Man vs. "Machine": What Impact Do Neural Networks Have Today?

India is often called the birthplace of the best IT specialists in the world, but now it faces a formidable new competitor. Neural networks are writing code, automating document processing, diagnosing illnesses, creating advertising content, and assisting in education.

The paradox is this:

the smarter the "machines" become, the more acutely humans feel their own vulnerability. Voices are faked in seconds, faces are transferred onto different bodies, and scammers hack accounts not with passwords but through digital copies. Algorithms know more about us than we do: they shape our desires and already influence our choices.

Unnatural facial expressions, blurred images, and robotic voices are usually signs that a person on the screen isn’t real. But what if today artificial intelligence is capable of more than we think?

An experiment was conducted: participants had to identify which of four videos featured a real person and which featured an AI. Out of 35 respondents, only two answered correctly in all cases. The two women in the videos were popular AI avatars from social media, where ordinary people earn money. But why do human eyes sometimes seem “blinded” by such images?

Svetlana Pashkevich, head of the Brain Center at the Institute of Physiology of the National Academy of Sciences of Belarus:

"Thanks to plastic surgery, our faces have become more similar to artificial intelligence faces. It’s becoming increasingly difficult to distinguish a living face from a fabricated one. Why do we perceive AI with some suspicion? First — it’s strange symmetry. Few algorithms can work with subtle variations. We perceive reflected light, which neural networks should reproduce, but they can’t do this as chaotically as our brains do."

"There are different types of faces: Mongoloid, Negroid, European. It’s very difficult to adapt to all of them. During our research, we found that to recognize a single cell without markers or glowing dots, you need over 20,000 photographs in various poses and angles. For AI to generate a face, it needs such a large number of images of one person, not many. That’s why faces are often a mix of different people, lacking individual uniqueness," Svetlana explained.

In such conditions, scientists believe that humans need more time—at least 80 seconds—to process information and make correct decisions. But digital intelligence can also deceive.

Artificial intelligence has long ceased to be solely a tool for good. The UN warns of new dangers for children. Besides deepfakes and cyberbullying, malicious actors can use neural networks to analyze children’s behavior, interests, and emotional states online, tailoring their manipulation strategies. In Belarus, one-third of offenses occur on the internet.

Alexander Ringeich, Deputy Head of the Main Directorate of the Ministry of Internal Affairs of Belarus:

"Malicious actors most often gain access to messaging app accounts. Using chat histories, they create fake photos, videos, and possibly voice messages voiced by the victim. Most often, this involves requesting bank card details or transferring money to accounts controlled by the attacker."

Fake content created with AI can lead to fines or even criminal charges if it misleads, disrupts public order, or threatens security. AI is mentioned in Decree No. 8 "On the Development of the Digital Economy," and a model law on AI technology has been adopted across the Commonwealth countries. However, there is currently no comprehensive regulation of neural networks in Belarus.

Vladimir Borovenko, Deputy Chairman of the Standing Committee of the House of Representatives of the National Assembly of Belarus:

"A working group has been formed, including representatives from all interested agencies. Work is actively underway to develop legislation. We mainly use foreign systems, but in sectors requiring high-risk AI applications, technological sovereignty is key. For systems that ensure national or individual safety, we need to rely on our own developments. Mechanisms used in some countries will be considered when drafting the AI law."

Artificial intelligence automates document processing, helps doctors diagnose illnesses, assists developers in coding, creates advertising content, and even produces music, displacing pop stars from top charts. But what limits might the digital mind cross? That remains a question for the future.