Who Controls AI? A Three-Polar World, Sovereign AI and the Robot Future

This episode of Teke Tek Bilim turns artificial intelligence from a technical curiosity into a direct question of power: in whose hands does this technology sit, under whose rules, and trained on whose data? The two guests are Prof. Altan Cakir, one of the first names that comes to mind for AI in Türkiye, and Ozan Sihay representing the technology side. The light-hearted opening soon gives way to a firm analytical frame.

The backbone of the conversation is a three-polar reading of the world. On one side the United States with a deeply interdependent ecosystem; on another China, carrying research and innovation onto a different track while rapidly lifting its universities to the top tier; and on the third side the European Union, whose path runs through regulation. Cakir highlights the contrast between the simplicity of American rules and Europe's reflex to steer by writing rules.

One of the most debated claims is that everyone will soon own a robot. Chatbots embedded into electric cars and plans for a dedicated robot factory are cited as evidence. The guests agree the timetable is optimistic; a helper in every home within a couple of years sounds early to them, yet the direction itself looks realistic. Childcare and shopping handled by machines are no longer science fiction but planning items.

This robot debate connects to the physical AI concept featured in major investment funds' five-year plans. The idea is that intelligence materializes not only on screens but inside electromechanical bodies acting in the real world. Cars and humanoid helpers are the first examples that come to mind, but the concept is broader: any system touching the physical world through sensors, motors and decision mechanisms belongs to this field.

A everyday example also shows the technology's limits. Driver-monitoring systems sometimes warn a driver who has just sat down about fatigue. A claimed 89 percent face-recognition success rate is discussed in this context; laboratory accuracy and messy road conditions are not the same thing. Such edge-deployed systems, the guests argue, still sit somewhere between simple automation and genuine understanding.

One of the liveliest segments concerns generated content becoming indistinguishable. The example of an ordinary user having a poem written, turned into a song, and genuinely enjoying the result shows the matter has left expert labs and entered homes. The claimed 90 percent accuracy of university originality-detection systems comes up too; their logic of flagging overly smooth text as machine-made rests on the idea that natural human roughness is itself a signature.

The conceptual center of the episode is sovereign AI. The definition offered here is not isolationism, no foreign products or banned clouds, but keeping control of data and models with the institution in charge. A vision is drawn where the public sector, private companies and universities each become their own node, sustaining a structure that serves the country's overall welfare. Where data resides, who governs it, and who bears the risk of loss or leakage form the core questions.

The observation from the creative-production side is that an era is ending. Image and video editing tools used daily for years are now opened rarely; describing the imagined work in words reportedly delivers a satisfactory result in most cases. As the craft of writing descriptions keeps evolving, the guests argue that in a world where everyone can produce everything, the difference returns to human imagination; the era of learning tools ends, the era of knowing what to imagine begins.

On the software side the picture is two-faced. Having code written for you has never been easier, yet outputs still contain errors, so expert review remains mandatory. The warning is stated plainly: companies should not upload trade secrets into public systems, because once the request reaches an outside server, control is gone. The good news is that open-source models have nearly caught up with their paid counterparts at this task.

Two big themes merge in the closing. First, the synthetic-data boom: the loop of feeding model-generated data into new models is expected to hit saturation, after which human-made original data will gain value again. Second, aging populations and the pressure of endless growth; questioning the compulsion to grow forever, the episode ends on the idea that AI could become a lever carrying humanity from a race of quantity into a field of quality.

sovereign ai · us-china-eu · robots · generative content · synthetic data · nodesdaily