The rapid growth of artificial intelligence (AI) technology has reignited humanity's long-held fundamental fears.

Questions like, "What if AI takes all human jobs?" and "Is a world where AI dominates humanity like in Terminator coming?" blend science fiction imagination with real-world anxiety.

However, when we look closely at what is happening at the forefront of AI, we encounter a bizarre and chilling twist. The real threat leading AI to destruction is not humans. AI is being consumed by 'another AI.' This technological paradox represents the most significant and essential crisis facing the current AI ecosystem. We will explore why AI is devouring AI and what this means for our future.

The Depletion of Data and the Curse of 'Inbreeding' (Model Collapse)

Today's large language models (LLMs) and generative AIs have grown by consuming the data (text, images, videos, etc.) that humanity has accumulated over thousands of years on the internet. Novels written by humans, paintings created by humans, and conversations exchanged by humans have all served as excellent nourishment for AI.

But a problem has arisen. The internet is now being filled more quickly with content that resembles AI-generated waste than with content created by humans. As high-quality 'human data' becomes depleted, the latest AIs have started to retrain on the text and images produced by previous generations of AI. Scientists refer to this as 'model collapse' or technological 'inbreeding.'

When AI repeatedly learns from the outputs of other AIs, subtle distortions and errors in the data are amplified. Ultimately, as generations pass, AI becomes dumber, producing bizarre results until it collapses on itself.

It is not humans destroying AI. It is the flood of fake data generated by AI that is contaminating and consuming the next generation of AI.

The Cannibalism and Monopoly of Large Platform AIs

From a market ecosystem perspective, AI is fiercely consuming AI. Just a short while ago, specialized AIs for translation, copywriting, and simple coding assistance were gaining attention.

However, every time big tech companies like OpenAI, Google, Microsoft, and Meta upgrade their 'general AI (AGI)' models with massive capital, numerous small AI companies disappear without a trace overnight.

As large language models become slightly smarter, small AI chatbots or tools that barely provided services on tiptoes lose their reason for existence and are absorbed. The speed at which lower-tier AIs are devoured by larger AIs is much faster than the speed at which human workers are replaced by AI. Ultimately, only a few super-large AI powers will remain in the ecosystem, waging war to consume each other.

The 'AI Judge' that Monitors and Eliminates Itself

Currently, the role of verifying AI ethics, filtering out fake news, and blocking harmful content is also moving away from human hands.

Due to the overwhelming amount of data being generated, humans cannot even begin to review it. Ultimately, a structure has been established where 'review AIs' monitor and control 'generative AIs.'

An interesting phenomenon occurs here. Numerous AI algorithms and models that do not meet the standards of review AIs are deemed unqualified and buried behind the scenes without human knowledge. The fate of another AI is determined by guidelines and logic set by AI. Humans only receive the final report from the review AI, unaware of which algorithm sentenced which algorithm to 'death.'

Where is the Place for Humans?

The proposition that "AI is not human, but it is being consumed by AI" paradoxically offers us both a significant comfort and a powerful warning.

The defense line of human uniqueness: As AI undergoes the process of copying and replicating each other and ultimately collapsing, the only key to breaking this vicious cycle is 'original human data.' The randomness, genuine emotions, and real experiences gained through direct engagement that AI cannot imitate will paradoxically become the most expensive and valuable resources in the age of AI. A warning against digital desertification: If we allow this trend to continue and let the internet world be filled entirely with AI imitations, humanity's intelligence may also regress.

Ultimately, the future is not a world where AI attacks humans like in Terminator. It is a struggle for humans to maintain their subjectivity and pure creativity in a swamp of AI that becomes monstrous through self-replication and mutual consumption.

In the midst of the great whirlpool of predatory AIs, we must focus even more on the value of 'the most human things.'