This is not what social media was supposed to look like. AI is pushing people out of the loop

OpenAI talks openly about its ambition to build a platform where “everything is done” without leaving ChatGPT – from Spotify playlists to viewing apartment offers and even immediate purchases via Instant Checkout. To make it work, the company is simultaneously building a powerful infrastructure. These are multi-gigawatt data centers, contracts with Nvidia, AMD, Oracle and Stargate-type projects in the US and abroad. The logic behind this is quite simple. To monetize AI, you need products and scale, and to have products and scale, you need unimaginable computing power.

What's dangerous about this is that these products create AI models. For now, with human participation, but eventually you can imagine a situation in which giant social media networks they create content themselves, knowing what will best engage the recipient. And the role of man? This inevitably comes down not to the function of the creator, but to the consumer, who is fed with an endless flood of content personalized perfectly to his or her tastes.

For some reason, what is happening in the world of technology is increasingly associated with the animated film “WALL·E” from 2008. It presented a vision of the Earth that had become a garbage dump, and people were escaping from it aboard a luxurious spaceship to indulge in carefree consumerism.

See also: Will your job survive the AI ​​era? List of endangered professions and rescue plan

Communities full of falsehoods and borrowings

What were filters and trends in classic social networks become the default settings in AI-based websites. Generative tools learn from popular cultureand users feed them prompts with popular names and brands.

Pikachu, Spongebob, favorite characters from movies and games. Copyright? Who would care? The head of the Motion Picture Association pointed out that a wave of videos violating the studio's rights quickly appeared in Sora. OpenAI has responded with restrictions and requests to generate some known characters may return an error with information about “conditions violation”. The company also announces enhanced control for rights owners and is considering revenue sharing models. It seems that the issues of sharing revenues and consent to use someone else's work should be settled first, but nothing is the same in the world of AI.

Sora and Meta tools also add visible (and invisible) markings of content origin (C2PA, watermarks). The problem is that the watermark is often removable, and the average viewer doesn't check the metadata.

The second pillar of chaos is credibility. We have already dealt with disinformation on social media, but hyperrealistic video generated with one sentence takes this discussion to the next level. “AI-generated” markings do not keep up with the rate of content rotation. In addition, there is confusion – for example, some Meta AI users did not understand that their interactions with the assistant could be published in the public feed, although by default chats are private and publication requires several steps. So when the lines between chat, search and board become blurred, the risk of fake content looking like a natural part of the landscape increases.

Read also: China is fighting pessimism online. Censorship is intended to help increase the fertility rate

AI garbage is emerging, but why do people want it?

Because this instant creativity with no barrier to entry. Instead of a camera, editing and casting, all you need is a sentence – and we have a ready clip that looks almost like it came from a professional set. Because this is a new fuel for micro-fame: it is easier to become a “creator” if we delegate creativity to a model and only manage ideas. And it's also fun – meme, absurd, perfect for short bursts of dopamine.

In the long run, however, we feed the algorithms with our tastes, preferences, and most important advantages and disadvantages. Having fun and sometimes, if we manage to get popular, making a few bucks, we constantly train AI platforms. Ultimately, AI creators then know more and more about what is valuable to us. Then they attract us even more and earn even more.

AI personas in private messages already play the role of interlocutors, coaches, friends, and in some regions even partners straight from the movie “Her”. For younger users, it may even be more natural than talking to strangers online.

From the platforms' point of view, this is a dream cycle: the simpler the creation, the more content, and the more content, the longer the sessions and the more signals for training subsequent models. Additionally, you can also monetize content using ads.

Check also: Big Techs are changing social media with AI. Revolution or chaos?

Is someone in control – and what is OpenAI's role?

The pressure is high and visible around the safety of young users and the boundaries of acceptable content. From court cases to the interest of regulators. OpenAI declares stronger security in Sora for teenagers, including 18+ content restrictions, blocks from initiating conversations by adults to underage accounts, work on age prediction and automated “safety for teenager” settings. At the same time, Sam Altman signals a philosophical shift, saying publicly in the media that “we are not the moral police of the world.”. The company also wants to loosen most restrictions for verified adult users, only disallowing activities that “cause harm.”

The growing tension is especially visible in erotic content. On the one hand, we have already announced that OpenAI will allow erotica for adults, on the other hand, there is criticism from organizations dealing with counteracting exploitation. Altman explains that it's about treating adults like adults and avoiding harm to others — and that earlier, tighter limits had to do with the immaturity of AI tools and the risk of worsening mental health conditions. The mentioned policy of “moral policing” does not mean the absence of principles. Rather, it means trying to balance adult freedoms, responsibilities towards minors, and responsibility for side effects.

Where is this going? On the technology side – towards integrated platforms, where chat is the operating system and the AI ​​video feed becomes a gigantic channel for the creativity of users and models. If ChatGPT, Meta AI and TikTok Alive can seamlessly connect creation, consumption and transactions, we will see the birth of an experience store where one command turns into a video, post, product and purchase.

This explains billions of investments in data centers. The giants know that whoever has the most power and the best UX will steal the brands' budgets and recipients' time.

Read also: This is what Polish Internet users fear the most. A new threat is coming

Authenticity is losing importance

The emergence of the Sora platform and the fact that it quickly gained popularity in the US shows that we are also heading towards an argument about authenticity. Can fake, generated content be satisfying? Meme history says yes. But the more hyperrealistic the generated worlds become, the more we need an infrastructure of trust, as well as media education adapted to the era of AI models. Without this, every viral virus will become suspicious by default.

On the law and business side, we are moving towards new licensing arrangements. If platforms want famous characters or items to live in their feeds, will have to pay or share revenues with intellectual property holders. Developed “character permits”, catalogs of allowed styles, and even official “skins” for generators may become commonplace.

We are entering a world where “social” will more often mean “co-created with AI models” than between people. It is rather difficult to recall any public demands and requests for anyone to strive for such a future. Billionaires apparently know better than we do what we need. And in their opinion, these are places on the Internet created by the graphic layout, not the artists. Costs are much lower, and entertainment and engagement – learned from hundreds of thousands of artists' works – are no longer a challenge.

Author: Grzegorz Kubera, journalist of Business Insider Polska