Connect with us

EduNews & Views

Knowing When to Think for Ourselves: AI, Intuition and the Education of Discernment

As students increasingly turn to AI at the first sign of difficulty, Dr Vijayakumar Parameswaran Unnithan examines the risks to independent judgment and explains how schools can nurture critical thinking and discernment.

Dr. Vijayakumar Parameswaran Unnithan

Published

on

Independent Thinking in the Age of AI
Encouraging students to work through difficult problems independently can help develop the critical thinking and discernment needed to evaluate AI-generated answers. Representational image. Image credit: Quí Trần/ Pexels

Every learner knows the moment when a problem stops being easy. The first attempt does not work, the next step is unclear, and a small discomfort sets in. For generations, that discomfort was where thinking began. Today, a chatbot increasingly fills that space.

As a career architect, I keep researching what students aspire to, how aware they are of their own paths, and where their motivation comes from. Between July and September 2025, I worked with 42 high school students, and about four in five of them used AI regularly. Four students described a habit that has stayed with me.

Whenever a homework problem felt a little tough, they turned to AI. They did not wait until they were stuck. The first sense of difficulty was itself the cue to hand the problem over, and to them this felt entirely reasonable. That moment, when difficulty arrives and is passed on before it can do its work, is what this article is about.

How AI Changes the Way We Judge Answers

Daniel Kahneman taught us to see the mind as working in two modes. System 1 is fast, associative and intuitive. It lets us read a face, finish a sentence or sense danger without deliberate effort, and it is also where anchoring, overconfidence and other familiar biases live. System 2 is slow and effortful. It questions a first impression, checks the evidence and revises a judgment.

System 2’s corrective work does not continuously scrutinise System 1. It is recruited when circumstances demand additional attention, for example, when something is difficult, surprising, or signals a possible error. Our own thinking can sometimes provide signals, such as difficulty, surprise, hesitation or contradiction, that prompt further scrutiny. The four students I spoke with seemed to have developed a habit of treating that very signal as a reason to stop thinking and ask the machine.

An answer produced by AI arrives without these signals. Whatever happens inside the machine, including the step by step reasoning that newer models display, the output reaches the student as a finished intuition. It is fast, fluent and confident, and fluency is persuasive in its own right. Decades of research on processing fluency show that people treat ease of reading as a cue to truth. In a classic experiment, statements that were easier to perceive were judged as true more often than the same statements presented with lower perceptual fluency.

Before generative AI, polished prose usually required education and editorial effort, so fluency carried some information about its source. Generative AI makes fluency nearly free, while readers continue to trust it as though it were costly to produce.

The result is an asymmetry. When a student’s own intuition is wrong, she at least has a chance of feeling it. When a judgment generated by AI is wrong, the burden of noticing falls entirely on her, and she is likely to process the machine’s fluent answer through her own System 1. Plausibility takes the place of scrutiny.

The Evidence Behind “Cognitive Surrender”

Researchers have begun to measure this. Steven Shaw and Gideon Nave at Wharton describe AI as a third system of cognition operating outside the brain, and they call the tendency to adopt AI outputs with minimal scrutiny “cognitive surrender”. Across three preregistered experiments with 1,372 participants, accuracy rose by 25 percentage points when the AI was correct and fell by 15 points when it was wrong.

People were not evaluating the answers so much as inheriting them. A survey of 319 knowledge workers by Microsoft Research and Carnegie Mellon points the same way. Workers who placed more confidence in AI engaged in less critical thinking, while those more confident in their own abilities engaged in more.

Why Discernment Matters More Than Ever

This is why the capacity that matters most now is discernment. Discernment is different from simply thinking more slowly. It begins with recognising that a judgment deserves examination in the first place, especially when nothing feels wrong. It continues by asking why an answer seems convincing, and whether that conviction comes from evidence or only from polish. It includes noticing what the answer leaves out.

It ends with a decision to accept the output, modify it, seek another view, or set it aside. Discernment is therefore not only the capacity to evaluate an answer. It is the capacity to recognise when accepting an answer without independent examination is itself a decision that requires scrutiny.

System 2 remains necessary, but what it scrutinises has changed. Students are no longer examining only the products of their own intuition. Increasingly, they are examining judgments produced by another cognitive system, one that operates continuously, rapidly and at scale.

Developmental Offloading: What Students May Lose

For adults, relying on AI is often a matter of cognitive offloading, handing a task to a tool while keeping the underlying capacity intact. For students, the issue may be different, because many of the capacities involved in independent judgment are still developing. When the environment continuously supplies ready-made interpretations, a young person has fewer occasions to generate an interpretation independently, tolerate uncertainty, compare alternatives and discover that a first impression was wrong.

Education: Young child looking thoughtfully at a computer screen, illustrating the role of independent thinking and discernment as children increasingly interact with AI.
As children increasingly turn to AI for answers, education must help them develop the independent thinking and discernment needed to question what they see on screen. Representational image. Image credit: Boris Hamer/ Pexels

The concern for education, then, is what I call developmental offloading. Capacities that are repeatedly delegated are exercised less often, and so they may not develop fully.

The evidence from classrooms supports this concern. In a field experiment with nearly a thousand high school mathematics students, Bastani and colleagues found that access to GPT‑4 improved performance while the tool was available, but students given unrestricted access performed worse when AI access was removed than students who had never had access.

A study of university learners published in the British Journal of Educational Technology found that ChatGPT significantly improved short term task performance without boosting intrinsic motivation, knowledge gain or transfer, and the authors named the pattern “metacognitive laziness”. The work looked better while the learning underneath it grew thinner.

The deepest risk goes beyond whether a student follows a particular answer. Over time, repeated interaction with AI may shape what the student notices, what she considers plausible and what she regards as worth questioning. At that point the machine is no longer only answering questions. It is quietly shaping the learner’s sense of which questions need asking.

How Schools Can Use AI Without Weakening Independent Thinking

None of this is an argument for keeping AI out of classrooms. The same Bastani study offers the most hopeful finding in this field. Safeguards made a real difference, and a tutor instructed to give teacher designed hints rather than hand over answers largely mitigated the harm. The technology did not decide the outcome. The design of its use did. AI augments discernment when its outputs provoke inquiry, comparison and reflection, and it erodes discernment when its outputs routinely replace them.

Schools can act on this. Teachers can ask students to stay with a difficult problem for a set time and form their own answer before consulting AI, and then to compare the two and explain the difference. They can set tasks where the question is not whether the AI’s answer is correct but why it sounds convincing and what it leaves out.

They can reward students for catching errors in machine output, which turns scepticism into a skill that earns recognition. Periodic assessments without AI can reveal what a student can do alone, so that neither the student nor the teacher mistakes assisted performance for capacity. And teachers can model doubt openly, showing students what it looks like to pause before a fluent answer and ask whether it has earned trust.

Helping Students Stay With Difficulty

Perhaps the most important support is also the simplest. A student turns to AI at the first sign of difficulty partly because the difficulty feels like evidence that she cannot do the problem. Teachers can change what that difficulty means. I saw this directly in my own work with these students. When a student hesitated during our sessions and was prompted with “think again”, “you know it” or “you can do it”, that prompt was enough to keep most of the students engaged with the problem. Instead of reaching for AI, they stayed with the problem and drew on what they already knew.

This is scaffolding in its truest sense, support that holds the student within the struggle long enough for her own capacity to take shape, and then steps back. The wider research points the same way. People who were more confident in their own abilities engaged in more critical thinking, while those more confident in AI engaged in less. A student’s belief that she can do it is not a soft extra. It is part of what keeps discernment alive.

The Question Education Must Answer

The four students I spoke with were not refusing to think. They knew they were turning to AI, and it seemed entirely reasonable to them. What they could not see was what they were handing over with it, the moment of difficulty in which discernment grows. The question facing education is not whether AI can think for our students. It is whether, through repeated interaction with AI, they continue to develop the capacity to decide when they should think for themselves.

Dr Vijayakumar Parameswaran Unnithan is a Career Architect and Director of Research at CareerTotus Consultants Pvt. Ltd, India. As a researcher, he works at the intersection of career development, employability, and organisation development.

EduNews & Views

Global Patent Applications Hit Record 3.9 Million in 2025; India Sees 37.5% Growth

Global patent applications reached a record 3.9 million in 2025, with China accounting for more than half the increase. India recorded a 37.5% rise, crossing 100,000 applications for the first time.

Published

on

Official document with a verification stamp, illustrating the legal process of patent applications and intellectual property protection.
Patent applications seek legal protection for new inventions, giving applicants exclusive rights if their inventions meet the requirements for a patent grant. Representational image. Image credit: Markus Spiske/ Pexels

Global patent applications crossed 3.9 million in 2025, registering a 6% increase over the previous year and the fastest annual growth since 2016. According to the World Intellectual Property Organization (WIPO), China accounted for more than half of the additional applications. India recorded a 37.5% rise, crossing 100,000 filings for the first time.

Patent applications are formal requests for legal protection of inventions. They allow individuals, companies and research institutions to seek exclusive rights over their innovations, preventing others from commercially using them without permission if a patent is granted. The number of applications therefore offers an indication of innovation and research activity, although not every application results in a granted patent or a marketable technology.

The increase of approximately 224,000 applications worldwide was concentrated largely in four countries: China, Japan, India and South Korea. Together, they accounted for about 94% of the net global increase, according to WIPO’s World Intellectual Property Indicators 2026 report.

China Accounts for More Than Half of Global Patent Application Increase

China contributed 113,693 additional patent applications in 2025, accounting for more than half of the worldwide increase. Inventors and applicants residing in China filed approximately 1.9 million applications globally, retaining the country’s position as the largest source of patent filings.

Global patent application
China led global patent filings in 2025, with its residents submitting approximately 1.9 million applications worldwide, followed by the United States and Japan. India also featured among the top 10 origins of patent applications. Source: WIPO, World Intellectual Property Indicators 2026, Figure A4.

The United States followed with 489,766 applications, while Japan recorded 466,301. These figures include applications filed both within the respective countries and abroad. Japan added 45,310 applications during the year, making it the second-largest contributor to global growth. India followed with an additional 28,690 applications, while South Korea contributed 23,169.

India Crosses 100,000 Patent Applications

India recorded 105,163 patent applications in 2025, marking the first time the country crossed the 100,000 threshold. The 37.5% increase also marked its seventh consecutive year of double-digit growth. The rise was driven primarily by an increase in applications filed by Indian residents, indicating growing domestic participation in patent activity. India was the third-largest contributor to the global increase, after China and Japan.

WIPO’s figures also show that applications received by India’s patent office increased by 27.8% in 2025. This marks the ninth consecutive year of growth at the office and the fourth successive year of double-digit increases. Kerala recorded 827 patent applications in 2024–25, up from 637 in the previous financial year, according to the Indian Patent Office.

However, a separate analysis by the Gulati Institute of Finance and Taxation has pointed to a gap between the state’s research spending and patent generation, raising questions about how effectively research investments are translating into intellectual property. The study also highlighted a gap between government research funding and patent output. Educational institutions received 44.81% of the state’s R&D allocation but accounted for only 7.29% of its patents.

Asia Accounts for 71.2% of Global Patent Filings

Intellectual property offices located in Asia received approximately 2.8 million patent applications in 2025, accounting for 71.2% of the global total. The region’s share has increased from 61.9% in 2015, reflecting a steady shift in global patent activity over the past decade.

China accounted for approximately 68.5% of the applications received across Asian offices. Africa was the other region, besides Asia, to record an average annual growth rate above the global average over the past decade.

Computer Technology Leads Patent Filings

Computer technology remained the largest field of patenting, accounting for 13.6% of global patent filings, according to WIPO’s technology-sector analysis. The field’s share has grown over the past decade, reflecting its expanding role in technological development.

WIPO’s report also tracks patent activity across fields such as digital communication, medical technology and electrical machinery, offering insights into areas attracting research and development activity.

Continue Reading

EduNews & Views

Woxsen University Ranks 12th Globally in MBA Entrepreneurship Ranking, Fourth in Startup Launches

Woxsen University has secured 12th place in the 2027 Poets&Quants MBA entrepreneurship ranking, with 15% of its MBA graduates from the 2021–2025 cohorts reportedly launching ventures within three months of graduation.

Published

on

Exterior view of Woxsen University's modern campus building, with the university's name displayed on the entrance sign.
Woxsen University's campus in Hyderabad, Telangana.Image credit:Woxsen university

Woxsen University in Hyderabad has been ranked 12th globally in the 2027 Best MBA Programs for Entrepreneurship ranking released by Poets&Quants, with the institution placing fourth worldwide in the category measuring startup launches by MBA graduates.

According to the university’s press release, an average of 15% of its MBA graduates from the Classes of 2021–2025 launched ventures within three months of completing their degrees. The figure contributed to the institution’s performance in a ranking that places considerable emphasis on entrepreneurial outcomes rather than relying solely on academic reputation or programme structure.

Woxsen’s overall position puts it ahead of several established business schools, including Arizona State University’s W. P. Carey School of Business, Brigham Young University’s Marriott School of Business, London Business School and Carnegie Mellon University’s Tepper School of Business.

The 2027 ranking assigns 25% weightage to the proportion of MBA graduates pursuing entrepreneurial ownership within three months of graduation. It also recognises business acquisitions as a form of entrepreneurship, broadening the scope beyond graduates establishing new ventures.

Woxsen
MBA

Students attend a classroom session at a business school.Image credit:Woxsen university

The university’s fourth-place position in the startup-launch category indicates the reported proportion of graduates moving into venture creation. However, the 15% figure represents an average across five graduating cohorts and does not, by itself, establish the long-term survival, revenue generation or employment impact of these ventures.

Woxsen first entered the Poets&Quants entrepreneurship ranking in 2026, when it featured among the global top 12 and was identified by the university as the only Asian business school in that group. Its position in the 2027 edition marks a continuation of that presence.

The ranking draws attention to a broader question facing MBA programmes: how effectively do business schools translate entrepreneurship education into actual venture creation? While business schools increasingly incorporate startup development, business planning and experiential learning into their curricula, graduate venture formation offers one measurable indicator of their outcomes.

Woxsen’s latest ranking provides one such indicator, although further information on the ventures launched, their subsequent performance and the methodology used to verify startup activity would offer a more comprehensive picture of the programme’s entrepreneurial impact.

Continue Reading

EduNews & Views

Indian universities lead Asia in research output per faculty, but the country’s top six all slip in QS ranking

Five Indian institutions are in the regional top 10 for papers per faculty member. IIT Delhi, still India’s highest-ranked university, falls from 59th to 66th.

Published

on

QS Asia University Rankings 2027
Image credit: S R Raju/Pexels

Indian universities lead Asia in research output per faculty, but all six of the country’s highest-ranked institutions have slipped in the QS Asia University Rankings 2027, with IIT Delhi falling to 66th.

All six of India’s highest-ranked universities have dropped in the QS World University Rankings: Asia 2027, released on Tuesday, even as Indian institutions recorded the strongest research output per academic in the region.

Top Six Slip in QS Asia University Rankings 2027

IIT Delhi stays at the head of the Indian list at 66th, down from 59th. IISc Bangalore is 79th (64th last year), IIT Bombay 83rd (71st), IIT Madras 85th (70th), IIT Kharagpur 90th (77th) and IIT Kanpur 97th (77th). The University of Delhi fell from 95th to 106th and left the top 100, which now holds six Indian institutions.

The declines came in a larger, more crowded table. QS Quacquarelli Symonds ranked 1,722 universities from 26 higher education systems this year, including 195 new entrants. Of India’s 325 ranked universities, 71 per cent fell, 18 per cent rose and 11 per cent held their position. Japan saw 85 per cent of its universities fall, while Taiwan and Indonesia saw 70 per cent each.

India has the second-largest number of ranked universities in Asia after Mainland China, which has 448. It added 34 this year. In 2016, only 24 Indian institutions were ranked.

Research output stands out

India has more universities in the regional top 10 for papers per faculty member than any other system. IISc Bangalore ranks first on the indicator and IIT Madras third. Twenty-seven Indian universities are in the top 50, up from 10 in 2018.

Doctoral qualifications show a similar pattern. Fifteen Indian universities share first place for the proportion of staff with a PhD, more than double the number from Iran, the next-best system. India also holds 47 of the top 100 places on that indicator.

“India has built remarkable research capacity,” said QS Senior Vice President Ben Sowter. “The challenge now is to translate this strength more consistently into research impact and international recognition.”

Those two indicators carry 5 per cent of the overall score each. The two reputation surveys carry far more: 30 per cent for academic reputation and 20 per cent for employer reputation. QS data show that 76 per cent of Indian universities lost ground on academic reputation and 83 per cent on employer reputation.

Citations and international links lag

Output has not translated into citations at the same rate. Only three Indian universities are in Asia’s top 50 for citations per paper, and one is in the top 10. India also sits below the regional average on international faculty, international students and student exchange, QS said, though the effect differs from one university to another.

Employer reputation is a relative strength. IIT Bombay is seventh in Asia on the indicator and IIT Delhi 10th, the first Indian universities to reach the top 10. Six Indian institutions are in the top 50, up from five. QS noted that this recognition is concentrated in a small group of institutions, in a country where about 45 million students were enrolled in higher education in 2023-24.

Climbers

Seven Indian universities in the regional top 200 moved up: Chandigarh University, Jawaharlal Nehru University, Lovely Professional University, Symbiosis International, Amity University, Jamia Millia Islamia and Manipal Academy of Higher Education.

Chandigarh University reached its best rank, 104th, on gains in international research, citations and papers per faculty. Manipal entered the top 200 at 195th. Further down the table, GLA University rose nearly 300 places to 491st, Jain University climbed 221 places to 374th and Woxsen University gained 195 places to reach 499th.

The wider region

The University of Hong Kong is first for a second year, followed by Peking University. Tsinghua University rose from ninth to third. The National University of Singapore fell from third to sixth and Nanyang Technological University from third to eighth.

QS pointed to India’s policy framework for the next stage. The National Education Policy 2020 placed research at the centre of reform, and the Anusandhan National Research Foundation, set up in 2024, backs efforts to widen research capacity. QS says India has built its output on national R&D spending of around 0.64 per cent of GDP.

“Deeper international research partnerships, more international faculty and students, and closer links with industry can help ensure that research produced on this scale achieves greater impact,” said Ashwin Fernandes, QS Vice President for Strategic and International Engagement.

He added: “With foreign universities opening campuses in India and Indian institutions expanding overseas, the country has an opportunity to become a major global centre for education and research.”

Continue Reading

Trending