A collaborative study by computer and social scientists at the University of California, Riverside, has revealed a troubling trend for the internet: as artificial intelligence systems become the go-to source for answers, the web is losing the very qualities that made it a vibrant and interesting space. The study, which compared how AI chatbots like ChatGPT and Gemini respond to subjective questions versus traditional web searches, found that human-written content is being replaced by sterile, logic-only responses that lack emotion, personal experience, and opinionated thinking.
The Rise of AI Summaries and the Decline of Human Nuance
Search engines have long been the primary gateway to online information. When a user types a query, the search engine returns a list of links to websites, blogs, forums, and articles, each written by real people sharing their knowledge, experiences, and perspectives. This diversity of voices gave the internet its character, its soul. But over the past few years, major search engines have started integrating AI-generated summaries directly into search results, often pushing organic human-written content below the fold or even eliminating the need to click through to a website at all.
The UC Riverside study specifically examined how AI systems handle opinion-heavy questions, such as whether governments should ban fossil-fuel cars or whether the US healthcare system needs reform. The researchers analyzed the reasoning behind each response and classified it using Aristotle's three pillars of persuasion: logos (logic and facts), ethos (authority and credibility), and pathos (emotion and human experience). They found a stark difference between AI and human responses. Human-written blogs and articles naturally weave all three types of reasoning together, creating content that feels authentic, relatable, and persuasive. In contrast, AI-generated responses rely almost exclusively on logos, offering a dry, factual, and often detached answer that reads more like a textbook than a conversation.
"What we found is that humans essentially use all three of those, whereas LLMs essentially only rely on logos," said Kevin Esterling, a professor of public policy and political science and co-author of the study. "It's not like talking to a person at all. It's just a machine that's predicting what words ought to be said in response to a prompt." This distinction matters because the internet has always been a place where people share not just facts, but also their emotions, biases, and unique perspectives. That messy, human element is what makes browsing the web interesting and educational beyond the mere transmission of data.
The Margarita Example: A Taste of What We Lose
The researchers offered a simple but powerful example to illustrate the difference between AI and human content. Suppose you search for a margarita recipe. An AI chatbot like ChatGPT will quickly provide a clean, competent list of ingredients and instructions: tequila, lime juice, triple sec, salt, ice, shake, strain, serve. It is accurate, efficient, and utterly forgettable. But if you browse a human-written cocktail blog, you might stumble upon a story about how the margarita was named after the Spanish word for daisy and was accidentally created by an Irish bartender in Tijuana who grabbed the wrong bottle while making a drink for a customer. Not only is the story more interesting, but it also gives you a small tidbit you can share with friends while enjoying a pitcher of margaritas, making the experience richer and more rewarding.
This example highlights what the internet is losing: context, character, and the human touch. A machine can spit out facts, but it cannot tell a story. It cannot convey the joy of discovery or the sense of connection that comes from reading something written by a real person. As more people turn to AI for quick answers, especially for topics like health, politics, ethics, and personal advice, society is gradually losing exposure to the diverse human reasoning that shapes how we understand the world. The internet's soul, built on shared human experiences and opinions, is quietly fading away.
The Illusion of Intelligence: Is AI Actually Reasoning?
The study also raises deep questions about the nature of reasoning itself. When a large language model generates a response, it is not actually thinking or reasoning in the human sense. It is performing a complex statistical calculation, predicting which words are most likely to come next based on the patterns it learned from its training data. This process can produce answers that appear thoughtful and logical, but they lack any genuine understanding, intent, or emotional awareness. The AI does not care about the margarita's history; it only knows that in its training data, the word "daisy" often appears near the word "margarita."
This distinction is crucial because people often anthropomorphize AI, attributing human qualities like wisdom, empathy, or creativity to the responses. But as the study shows, when you strip away the logos, there is nothing left. AI does not have a lived experience, a personal opinion, or a sense of humor. It cannot offer a unique perspective born from a lifetime of experiences. It is a mirror reflecting back a statistical average of everything it has been trained on, which increasingly means it is reflecting the most common, bland, and uncontroversial viewpoints.
Implications for Public Discourse and Information Quality
The shift toward AI-generated summaries has significant implications for public discourse. Traditional web searches allowed people to explore a variety of viewpoints, including minority opinions, passionate rants, and emotionally charged arguments. That diversity is essential for a healthy democracy and for fostering critical thinking. When someone searches for a controversial topic like healthcare reform, an AI summary might present a balanced, neutral overview that seems objective but actually sanitizes the debate. It removes the emotional stakes, the personal stories of people who have struggled with the system, and the passionate arguments from activists. In doing so, it deprives the user of the full picture and of the opportunity to engage with the human dimensions of the issue.
Furthermore, the reliance on AI for answers could accelerate the homogenization of online content. If everyone gets the same AI-generated summaries, and if those summaries are designed to be inoffensive and broadly acceptable, the internet could become a monoculture of shallow, unopinionated information. The very things that made the web exciting, the unexpected connections, the angry disagreements, the heartfelt stories, could disappear. The internet would become a vast, efficient, but soulless library of facts, devoid of the human spirit that built it.
The Economic and Creative Toll on Content Creators
There is also an economic dimension. Bloggers, journalists, and independent content creators rely on search engine traffic to sustain their work. If search engines increasingly answer user queries directly with AI summaries, fewer people will click through to the original source. This reduces ad revenue, subscription income, and the incentives for creating original, in-depth content. Over time, the web could lose its incentive for producing high-quality human-written material, further accelerating the decline of the internet's soul. Creators who pour their heart into writing about their passions, from cocktail recipes to political analysis, may find their work devalued and their audiences siphoned away by machines that repackage their labor without attribution or compensation.
Some might argue that AI summaries are simply a convenience, saving users time and effort. But convenience comes at a cost. The study suggests that the cost is the richness and diversity of human expression. A healthy internet needs messy, opinionated, emotional content. It needs people sharing their stories, their failures, and their successes. It needs debates that are not always polite or balanced but that reflect the complexity of real life. When AI replaces that, it may provide quick answers, but it also saps the life out of the web.
What Can Be Done?
While the trend toward AI summaries seems inevitable, the study's authors hope their findings will encourage a broader conversation about the role of AI in information retrieval. They suggest that users should be aware of the limitations of AI and seek out human-written content for topics that require depth, nuance, or personal perspective. Search engines and AI companies could also design their systems to better highlight and link to original human sources, rather than presenting AI summaries as the final answer. Additionally, policymakers and educators can help raise awareness about the difference between machine-generated and human-generated content, and about the value of engaging directly with the messy, emotional, and personal web.
Ultimately, the internet's soul may not be gone yet, but it is certainly at risk. As AI spills everywhere with quick answers, the richness of human experience is being quietly filtered out. The margarita story is just one small example, but it represents a much larger loss. The web was built by humans for humans. If we let machines become the default interpreters of that world, we risk losing not just interesting anecdotes, but the very spirit of curiosity and connection that made the internet so transformative in the first place.
Source: Digital Trends News