It encouraged readers to identify mushrooms by "smell and taste," an extremely dangerous practice because even tasting a tiny amount of a deadly mushroom can cause acute liver failure and even death.

The problem came to light after Donovan Thiesson, a mushroom expert, raised the alarm.

The book was later found to have been written by AI.

It is one of the more chilling examples of shallow information in the AI era, where both writers and editors fail in judgment as much as execution.

A study by researchers at UC Berkeley and Cornell University in the U.S. found that AI has become a powerful tool for increasing the quantity of research, though not necessarily the quality.

By analyzing more than two million manuscripts uploaded between 2018 and mid-2024 on non-peer-reviewed scientific platforms arXiv, bioRxiv, and SSRN, they found that AI users increased their output by 50%, and by nearly 90% in biology and social sciences.

The problem is that many of these papers are polished, dense and highly "scientific" in appearance while being of poor quality and less likely to survive peer reviews.

The flood of such manuscripts forces reviewers, research evaluators and funding bodies to spend more time identifying works that genuinely matter.

AI has made it harder to distinguish real quality from among a clutter of convincing-looking but dubious material.

I have encountered a milder version of the same issue in my own work. More and more reports arrive with slick presentations and heavy use of specialized jargon.

Even a new employee can produce speeches that sound authoritative. But once conversations move beyond the surface, it often becomes clear that the writer or presenter does not truly understand the subject.

Behind AI fluency lies a more troubling problem: The illusion of understanding.

AI delivers quick and polished answers to almost any question. As a result, the line between "having information" and "having knowledge" has started to blur. A smooth explanation makes people think they understand the lesson. A good summary makes them think they have read the book. A rewritten paragraph creates the illusion of sharp reasoning. In many cases, what people gain is only the surface of understanding, not its structure.

The human brain does not work like a simple read-write device. Knowledge does not form merely by receiving and memorizing answers, but grows through a slower process of reading, questioning, comparing, and explaining ideas in one’s own words.

That quiet intellectual labor takes time and effort. It rarely delivers instant rewards. But it is what builds real capability, allowing people to understand why something works instead of simply knowing what it is.

Does this AI-driven illusion of understanding pose a real danger?

As is often the case, what comes easily tends not to stay long in memory.

Information delivered through a single click creates a sense of convenience that can weaken curiosity, especially when the demand for information comes from outside pressures such as bosses or teachers rather than personal interest.

This risk is common to everyone, but is especially serious for young people, who still need time to build intellectual foundations in an increasingly competitive future.

For people with strong foundations, AI can expand thinking, test assumptions and improve productivity. McKinsey's 2025 report, "Superagency in the Workplace: Empowering People to Unlock AI's Full Potential", found that experienced workers aged 35 to 44 were the most optimistic and confident about AI.

Over years of work, they have acquired deep expertise in both professional and social realms, and for them, AI functions as an effective assistant for routine tasks.

But for people still learning how to think, a tool that powerful can become a crutch too early.

The future job market will not lack people who know how to use AI. Tasks such as summarizing, drafting, creating spreadsheets, translating, and brainstorming will become cheaper and more common.

What businesses will need are people who can evaluate information, frame the right questions and connect technology to business, society, and human realities.

Future demand may lean toward work that is less machine-perfect but carries a clearer human signature.

Signs of slower hiring and tighter recruitment for younger workers are already emerging. Instead, companies are increasingly giving AI assistants to experienced employees so they can handle workloads two or three times larger than before. Overall costs fall while short-term efficiency rises.

But this could create a dangerous skills gap since experienced workers will eventually leave the labor market even as younger people lose the chance to learn slowly, make mistakes and grow.

Competitive pressures will narrow opportunities further and raise expectations higher. In that environment, there is little room for the illusion of "knowing everything."

In 2002, during a press briefing on national security, then-U.S. Defense Secretary Donald Rumsfeld spoke about different "zones" of information and understanding, later known as the "Rumsfeld matrix."

It includes four categories: Known knowns or things we know that we know, such as verified facts and established information; known unknowns or things we know we do not know, where we recognize a gap in understanding; unknown knowns or knowledge we possess but fail to recognize or use; and unknown unknowns or unforeseen possibilities beyond existing experience or prediction.

The first two, known knowns and known unknowns, are essential starting points for building knowledge. Understanding what we know and what we lack creates the basis for exploring the far more dangerous territory of unknown unknowns.

In simpler terms, self-awareness is the foundation of curiosity and discovery.

AI may help people move faster, but it cannot walk the path of intellectual growth for them.

A society flooded with instant answers is not necessarily a society with deeper understanding.

What needs protection in the AI age is not slowness itself, but the balance between thinking, exploration, speed, and output.

If that difficult process disappears, humanity may eventually face a paradox, an excess of meaningless answers and a shortage of worthwhile questions.

The world will always need people capable of asking the right questions to push further, dig deeper, and uncover new knowledge.