AI: Will We Still Think Deeply When AI Reads for Us?

*Will We Still Think Deeply When AI Reads for Us?*

For subscribers only

Translated by ChatGPT

https://www.zaobao.com.sg/forum/views/story20260522-9091900?utm_source=android-share&utm_medium=app

2026-05-22

Lianhe Zaobao (联合早报）

Author: Dr. John C. Keng (耿庆文博士)

The author is a Canadian cultural observer and holds a doctorate from Cornell University in the United States.

=====

Since the beginning of this year, two seemingly unrelated news developments have in fact pointed to a common issue worthy of concern.

First, many universities in the United States have begun reinstating paper-based examinations and handwritten classroom assignments. The reason is not complicated: after the widespread adoption of generative artificial intelligence (AI), more and more students are directly using AI to complete essays, summaries, and reading reports, making it difficult for teachers to judge students’ true abilities.

At the same time, technology companies such as OpenAI and Google have successively launched “AI summaries” and “AI agent” functions. Users no longer even need to open the original webpages, as AI can automatically condense news, organize key points, and generate conclusions.

On the surface, this appears to be an efficiency revolution. But the deeper issue is this: when AI begins to “read” on behalf of humans, are humans also gradually losing the interest or ability for “deep reading”?

Over the past 20 years, the internet has already changed the way humans read. Short videos, instant notifications, and social media have made people increasingly accustomed to receiving fragmented information. People continuously scroll through screens, yet become less and less able to remain engaged with complex discussions for long periods. Generative AI may further reinforce this kind of “cognitive shortcutting.”

AI’s greatest temptation lies in allowing people to avoid the process of “difficult thinking.” A book that once required days to read can now be summarized within minutes; before a long article is even carefully examined, AI has already extracted the key points. Even comparison of viewpoints, logical organization, and conclusion synthesis can now be completed instantly.

As a result, people increasingly mistake “knowing the conclusion” for “understanding the content.”

But true reading is not merely about receiving information. The process of deep reading is essentially a slow construction of the mind. During prolonged reading, people experience hesitation, comparison, reflection, questioning, and renewed understanding. What truly matters is not merely the final answer obtained, but how one’s thinking ability itself is trained.

French philosopher Bernard Stiegler long warned that digital capitalism is creating a phenomenon of “spiritual proletarianization” — human memory, attention, and judgment are gradually being outsourced to and weakened by technological systems.

American cognitive neuroscientist Maryanne Wolf has also pointed out that prolonged fragmented reading weakens humanity’s capacity for deep reading. She worries that once people become accustomed to rapid browsing and instant stimulation, the brain will increasingly struggle to enter the state of complex thinking that requires patience and concentration.

Similar warnings also appear in the research of American technology critic Nicholas Carr. In his book The Shallows, he argues that the internet is not merely changing the way information is transmitted, but is also reshaping the structure of human thought.

One of the greatest risks of the AI era is not that machines become smarter than humans, but that humans become increasingly unwilling to engage in “high cognitive-cost” thinking. Deep reading already runs counter to humanity’s instinct for immediate gratification. It requires patience, solitude, and sustained concentration. Yet today’s platform economy and AI systems are built upon the logic of “reducing cognitive friction.” Algorithms want people to obtain stimulation faster, get answers in less time, and remain continuously engaged on platforms.

As a result, human civilization faces a dangerous contradiction: technology continuously improves the supply of information, while human understanding may simultaneously decline. This is why more and more educators are beginning to rethink what schools should truly cultivate in the future.

In the past, education emphasized the “acquisition of knowledge.” But when AI can retrieve massive amounts of information within seconds, the truly scarce ability instead becomes “deep understanding.”

The most important talents of the future may not necessarily be those who are best at using AI, but those who can still read complex texts for extended periods, independently analyze problems, and maintain deep concentration. AI can help organize information, but it cannot truly build character on behalf of a person.

A society that loses its capacity for deep reading will also gradually lose its ability for deep democratic discussion. Public issues will become increasingly emotionalized, instantaneous, and label-driven. Complex problems will no longer receive patient understanding, and society may ultimately become dominated by short-video-style politics and algorithmic recommendations.

Therefore, the decline of “deep reading” is not merely an educational issue; it may also become a civilizational crisis. What truly needs to be defended in the AI era may not only be copyright, data, and privacy, but also humanity’s capacity for “slow thinking.”

Future education may need to reestablish several core values: protecting the ability to concentrate for long periods, emphasizing the importance of the reading process, and helping younger generations understand that AI is a tool, not consciousness itself. What ultimately determines the height of civilization is not technology itself, but whether humanity still retains independent thinking ability.

The author is a Canadian cultural observer and holds a doctorate from Cornell University in the United States.