According to a new survey, the average American is less than half of what they see and read online.
A survey of 2,000 US adults found that skepticism was the highest ever in the internet.
The results show that Americans believe that only 41% of what they consume online are completely accurate, factual and created by real people.
In fact, they also believe that 23% are completely false and intentionally inaccurate or misleading, while 36% falls in the middle.
Three-quarters say they trust the internet more than ever.
Another 78% agree that the internet has “never gotten worse” in terms of distinguishing between real and artificial.
To put things in perspective, the average American comes across information they know or suspect, showing that AI is generated about five times a week, with 15% being more than 10 times.
Social media posts (48%), news articles (34%), and chatbots (32%) are the top three most likely perpetrators when it comes to content that is generated or misleading to AI.
In fact, voters believe that 50% of news articles and articles they encounter online have an AI element, whether images or actual written content.
Despite doubt, less than a third (31%) are confident in their ability to distinguish when reviews of products or services are written by AI or real people.
This survey was conducted on behalf of Talker Research. worldeven tested Americans for their ability to do that.
Showing a series of business reviews written by AI and humans, only 3 out of 10 (30%) were able to correctly identify which.
Of the three options written by people, two are ranked at the bottom of the list, indicating how easy it is for this day and age.
Because 80% of Americans rely on some level of review when choosing a business to support, bot-created reviews (62%), AI customer service representatives (50%), and AI-generated images (49%) are less likely to love the company.
Unfortunately, 46% of those who voted actually bought something that was not actually advertised, with 24% of those unable to receive a refund or return the item.
“The trust in the Internet has not only faded, it collapsed in an avalanche of noise generated by AI. The Internet has become a mirror home where 78% of Americans can’t distinguish themselves from reality and artificial.” We offer an elegant solution: instant human verification without compromising privacy. No personal data or tracking – simple evidence to restore what’s lost: trust in those we are actually connected online. ”
The survey also revealed the most stressful situation when it comes to distinguishing whether you are dealing with people or chatbots (43%) when talking to customer service representatives.
Remittances through accommodation or hotel bookings (23%) and third-party apps (22%) were also ranked among the top three most uncertainty-inducing situations.
To verify human status, Americans have some tricks on their sleeves. One respondent said, “We often test open-ended questions and human-like responses, such as seeking personal opinions and experiences.”
“What is your birthday? What is your mother’s maiden’s name? Where were you born? Are there any identifiable marks on your skin?”
Beyond that, 24% search for entities online to check the status of users, or 23% ask for phone or video calls.
For all struggles, 82% of Americans agree that businesses and vendors should be legally required to disclose whether AI is used in marketing, content, customer service or websites.
“Being able to prove you’re human online is becoming as important as having an email address from 20 years ago,” Hahn said. “Our research shows that Americans are desperate for tools to restore confidence in digital interaction. Human validation pioneers a new paradigm that becomes the foundational layer of the Internet. It is simple, safe and universally accessible. This does not only solve today’s crises of trust.
Research method:
The Talker survey looked at 2,000 Americans in general population. The survey was commissioned by the world and was managed and conducted online by Talker Research between March 28th and March 31st, 2025.
