Enlarge image

OpenAI's new AI Sora creates videos from short text input: The AI ​​community is absolutely thrilled

Photo: Drew Angerer / AFP

Two seemingly unstoppable technological complexes are currently moving towards each other, which alone would have the power to change our world. We can only roughly imagine what will happen when they merge. One is the social network TikTok, into which the EU has just opened an investigation. Above all, to find out how safe children and young people are there in terms of addiction potential, the “rabbit hole effect” or age verification. The other is a technology, of course from the area of ​​generative artificial intelligence, which goes by the unwieldy genre term “text to video”. ChatGPT creator OpenAI recently introduced Sora, an AI that can quickly generate more or less realistic-looking film sequences from short, thrown commands.

The question of why you have to look at both together and in terms of their potential impact starts with TikTok. The Chinese video app is one of the most important entertainment, information and news platforms for the generation up to thirty years old in the Western world. By radically focusing on a kind of algorithmic populism, TikTok has managed to show a lot of people exactly the content they want to see. Even if they didn't know it before. A term coined for this in the early days of internet culture was “serendipity,” which can be translated as “finding without looking.”

TikTok has incentivized a generation and a half of how to hyper-personalize entertainment and translate creativity into power. Ultimately, TikTok offers its audience endless and really good entertainment. But for content producers, TikTok offers reach, monetization and a relevance that is still drastically underestimated by most older people. What is particularly attractive about its reach is that TikTok has completely redefined how so-called virality arises in social networks, i.e. the explosive spread of individual content. This means that even people with very few followers can reach an audience of millions - with the right idea for a video.

Fireworks of manipulation

TikTok currently has arguably the best recommendation algorithm in existence. Unfortunately, it can be exploited, especially by people who work with truncated or outright fake information to achieve a particular form of TikTok sensationalization. TikTok sensationalization has many different cultural facets, and most of them are structurally positive. At its core, it's usually about the creative ability to produce an interesting to spectacular second production using wit, charm, sometimes intelligence and music snippets as well as a few technical gadgets.

more on the subject

  • Protection of minors: EU Commission opens proceedings against TikTok

  • Text-to-video generator Sora:OpenAI now also generates AI videos

  • AI luminary Mustafa Suleyman: “This will be the most productive decade in human history” An interview by Patrick Beuth

  • Regulation of online platforms: “The fact that the EU has agreed on this set of rules is a big step” An interview by Sophie Garbe

Politically, socially and news-wise, TikTok is by no means just a firework of manipulation, but is now also a firework of manipulation that is ideal for spreading ideologies. Too often, the radical staging on TikTok means that what sounds good and interesting is taken as true and correct. Admittedly, this isn't something TikTok invented or even exclusive to - but the speed, extreme virality potential, hyper-personalization, audience and state of the world amplify this effect.

In recent months, the question has arisen frequently as to why the AfD is so successful on TikTok. On the one hand, this is due to reactionary trends that are often subtly but ultimately quite openly conveyed there, such as anti-feminism, racism, anti-Semitism and, especially, social Darwinism in various forms. For example, a large sphere of interest with its own stars who reach millions is almost exclusively concerned with how to become as rich as possible as quickly as possible. This fits seamlessly into the omnipresent self-optimization on TikTok. And because social media like Instagram and TikTok are so perfect for showcasing wealth and luxury, both trends quickly give rise to the attitude that poor people are fundamentally to blame for their economic failure. In addition to the plan on how to supposedly get rich on your own.

The call for authoritarian leadership is also finding surprisingly strong resonance, and not just in right-wing circles. In my opinion, central to the success of the AfD among young people on TikTok is a replacement of the classic news and world-explaining media from the everyday experiences of many young people. For example, when absurd excessive bureaucracy, policy failures or problems with integration are experienced personally or are present in the media, but only cramped, abstract explanations and certainly no solutions are offered. Suddenly racist or anti-democratic populism appears as both an explanation and a solution. Spectacle and staging do not always mix well with complexity, shades of gray and the balancing of democratic values.

Calling TikTok a fake news slingshot is definitely reductive and doesn't take into account the many forces that are opposing the tide of anti-progress right there, with enlightenment, education, counter-analysis and really good entertainment. But TikTok is currently the most powerful, least corrective and most attention-grabbing medium of influence in minds under the age of 30.

A scary sensation

Which leads directly to Sora, which is intended to serve as a symbol for the entire technological complex of AI videos. The AI ​​community on social media has been going crazy since Sora was introduced. This technology is already available to a number of people who are particularly trustworthy from OpenAI's perspective, and these people continue to publish new videos, often including the instructions, the prompts, with which the AI ​​generated the videos. That's what OpenAI did on Sora's introduction page. You can currently use Sora to create video clips up to 60 seconds long, with two or three short sentences. The results open up a world almost as large and astonishing as the one that ChatGPT opened to the public. The following video gives an insight and it can be described as a scary sensation. The prompt “A litter of Golden Retriever puppies is playing in the snow. Their heads pop out of the snow” – leads to a video that, even for experts, can hardly be distinguished from a real video. Not at all at first glance.

You should not approach artificial intelligence with your current understanding of technology. Even classic, digital insights are often not enough to grasp the current potential of this technological complex. And even less to be able to meaningfully estimate future developments. AI must be thought of in terms of constant, sometimes exponential improvement. This is not trivial because our basic human understanding often incorrectly classifies what is easy for AI and what is difficult.

Sora must therefore be viewed as the starting point of a new world. In my opinion, all the small inadequacies and major oddities in the videos so far are negligible. We are at the beginning of a new media and therefore also political era, because with Sora and comparable technologies, the term “fake news” is no longer sufficient.

On the way to the perfect propaganda machine

Text-to-video on the level of Sora is the starting signal for the era of “fake reality,” even if the term seems contradictory. AI can create worlds whose image will no longer be distinguishable from the images with which we have previously learned to assess our reality. ChatGPT has mastered the creation of communication strategies, these skills should improve dramatically in the next iteration, and they will eventually be able to be linked to Sora. A prompt from the future: “Create a series of 10 videos that convince as many people as possible that Vladimir Putin is the only hope for Europe.” The scripts would take into account the latest scientific findings in persuasive communication and propaganda; they would be clever, entertaining and virality-oriented.


Sasha Lobo

The great crisis of trust

Publisher: Kiepenheuer & Witsch

Number of pages: 336

Publisher: Kiepenheuer & Witsch

Number of pages: 336

Buy for €25.00

Price inquiry time

February 21, 2024 6:07 p.m

No guarantee

Order from Amazon

Order from Thalia

Order from Yourbook

Product reviews are purely editorial and independent. We usually receive a commission from the retailer when you make a purchase using the so-called affiliate links above. More information about this here

And this is where TikTok connects seamlessly. Because even the best generative AI can only increase the probability of achieving a communicative goal. But because AI has to think in learning loops, the AI ​​would create thousands of versions that can easily be checked for reach, virality and audience impact via TikTok. This data flows into the further creation process, can be divided according to its effectiveness with target groups and improved with ever greater precision.

With the principle of TikTok plus the technology complex surrounding Sora, we are on the way to the perfect, fully automated propaganda machine of the 21st century. It can create fake realities, tailored to each individual person if in doubt. What has protected us so far is that the quality of Sora requires gigantic resources and expertise that only very few large companies have mastered - with which you can negotiate. But it won't always stay that way. We have barely gotten the problem of fake news under control so far and are already having to deal with a much more powerful fake reality.