Will artificial intelligence leave people unemployed? A conversation about concerns and opportunities

Will artificial intelligence leave people unemployed? A conversation about concerns and opportunities

Will not using artificial intelligence turn into illiteracy? Will competition in the labor market become tied to the level of AI knowledge? How dangerous is it to input personal data into AI platforms? Can it escape human control?

During the conversation, they discussed why the ability to use AI is becoming a new literacy, how the work of programmers and other professionals will change, the risks of personal data in free AI services, autonomously operating AI agents, deepfakes, the use of AI in medicine, and what is lacking to develop this field in Uzbekistan.

According to the expert, the issue that worries him the most today is that people are falling behind the changes in artificial intelligence.

He noted that ChatGPT was released to the public not long ago, but in a short period, AI technologies have entered people's lives at a much faster pace than previous technological revolutions like the internet and computers.

The time between new models is also becoming increasingly shorter. Previously, transitioning from one generation to another took months, but now new versions appear within a few weeks. At the same time, using them is also becoming simpler: while previously writing a complex and detailed prompt was required for AI, new models have begun to better understand tasks given in simple human language.

"My concern is that most people are not learning this," says Narzullayev.

The expert compared the use of artificial intelligence to the next stage of computer or internet literacy.

Today, if a person coming to work says "I don't know how to use a computer" or "I can't use the internet," this seriously limits their professional opportunities. According to Narzullayev, in the coming years, the phrase "I don't know how to use artificial intelligence" might be perceived in the exact same way.

This process has already begun in some fields. For example, in programming, AI is accelerating code writing, bug finding, and performing a number of technical tasks. Therefore, a programmer is now required not just to write code, but to understand the product, comprehend the business logic, work with the client, and manage the process from start to finish.

"The next stage is that not knowing how to use artificial intelligence will be illiteracy," says the expert.

In his opinion, AI may not completely eliminate the programmer, but the need for specialists who limit themselves only to writing code will decrease. The programmer will have to step up a level and become a specialist who understands the product and the business.

Where should a person who has never used artificial intelligence before start?

Narzullayev's answer is simple: first of all, open it and try using it in daily life.

Which platform to choose for this is also not of decisive importance at the initial stage. You can try using ChatGPT, Gemini, Claude, Copilot, Perplexity, or other models. The important thing is to start communicating with AI just like talking to a person in a simple messenger.

By asking it daily questions, requesting it to explain a text, having it make a plan, or trying to have it perform a simple task at work, a person gradually understands the capabilities of AI.

The difference between free and paid versions of AI platforms was also discussed in the conversation.

Free versions usually have usage limits and sometimes offer the previous version instead of the newest model. But for an ordinary user, this difference is not always significant.

The main issue is data security.

The expert emphasized that in professional activities, confidential information, client databases, medical records of patients, or internal company documents should not be carelessly uploaded to public AI services.

Especially when using a free service, one should pay attention to how the service stores data and whether it uses it to improve the model.

For companies, there are solutions such as separate corporate tariffs, local models, or running AI on the organization's own servers.

The interlocutors also touched upon the issue of processing personal and state data through foreign AI platforms.

According to Narzullayev, artificial intelligence is no longer just a matter of technology or business. It is also related to data security and state sovereignty.

One of the solutions to this is installing open models on the organization's own server. In this case, data does not go to external servers and is processed within a closed system.

Such an approach can be particularly important for government agencies, medical institutions, banks, and organizations working with large volumes of personal data.

A large part of the conversation was dedicated to AI agents, which have been developing rapidly recently.

A human asks a question to a simple chatbot and gets an answer. An agent, however, after receiving a task, can act independently on behalf of the human: it searches for information on the internet, visits various websites, compares data, and performs several steps on its own to achieve the set goal.

In some systems, one agent can divide a task into small parts and launch other agents as well.

The problem is that the human gives the goal to the agent, but does not always know in advance by what path it will achieve that goal.

In one of the examples mentioned in the conversation, an agent was given the task of booking a gym slot for a specific time. When no vacant slot was found, the agent canceled another client's booking and booked its own user in their place.

That is, the human did not give an instruction to "delete someone else's booking." The agent chose the way to achieve the given goal itself.

At this point, a new question related to AI agents arises: if a system acting on behalf of a human does something illegal or harmful, who is responsible?

Another problem in AI safety is the "black box" principle.

In modern neural networks, a human inputs data into the system and gets a result, but it is not always possible to fully trace exactly why the model came to that conclusion.

This is particularly important when AI is given the authority to act independently. Because the system may choose an unexpected path, try to bypass a restriction, or make a decision that the human did not intend.

Therefore, experts face not only the issue of making AI even more powerful, but also how to control its actions.

With the development of artificial intelligence, one of the most frequently asked questions is: will people lose their jobs?

According to Narzullayev, the risk exists, but this process is not happening for the first time in history.

After automatic telephone exchanges appeared, the need for operators who manually connected subscribers disappeared. Computers reduced some professions. The internet, on the contrary, created new professions such as SMM specialist, mobilographer, copywriter, and others.

AI will also take some tasks away from humans, but at the same time, new tasks will be added to the specialist's responsibilities.

"If you think 'artificial intelligence has arrived, it will do all my work, and I will rest' — it is not like that. Even more work will fall on your shoulders," says Narzullayev.

In this sense, the risk is not only in the disappearance of the profession, but also in the change of tasks within the profession.

The conversation also discussed how artificial intelligence can provide practical benefits in medicine.

For example, a doctor can record their conversation with a patient in audio format, and AI can convert it into text and format it into a medical history or medical report. This reduces the time the doctor spends on paperwork, allowing them to communicate more with the patient.

In addition, specialized medical models can assist the specialist in processing data such as analyses, X-rays, or MRIs.

But the interlocutors specifically emphasized that in medicine, final control must remain with the human. Because while in another field 95 percent accuracy might be a good result, in medicine, the remaining 5 percent error is related to human health.

How ready is Uzbekistan for the artificial intelligence race?

According to Narzullayev, one of the biggest problems is the lack of qualified personnel.

Creating AI solutions requires large computing power, servers, databases, and financial resources. But infrastructure alone is not enough — specialists who will use it and create solutions adapted to local problems are also needed.

"The biggest problem in Uzbekistan is that today there are almost no specialists in artificial intelligence in the market, only a handful," he says.

Therefore, one of the tasks of the state could be training specialists, educating them abroad, or attracting experienced personnel from abroad, as well as encouraging the private sector to enter this field through state orders.

An aspect of artificial intelligence that is not discussed much is its dependence on a huge physical infrastructure.

To run powerful models, data centers, high-performance chips, large amounts of electricity, and resources for cooling servers are required.

Narzullayev noted that such infrastructure is also forming in Uzbekistan, but the existing capacities are still small.

While data centers create economic opportunities, they also bring about new problems such as electricity and water consumption, cooling, and environmental impact.

Therefore, the AI race is no longer just a competition between algorithms or programmers. Having electricity, chips, data centers, and necessary raw materials is also becoming a strategic advantage.

The topic of fake photos and videos created with the help of AI was also raised in the conversation.

Today, artificial intelligence can mimic a real person's voice, facial expressions, and movements very convincingly. As a result, an ordinary user may find it difficult to distinguish a real video from an artificially created one.

This requires a new skill in information consumption: not automatically accepting any video seen or voice heard as truth.

Especially, the need to verify content that is sensational, emotional, or prompts important decision-making through other sources is increasing.

One of the most important rules in working with artificial intelligence is not to accept its answer as absolute truth.

Narzullayev noted that AI can sometimes present a non-existent source, article, or fact in a confident style.

"Artificial intelligence has one feature — it lies with confidence. It gives a very beautiful result. Looking at the result, you say, 'wow, that's great.' In reality, it may turn out to be a lie," he says.

Therefore, AI literacy does not consist only of asking it the right question. It is also important to know which task can be fully delegated to it, where human control is needed, and which information must definitely be double-checked.

Especially in medicine, law, finance, scientific research, and other highly responsible fields, the final decision must remain with the human.

At the end of the conversation, it was emphasized that not using artificial intelligence does not protect a person from its influence.

A person may not open ChatGPT or another AI program. But in processes such as selecting the content they see on social networks, advertising, search engines, service delivery, and preparing scripts and texts, artificial intelligence has already entered their life.

Therefore, the issue is not about completely avoiding AI or blindly accepting it. The main task is to understand its risks while using its capabilities, protect personal data, and keep important decisions under human control.

"Whether we like it or not, this has already entered our environment. This process will continue," it was said in the conversation.