in a huge database, these systems open up a new attack vector for manipulating our behavior. (And I haven’t even started with the possibilities of self-radicalization through finely tuned open source language models.) The article in Der Spiegel offers a successful introduction to all these dimensions of the new AI post-truth, in which the boundaries of reality and our perception become blurred can never be secure in the digital space.