po polsku
illustration: DALL-EMore and more often, we feel like the internet has lost its soul. Discussion forums seem empty, and social media is flooded with the same repetitive content. What once was a space for authentic exchange of ideas now resembles a digital wasteland. These feelings fit into the increasingly popular dead internet theory. It suggests that most online content and interactions are no longer created by humans, but by bots and artificial intelligence.
Online traffic dominated by bots
Data seems to confirm these concerns. The latest "2025 Bad Bot Report", prepared by analysts at Imperva, sheds light on the real scale of this phenomenon. It turns out that in 2024, humans accounted for less than half of all internet traffic. It`s machines, not us, generating most of the activity online.
This might be shocking, but bots aren`t just spam in your inbox. These are advanced programs that create fake social media accounts, write comments, and even generate full articles. The rapid development of artificial intelligence has only accelerated this process, making it nearly impossible to tell humans from machines.
| Year | Human-generated traffic | Bot-generated traffic |
|---|---|---|
| 2024 | 49.0% | 51.0% |
| 2023 | 50.4% | 49.6% |
| 2022 | 52.6% | 47.4% |
| 2021 | 57.7% | 42.3% |
| 2020 | 59.2% | 40.8% |
This digital shift makes it harder and harder to find real, human opinions. Authentic user experiences are lost in a sea of artificially generated content. We all feel the consequences, even if we’re not fully aware of them. Our everyday online experience is degrading, and the internet is becoming less useful and trustworthy.
Fake reviews powered by AI
The issue of fake online reviews is one of the most tangible examples of digital inauthenticity. Global statistics are alarming. According to David Ehrentreu, who gathered key stats on Shapo.io, on average 30% of all online reviews are considered fake, and on some major platforms this rate may reach even 47%. This isn’t surprising, considering that 95% of internet users check reviews before buying, and reviews serve as "social proof" that is twelve times more trusted by customers than any other form of marketing.
The fake review phenomenon has evolved from manual operations to mass, automated AI-driven production. Tools like Scalenut Online Review Generator, which create "authentic"-sounding reviews, have drastically lowered the entry barrier for scammers.
According to The Transparency Company, AI-generated reviews have been growing 80% month-over-month since June 2023. Meanwhile, a report by Originality.ai found that in 2025, 23.7% of real estate agent reviews on Zillow were likely created by AI - a huge jump from just 3.63% in 2019.
CapitalOne Shopping Research estimates that this year, fake reviews will cost the average e-commerce customer $125. Globally, this fraud costs us $770 billion. The dead internet is no longer just a theory - it`s a very real cost. For every one of us.
Dead communities
From a global perspective, the social media platform most saturated with bots seems to be X (formerly Twitter). A study conducted in 2024 by Internet 2.0 using the 5th Column AI tool predicts that around 64% of all accounts on the platform are likely bots. Meanwhile, the BotNot team from Yofi.ai`s "flash experiment" from May 2022 estimated that between 24% and 37% of X`s daily active users were bot accounts. Although the methodologies differed, the trend of the internet "dying" is clearly visible over just two years.
But the issue affects all major social platforms. Data analysis shows there is no single consistent number defining the bot problem. This discrepancy stems from differences in research methodologies and the dynamic nature of bots themselves.
| Platform | Metric | Estimate | Source |
|---|---|---|---|
| X (Twitter) | accounts | 5% | |
| active users | 24%-37% | Yofi.ai | |
| accounts | approx. 64% | 5th Column AI | |
| active users | >5% | META | |
| ads | 50%-70% | Reddit / industry research data | |
| active users | 4.75% | Basic Thinking | |
| accounts | 10% | MDPI | |
| TikTok | Deleted accounts | 171 million | Fake accounts removed in Q1 2024 |
| ads | 75%-97% | Fraud0.com |
Most importantly, there`s no unified definition of a "bot". Tech companies like Meta or X often use narrow definitions focused on "non-human accounts", often separating them from spam accounts. Independent researchers, on the other hand, use broader criteria, classifying accounts based on behavior patterns suggesting inauthenticity. As a result, platform-reported numbers are usually much lower than those from external sources. Companies have a clear interest in showing low numbers to maintain trust among advertisers and investors.
Disinformation
AI-driven disinformation was identified by the World Economic Forum in 2024 as a top short-term threat to global stability.Using the same content-creation tools as real users, bot armies can easily produce and widely spread fake news.
The scale of bot involvement in disinformation campaigns is massive. A study from the National Institute on Drug Abuse in Baltimore on COVID-19 disinformation found that up to 66% of bots actively discussed the virus. The findings were published in the Journal of Medical Internet Research.
In Poland, a report from the National Security Bureau showed that in the first months of 2024, a pro-Russian "bot army" (SDA) generated 33.9 million comments and nearly 40,000 content pieces on polarizing political and social topics. This highlights the scale and organized nature of such campaigns. Worldwide. Because the mechanisms are similar and can now easily be adapted to local contexts. And almost every topic can be framed in a polarizing way.
Internet for replacement?
Despite impressive technological progress, we won’t win the fight against the dead internet with algorithms alone. Technology is a double-edged sword. Bots are becoming more advanced, and social media algorithms promote content that generates engagement. In the end, the most effective line of defense remains our human ability to think critically.
Researchers are also studying the phenomenon. In the analysis "The Dead Internet Theory: A Survey on Artificial Interactions", published on ResearchGate, scientists examine how artificial interactions affect us and our behavior. The conclusions show growing user fatigue with inauthenticity and a search for more genuine online relationships.
In practice, this means we increasingly value niche communities, small internet forums, or private groups where we can still count on real contact with another human. That`s where, far from the noise of algorithms and bots, the internet still lives. For now.
COMMERCIAL BREAK
New articles in section Media industry
Why do we believe fakes? Science reveals the psychology of virals
KFi
Why do emotions grab more attention than evidence, and why can a fake authority overshadow scientific data? Researchers from Warsaw University of Technology, Jagiellonian University, and SWPS University in Poland sought the answers. Here are their findings.
Investigative journalism in Europe. Newsrooms face pressure
KFi, Newseria
Media and political representatives point to the difficult situation of investigative journalism in Europe. Newsrooms are reluctant to invest in this segment due to high costs and the large amount of time and effort required. Most of all, however, they fear legal proceedings.
Energy under attack. Disinformation threatens Poland’s power transition
KFi
One in five online messages about energy may be fake. Between 2022 and 2025 nearly 70,000 publications warning and condemning disinformation in this strategic sector were recorded in Polish media. They generated a reach of 1.19 billion impressions.
See articles on a similar topic:
Investigative journalism in Europe. Newsrooms face pressure
KFi, Newseria
Media and political representatives point to the difficult situation of investigative journalism in Europe. Newsrooms are reluctant to invest in this segment due to high costs and the large amount of time and effort required. Most of all, however, they fear legal proceedings.
Artificial intelligence in newsrooms. Three realities of the AI era in media
Krzysztof Fiedorek
According to a report by the European Broadcasting Union, many newsrooms already use AI but still do not fully trust it. Audiences do not want "robotic" news, and the technologies themselves though fast can be costly, unreliable, and surprisingly human in their mistakes.
How do we assess news credibility? Data analysis from 40 countries
Krzysztof Fiedorek
Are people defenseless against false information? Do they really fall for clickbait and fake news? A meta-analysis of 67 studies involving 200,000 people shows the problem is different than we thought. Instead of excessive gullibility, we are dealing with the opposite.
Women in media 2025. Editorial power knows no equality
KFi
Only 27% of editors-in-chief in the media are women, even though they make up 40% of journalists. In 9 out of 12 countries studied by the Reuters Institute, women in media are less likely to get promoted. It seems that equality in newsrooms is lagging behind broader society. And the gaps go much further.
