Can neural networks replace digital specialists?

Contents:

• «Content volume has tripled while quality has decreased»
• «Decreasing quality and reducing reach»
  
• «Many articles have received comments expressing bewilderment: "What is this?"»

  «Readers have left comments full of doubt: "What nonsense?"

  «Many comments under articles have expressed bewilderment: "What nonsense?"

  «Under various publications, opinions have appeared asking: "What nonsense?"

  «In the comments to the articles, voices have been heard criticizing: "What is this at all?’”

• “Copywriting is under threat: how neural networks can displace professions”

  “The future of copywriting: the risks of replacing specialists with artificial intelligence”

• “Changes in content lead to a loss of subscriber trust”
  “Subscribers notice changes and begin to doubt the reliability of the brand”
  “Inconsistency of content causes distrust among the audience”
  “Users realize the difference and begin to lose faith in the company”
  “Inconsistencies in content undermine subscriber trust in the brand”

Every day, employers are increasingly looking for ways to optimize their content creation costs, replacing editors, social media specialists, marketers, and copywriters with solutions based on neural networks. We discussed this situation with digital experts who have already encountered similar changes.

According to participants in a survey conducted by the Skillbox Media Marketing editorial team, artificial intelligence is ineffective in solving these problems: the quality of the content produced is declining, audience reactions are becoming less positive, and the time that should have been saved is spent on rework. We present to you five real cases where they tried to replace authors with AI.

• After cutting costs on writers, the project stopped making a profit.
• The screenwriter was replaced with a neural network, and as a result, reach decreased significantly.
• By entrusting content creation to artificial intelligence, companies faced a drop in sales and a loss of subscribers.
• Copywriters were fired, and now content managers are editing texts.
• They tried to increase the number of leads using artificial intelligence, but the result was zero.

"Tripling the volume of content while reducing its quality"

Social media specialist, creator of the Telegram channel "Dima, what other SMM?"

I had a client - one of the leading marketplaces in Russia. They asked me to create a network of Telegram channels covering various product categories and target audiences. We started working on the project, effectively developed content and achieved sales. We primarily posted content in the text and image format.

One day, a client decided to reduce content creation costs and outsource post development to neural networks. The artificial intelligence independently found products, extracted images from them, formulated the text, and posted it all to the channel. As a result, the volume of content tripled, but the quality left much to be desired and dropped threefold.

Initially, the texts generated by the artificial intelligence were not edited, but their quality raised numerous complaints. Later, we decided to train the neural network on our own content, but even then, various artifacts emerged. We also experimented with manually rewriting the texts generated by the neural network, but the editing work was limited as we aimed to minimize costs.

Over time, the situation with the metrics worsened: we observed a decrease in engagement and an increase in unsubscribes. About 20% of subscribers decided to unsubscribe, and approximately 40% stopped being active in relation to content. After several months of work in this format, the project became unprofitable and was closed.

So, friends, don't worry: neural networks will not replace us, but will only serve to automate certain tasks.

Social media specialist, creator Telegram channel "Dima, what SMM?"

I had a client—one of the leading marketplaces in Russia. They approached me with a request to create a network of Telegram channels covering various product categories and target audiences. We enthusiastically took on the project, developing high-quality content and achieving sales. Our publications were mainly a combination of text and images.

One day, the client decided to save on content creation and entrust this process to neural networks. The artificial intelligence independently found products, extracted images, generated texts, and published them in the channel. As a result, the volume of content tripled, but its quality significantly deteriorated.

Initially, the texts generated by the artificial intelligence were unedited, but their quality was quite low. Then we began training the neural network on materials created by our specialists, but, unfortunately, this also did not produce ideal results, as various artifacts appeared. We also tried manually editing the texts generated by the neural network, but kept editing to a minimum to avoid increasing costs.

Over time, the performance began to deteriorate: we observed a decrease in user activity and an increase in unsubscribes. Approximately 20% of subscribers unsubscribed, and about 40% stopped interacting with the content. After several months of such work, the project no longer generated income and was closed.

So, colleagues, don't worry: neural networks will not replace us, but will only facilitate the automation of some tasks.

"Decreasing quality and reducing coverage"

Screenwriter and creator of the Telegram channel "Screenwriter Mari Santi."

Previously, I worked with a small company organizing unique tours. I developed a content plan and wrote scripts for Instagram reels. The business owners filmed, edited, and published the videos themselves.

The results were impressive. In just a few months, the account's following increased from 4,000 to 24,000. Views also increased significantly: while previously there were only 200-300, they soon rose to 10,000, then 20,000, and eventually reached 100,000. I managed to create several videos that each garnered over 1 million views. I deliberately chose current trends and simple formats to make the content creation process effortless.

At some point, the clients reported that they had basically mastered creating scripts and decided to outsource my tasks to neural networks. This approach will be simpler and more cost-effective than hiring someone to do this work full-time.

Based on my information, they have been using artificial intelligence to generate scripts for some time and even published content based on them. However, the quality of the materials has significantly declined, and, as a result, their audience reach has also decreased. Their social media account currently does not exist - it has likely been deleted or renamed.

I believe that neural networks are currently incapable of generating creative content on par with humans. Artificial intelligence lacks an understanding of current trends, cannot adequately convey emotions, and cannot make jokes. Even if detailed instructions are provided, the scripts generated by neural networks are likely to be somewhat strange and absurd. Perhaps this is the reason why my clients have not been able to replace specialists with artificial intelligence.

Screenwriter and creator of the Telegram channel "Screenwriter Mari Santi".

Previously, I collaborated with a small company that organized signature excursions. My task was to develop a content plan and create scripts for Instagram videos. The company owners handled the filming, editing, and publishing of the content themselves.

The results were impressive. In just two months, the account's following increased from 4,000 to 24,000. Views also increased significantly—from 200-300 views, we began getting 10,000 to 20,000, and then up to 100,000. Some videos even surpassed 1 million views. I deliberately chose current trends and simple formats so they could be easily filmed without outside assistance.

Some time ago, clients told me they had largely mastered script creation and decided to outsource my tasks to neural networks. This approach will be simpler and more cost-effective than hiring a human.

As far as I know, they did indeed use artificial intelligence to create scripts for some time and even published content based on them. However, the quality of these materials has noticeably deteriorated, which has naturally led to a decline in reach. Their social media account is currently unavailable—perhaps it has been deleted or renamed.

In my opinion, neural networks are currently incapable of generating creative content at a human level. Artificial intelligence is not yet aware of current trends, and it lacks the ability to convey emotions and create jokes. Even with a detailed query, the results of neural networks can be somewhat absurd and illogical. Perhaps this is the reason why my clients have not been able to replace a professional with artificial intelligence.

"Many articles have received comments expressing bewilderment: "What is this?""

"Readers have left reviews full of doubt: "What nonsense?""

"Many comments under articles expressed bewilderment: "What nonsense?""

"Under various publications, opinions appeared asking: "What nonsense?""

"In the comments to the articles, critical voices were heard: "What is this anyway?""

Content management specialist, creator of the Telegram channel "Content Alignment".

I held the position of content manager and copywriter in a small online school focusing on health issues.

All the provided material was prepared by an in-house expert, and I, based on it, developed a content strategy, including even the creation of stories. I also created technical specifications for the editor and designer. I also managed three Telegram channels and wrote scripts for reels. Our work was organized efficiently, and we were pleased with the results. For example, we were able to develop a quarterly sales plan in just one week.

I decided to leave the project on my own, striving for professional growth. Our parting was a positive one. I even offered to help train a new copywriter when the company found my replacement. However, unfortunately, no one ever contacted me about this.

A month later, I noticed that I still hadn't left the school's work chat. In a conversation with colleagues, they were discussing declining sales and the need to urgently find a solution. From their messages, I realized that a new candidate had not been found. Ultimately, all content was created using neural networks, and its editing and publication was handled by an employee who had never worked with texts before.

As a result, there was a significant decline in the number of subscribers on the school's accounts. Over the course of a month, their number dropped from 12,000 to 9,000. A number of posts were accompanied by disgruntled comments, such as, "What kind of nonsense is this?" and "Why have all the posts lost their sincerity?"

In my professional work, I actively use neural networks. For example, I might turn to ChatGPT to get a summary of a lengthy text or to ensure I've captured its main idea correctly. However, relying on neural networks for writing materials without any editing is not the smartest solution. Currently, these systems are unable to accurately convey the author's individual style, sometimes make factual inaccuracies, and overall write rather lifelessly.

Content specialist, creator of the "Content Alignment" Telegram channel.

I held the positions of content manager and copywriter at a small online school focused on health.

All information came from an in-house expert, based on which I developed a content strategy, including even the creation of stories. I also formulated technical specifications for the editor and designer. In addition, I managed three Telegram channels and wrote scripts for reels. Our work was organized, and the results were immediate. For example, we managed to create a quarterly sales plan in just a week.

I independently decided to leave the project, as I sought to develop in my professional field. Nevertheless, our parting was amicable. I even expressed a desire to train a new copywriter as soon as the company found my successor. However, in the end, no one contacted me.

A month later, I noticed that I was still in our school's work chat. Discussions among colleagues revealed that sales had dropped significantly and urgent action was needed. From what I read, it became clear that a new employee had never been hired. Instead, the team decided to use a neural network to generate all content, and delegated the review and posting of materials to someone with no previous experience working with text.

As a result, there was a significant decline in the number of followers on the school's account. Within a month, their number dropped from 12,000 to 9,000. Numerous posts were followed by comments expressing dissatisfaction, such as, "What is this?" and "Why do all the posts seem so insincere?"

In my professional work, I actively use neural networks. For example, I might ask ChatGPT to summarize a lengthy text and evaluate how accurately I captured its main idea. However, relying on automatic text generation without subsequent editing is not the best approach. Neural networks are not yet able to accurately convey the author's individual style, sometimes make factual errors, and, in general, create materials devoid of depth and emotion.

"Copywriting under threat: how neural networks can displace professions"

"The future of copywriting: the risks of replacing specialists with artificial intelligence"

Copywriter

— My experience includes working at an agency specializing in SEO website promotion. I primarily wrote articles for clients in a wide variety of industries, from automotive to high-end furniture production.

In May 2023, when neural networks were beginning to gain popularity, the agency's director insisted on using them in their daily operations. His goal was to stay ahead of the curve and improve overall productivity.

Ultimately, all copywriters completed training on using artificial intelligence and began actively using the paid version of ChatGPT. Over time, the writing process for certain articles did indeed speed up, but editing the texts generated by the neural network and fact-checking remained necessary steps.

In April 2025, it became clear that top management was still dissatisfied with our achievements. The HR manager expressed the opinion that the copywriter profession was already outdated and that we could soon be displaced by neural networks. To maintain our position and continue our careers in this field, we should reconsider our approaches to organizing our work.

In June 2025, all of the agency's copywriters were fired, and the text creation process was completely transferred to artificial intelligence. Now, content managers, who were previously uninvolved in this process, are responsible for editing and publishing finished materials. I had the opportunity to review the texts generated by AI and noticed that errors appeared even in the first paragraph.

I currently work for a different SEO agency that uses a completely different approach. Previously, they used neural networks in conjunction with SEO experts to write articles. However, over time, they came to the conclusion that artificial intelligence is not capable of producing texts as high-quality as humans. As a result, they decided to hire a professional copywriter and were pleased with the results achieved.

Copywriter

— My work experience is related to an agency specializing in SEO website promotion. I primarily focused on creating content for clients in a variety of sectors, ranging from automotive to high-end furniture.

In May 2023, as neural network technologies began to rapidly develop, the agency's director decided to integrate them into his workflows. His goal was not only to keep up with current trends but also to improve overall efficiency.

As a result, all copywriters completed a course on using artificial intelligence and began using the paid version of ChatGPT. Over time, the writing process for certain articles did indeed speed up, but the need for thorough editing of texts generated by the neural network and fact-checking remained urgent.

In April 2025, it became clear that management remained dissatisfied with our results. The HR manager expressed the opinion that the copywriting profession had essentially lost its relevance and that artificial intelligence would soon take over our position. To save our jobs and remain in this field, we need to rethink how we organize our work.

In June 2025, all of the agency's copywriters were laid off, and the copywriting process was completely handed over to artificial intelligence. Now, content managers, who previously had no role in this task, are responsible for editing and publishing finished materials. I've seen texts created by AI, and errors can be found in the very first paragraph.

I currently work for another SEO agency that uses a completely different strategy. Previously, they used a combination of neural networks and SEO specialists to create articles. However, over time, they concluded that artificial intelligence is not capable of providing the same level of content quality as humans. As a result, they decided to hire a copywriter and were pleased with the results.

"Changes in content lead to a loss of subscriber trust"
"Subscribers notice changes and begin to doubt the reliability of the brand"
"Inconsistency in content causes distrust among the audience"
"Users realize the difference and begin to lose faith in the company"
"Inconsistencies in content undermine subscriber trust in the brand"

Editor, content marketing specialist, creator of the telegram channel “Sweet Editor.”

One of my clients showed a keen interest in creating content using neural network technologies. He strongly recommended that I take an extensive course on the topic and then asked me to develop a sales funnel for him using the capabilities of artificial intelligence.

The funnel was impressive in its scale: 25 scenarios, each with five branches. Multiply this, and the results are staggering. The client was striving for deep segmentation of their audience and a personalized warm-up process. All this had to be implemented in just two weeks. A true classic.

We had to use social media and work late into the night. Ultimately, the team consisting of Kolya and artificial intelligence created something acceptable. While I can't say I'm proud of the results, at least we managed to meet the established deadline.

Ultimately, this funnel did not bring the client the expected results. The main problem was the lack of resources for traffic acquisition, and even with a limited budget, the funnel showed only average results. In fact, it couldn't have been any other way, as initially, it was just a FOMO-fueled idea that someone could create a large volume of polished content using artificial intelligence and successfully implement it. This very concept was actively presented in the course on neural networks. However, in reality, the created content was of low quality and did not contribute to high conversions. Moreover, the large number of scenarios diluted the budget, making it difficult to focus on improving a single funnel for effective results.

Overall, I have serious doubts about creating content using artificial intelligence. Such works are often characterized by low quality and lack of originality. Subscribers who encounter such generated texts are able to detect the difference, which ultimately leads to a loss of trust in the brand. It's acceptable to use AI to generate ideas or refine individual sections, but using it to write entire content is, in my opinion, unacceptable. And, of course, we shouldn't forget about the so-called AI content factories that display hashtags like #replaced5editorswithAI at every turn.

Editor, content marketing specialist, creator Telegram channel "Sweet Editor".

One of my clients showed a keen interest in creating content using neural network technologies. He even insisted that I take an extensive course on this topic, and later asked me to develop a sales funnel for him using artificial intelligence.

The funnel was truly colossal: 25 scenarios, each with five branches. We multiply and are amazed by the result. The client strove for maximum segmentation of their audience and personalization of the warm-up process. All this had to be implemented within two weeks. A true classic.

We had to use social media and work at night. Ultimately, the team consisting of Kolya and artificial intelligence created something acceptable. Although I can't say I'm proud of the result, at least we managed to meet the deadline.

Ultimately, this funnel did not produce the expected results for the client. The main problem was insufficient financial resources for attracting traffic, and even with a small investment, the funnel's effectiveness was mediocre. In fact, even more money wouldn't have guaranteed results. The idea that someone could create a ton of high-quality AI content and use it successfully was inspired by the fear of missing out (FOMO). This very concept was actively promoted in the course on neural networks. In practice, the content created was of low quality, leading to weak conversions. Furthermore, the large number of scenarios dilutes the budget and makes it impossible to focus on improving a single funnel for more effective results.

In general, I am skeptical of content creation using artificial intelligence. Such materials often turn out to be quite mediocre in content. Subscribers who encounter AI-generated texts begin to notice the differences and, ultimately, lose trust in the brand. Using AI to generate ideas or edit individual sections is one thing, but writing full content is quite another. And I certainly don't want to mention those well-known AI content factories that use hashtags like #replace5editorswithAI.

If you're interested in neural networks, check out the materials from Skillbox Media.

• A neural network is a complex computational model inspired by the structure and functioning of the human brain. It consists of many artificial neurons that are combined into layers and interact with each other. The main task of a neural network is to recognize patterns and process data, allowing it to learn and improve its results based on experience.

  A neural network begins with input data, which passes through several layers of neurons. Each neuron performs a specific mathematical operation, transforming the information and passing it on. The network is trained using large volumes of data, which allows it to identify patterns and dependencies. During training, the neural network adjusts its internal parameters to minimize prediction errors.

  Once training is complete, the neural network is able to accept new data and make predictions or classifications based on the knowledge it has learned. This process makes neural networks a powerful tool in various fields, such as computer vision, natural language processing, and many others.

• 12 free neural network services available in Russian.
• DeepSeek neural network: its capabilities and reasons for its popularity

  DeepSeek is an artificial intelligence system with many impressive features that has attracted the attention of many. It is capable of performing complex tasks that were previously considered difficult for machine learning.

  One of DeepSeek's key features is its ability to process huge amounts of data, extracting useful information from it with high accuracy. This makes it an indispensable tool in various fields, including data analysis, natural language processing, and computer vision.

  DeepSeek's popularity is due not only to its functionality, but also to the fact that it opens up new horizons in the field of technology. Many professionals and companies are interested in using it to optimize processes and increase work efficiency.

  Thus, DeepSeek attracts attention due to its outstanding capabilities and significant impact on the modern technological environment.

• Neural network tools for SMM specialists: 4 platforms that will help optimize activity on social networks.
• Artificial intelligence systems for marketplaces: four tools that will significantly increase work efficiency.