..a world- and society-kindly, balanced, moderate, nuanced statementfrom an authentic scholar [..transcript from the ever-curious if not psycho-spiritually lost New Statesman our readers will find valuable]
Terence Tao is a man who understands mathematics. He has a Fields medal – often called the Nobel prize of mathematics – to name just one of his many accolades. So, when he says that OpenAI’s recent blockbuster mathematical breakthrough is indicative of how big tech companies are negatively affecting the field, people are likely to listen.
Tao says the way in which discoveries are being made and dumped on the mathematical community may actually harm mathematics.
There has always been competition among mathematicians to be first, he says, which was fine because mathematics was so difficult that it created a natural brake to stop things getting out of hand. “Now, there’s no speed limit, and suddenly things are breaking down,” he says.
Tao fears that new findings will keep landing at an increasing pace and that there will be no time to absorb the results, fully understand them, put them into textbooks and teach them to students. This means the staggering ability of AI to bring us new mathematics could actually harm the field, instead of boosting it.
Mathematics needs to overcome five hurdles in order to beome useful, he says. First, it needs to be created, then checked, explained, accepted and finally digested into the corpus of work that is taught to the next generation. He says AI firms are only concerned with the first two.
These companies are dumping carcasses of raw meat onto our communal village table and saying: ‘Here you go, I solved your food problem.’ And then they just leave.
As an example, he points to the situation surrounding Tristan Buckmaster at New York University and Levent Alpoge at Ai company Anthropic.. They had been “working slowly on some very nice results” around the Navier-Stokes e quations says Tao, and had developed a solution to the Euler equations, a close cousin of Navier-Stokes that would act as a stepping stone to solving them. But they say they were suddenly forced to rush out papers – which mathematicians told New Scientist lacked the clarity they would normally accept – once they found out that OpenAI only began working on the Navier-Stokes problem once it heard rumours of Buckmaster and Alpoge’s work.
Tao says that what has developed is a race that benefits nobody (except perhaps the marketing departments of AI companies), erodes clarity and prioritises who is first over who is accurate, clear and concise.
“This is not how science is supposed to work,” says Tao. “They’re focusing on what traditionally has been the most prestigious part of mathematics. but then there’s sort of the lower-prestige work of explaining it, communicating it, teaching it, refereeing it for journals – all that work is left for us to clean up, and it’s demoralising.”
Increased competition wil also jeopardise the open, sharing culture that exists in mathematics and push research underground, he says, for fear of others swooping in on their ideas and completing them before they can publish.
It used to be that, if you announced you were working on a problem, there were maybe only a dozen people in the world who could plausibly outcompete you, and you knew all of them, you were on a first-name basis. There was no incentive to screw over your colleagues because they would screw you back,’ says Tao. Knowing that someone’s working on a problem is now valuable information.”
Despite his pessimism, Tao believes that mathematicians will eventually do the hard work of understanding, categorising and teaching AI company output.
“It will eventually end up in textbooks, but it’s just a completely sordid way to get there,” he says.
MS 19 September 2026 l New Scientist
Courtesy Billionaire Monkeys with Typewriters and the now famous @Xcomms account since 2012 @greeneconomyact. Also JohnBlundell3 @Substack for Philosophy AND Science emphasis amongst the bewildered & confused of consciousness and the actual purposes & all-interlocking systems of Life on Earth
Directionless under the leadership of Science-triumphalist/ ‘Perma-war’/ Landlord Class & Trades Union movement multi-millionaire ENVIRONMENTALLY BLIND politicians effective vassal state of the United States of America Australia
Invoking AI’s harm to critical thinking is a weak defence for universities
Calls to ditch AI because it is destroying students’ analytical skills ignore institutions’ questionable record at developing complex reasoning, says Ian Richardson
A historical pattern has emerged in higher education when it comes to new technology. Something arrives. Institutions recoil. Arguments are voiced about standards, about the sacred nature of knowledge, about the irreplaceable human element of learning. In time, the technology is accepted and the arguments are quietly retired. Until the next thing arrives – when the same reassuringly familiar laments can be aired once again.
It’s interesting to question where we might have been if only we’d managed to resist the tide of technological “progress”. The printing press wouldn’t have debased the authority of texts by disintermediating venerable and learned hands. The photocopier wouldn’t have allowed accumulation of vast quantities of text without proper engagement. Scientific calculators wouldn’t have hollowed out numerical intuition. Searchable databases wouldn’t have stripped away contextual judgement by preventing us from spending years thumbing card indexes and scouring dog-eared journals. Word processors wouldn’t have produced sprawling, undisciplined prose. The internet wouldn’t have degraded research quality and distracted lazy students with a swamp of unreliable content.
Each generation of educators has convinced itself that new technology poses a unique threat to authentic learning, intellectual rigour and human connection – only for those fears to fade as the technology becomes normalised. It’s a pattern that reveals a principled and cautious engagement with technological change and, more discreetly, a near-perfect record of confusing intellectual virtue with institutional and parochial self-interest.
In a Guardian article published on 10 March, a dozen humanities professors describe mostly despairing experiences at the arrival of artificial intelligence in their classrooms. “It’s driving so many of us up the wall,” confided one. Another described it as “the bane of [her] existence” and wished she could “push ChatGPT off a cliff”. Elsewhere, the question of what it would do for us as a species reflected broader existential concerns at AI’s emergence ..
These are not fringe voices. They’re senior academics at Stanford, Berkeley and Penn State, among others. And the argument they’re making is, at its heart, a straightforward one: AI threatens critical thinking, and critical thinking is what universities exist to cultivate.
It’s a powerful argument, and one that has become a rallying cry within the sector. It’s also, when held up to even modest scrutiny, one that many institutions are making with little support beyond fine words and good intentions.
It’s 15 years since Richard Arum and Josipa Roksa’s Academically Adrift was dropped on an unsuspecting sector, highlighting the roughly 45 per cent of US undergraduates who demonstrated no significant improvement in critical thinking, complex reasoning or writing skills during their first two years of college – findings that were later extended to all years of college life and, the authors claimed, left graduates poorly prepared for workforce transition.
Not everyone agreed, of course. Some suggested flawed methods could account for the findings, while others pointed to differences between disciplines. Elsewhere, studies were produced showing higher levels of gain. Nevertheless, the central criticism landed, and the sector was clearly on the defensive. Outcome-based learning approaches were embedded, writing across the curriculum programmes were expanded, “high-impact practices” were formalised, and the Collegiate Learning Assessment (CLA/CLA+) was adopted by many as a measure of value-added learning. These developments, initiated in the US, have significantly influenced international approaches – notably through quality assurance and accreditation frameworks designed to address rising demands for accountability.
However, a decade later, the 2022 OECD report Does Higher Education Teach Students to Think Critically?, which assessed more than 120,000 students across six countries, found only a modest improvement in students’ abilities. Given “the importance that most higher education programmes attach to promoting critical thinking skills, the learning gain is smaller than could be expected”, it reported. With half of all exiting students performing at the two lowest levels of mastery, the conclusion was unequivocal: a university qualification is not a reliable measure of critical thinking expected by the global marketplace.
Despite this, critical thinking appears prominently in institutional mission statements, quality frameworks, programme descriptions and learning outcomes across the sector. But the gap between aspiration and evidenced attainment remains wide – and it’s clear that institutions have been evasive in this area when it comes to comparative measures. The OECD’s Ahelo project, for instance, was abandoned in 2015 because of insufficient institutional support – notably from elite universities. As Andreas Schleicher, director for education and skills at the OECD, said at the time, the biggest resistance had come “from those institutions that fear they will never do as well as their reputation suggests”.
So what is actually being defended in the Guardian article?
The professors interviewed are intelligent people grappling with a genuine disruption. AI does present real challenges to meaningful assessment. The case for preserving the productive difficulty of unaided thought – the struggle with a text, the construction of an argument from scratch – is defensible and important. The argument that universities must move towards metacognition, teaching students to interrogate their own reasoning rather than merely deploying it, reflects serious engagement.
But the framing – seriously? “Self-lobotomisation.” Students deliberately made “helpless” by tech companies. “Intellectual ravages.” AI threatening us “as a species”. Beneath the intellectual veneer lies an emotional texture entirely familiar from the history of technological resistance: the language of desecration, of theft, of an irreplaceable thing in jeopardy.
And what is this irreplaceable thing, exactly – unaided human cognition, or something else entirely? For many, the concerns are sincere: a genuine belief that something important about how humans learn and reason is being bypassed, and a commitment to students that deserves to be taken seriously. But sincerity and self-interest are not mutually exclusive.
Resistance to technology has never really been about technology – it is about what technology does to existing structures of power and privilege. Those structures depend on control, and technology has a habit of loosening it.
Each transformation has been resisted most fiercely by those with most to lose. AI threatens to perform functions that have long conferred authority and status on those who perform them – and it is doing so at the very moment when institutional performance is being questioned like never before.
None of this argues against taking AI’s risks seriously, but it does argue that we need to be driven more by a sense of what we’re trying to achieve than by what we’re trying to defend (not least because the thing we’re defending is on rather shaky ground). Responses to AI within the sector – for instance, oral examinations, competency-based assessment, explicit instruction in critical reasoning – are frequently presented in opposition to the technology when in fact, they all can be facilitated rather than undermined by AI. AI’s broader potentialities must be seriously explored – the flat suggestion that the technology is antithetical to the development of critical enquiry is emotional, oversimplistic nonsense.
Professors are asking their students to trust that the struggle is worth it – that the friction of unaided thought produces something no machine can replicate. They may well be right. But a sector that has spent 15 years evading serious measurement in this area, and 500 years resisting technologies that threatened its authority, cannot demand that trust on faith alone.
The critical thinking stance is available to higher education, but it has to be earned – and not by pushing technology off a cliff.
Ian Richardson is a faculty member and director of executive education at Stockholm Business School at Stockholm University. He is co-founder of the national Swedish programme AI for Executives, which seeks to drive board-level understanding and organisational adoption of AI across industries and sectors
Rather than resisted by those with mot to use another view is that the “transformation” have been most resisted by those with the greatest understanding of what is at stake. Deeply disappointing to read another advertisement masquerading as an article. If the critical thinking stand has to be earned perhaps the author could be hoist on their own petard.
#2 Submitted by oliver…. on April 15, 2026 – 7:41am
This piece is upside down. Universities are now the only source of critical thinking really, and new tech seems to be now linked to oligarchy and authoritarianism. Students and academics are right to distrust it.
#3 Submitted by m.o…. on April 15, 2026 – 10:04am
The best students strive to develop critical thinking and this is evident in their contributions to discussion, oral presentations, in dissertation supervision meetings and written work. Such students tend to be wary of AI and are are able to justify their distrust of it. Importantly, the student’s critical thinking is consistent and can be tested at any time including via impromptu questioning about a text or their own work. They are a joy to teach. AI cannot beat this.
#4 Submitted by t…. on April 15, 2026 – 2:40pm
“He is co-founder of the national Swedish programme AI for Executives” ’nuff said
#5 Submitted by … on April 15, 2026 – 3:55pm
In fairness, just because he has a role to perform, does not mean he does not believe in the values of that role. People who strongly believe in EDI for example, often take up roles in that sector, people who believe in education, become teachers.
#6 Submitted by tom…. on April 23, 2026 – 3:49am
Fair critique but what I feel is missed is that AI is not just another technological innovation. Word processing is a replacement technology, not a disrupting technology as AI is.
#7 Submitted by mlarsen… on April 24, 2026 – 1:00am
Those joining the unconstrained AI fan club seem to have lost sight of the purpose of higher education. Although there’s no doubt students will use AI professionally, I believe we have a moral and ethical obligation to ensure they retain the critical thinking and communication skills to productively participate in society and the economy. Redefining learning to match the products released by Silicon Valley is the very definition of cognitive surrender
Corrupt cartels and monopolies dominate the global economy with impunity, in part by exploiting asymmetric information to lie, cheat, steal, and deceive. Asymmetric information is the disparity in knowledge that exists among buyers and sellers in a market. For example, a used car salesman might know that a given car on the lot could be a lemon, but the buyer has no clue. A financial advisor may recommend investments to to the client that inflate commission fees, but the client may think that the advice is being given on good faith. The asymmetry of information between buyers and sellers can affect everything from prices to product quality, and since it exposes some of the inherent weaknesses and imperfections in a market economy, it has often been used as a justification for government regulations to various degrees.
But the libertarian economists Tyler Cowen and Alex Tabarrok argued in a 2015 essay that asymmetric information is becoming less relevant because of technological innovation (“The End of Asymmetric Information”). They brought up several examples to support their argument, such as the declining frequency of odometer fraud in used car sales or reputation mechanisms on Yelp and Angie’s List that allow consumers to rate sellers and businesses. Their central point was that increased technological innovation and the democratizion of knowledge from the Internet have collectively reduced the importance of asymmetric information, and therefore we don’t need additional government laws and regulations to achieve the same goal. Buyers and sellers are now apparently capable of policing each other, so we should just get the government out of the way. That’s their thesis.
They are flatly wrong. Technological innovation has massively exacerbated asymmetric information, because it’s allowed the rich and powerful to protect and amplify their advantages over rivals and competitors by deploying new methods of evasion and subterfuge. Many new technologies are not even shared or publicized. They’re concentrated in the hands of powerful elites, who exploit them to ensure their ongoing class domination of society. There are thousands of critical patents that are kept secret by the United States government because they involve sensitive technology for national security. For another example, Volkswagen fooled government regulators by controlling its diesel engines with highly sophisticated software that could quietly cheat during emissions tests.
And for a final example, large brokerage firms on Wall Street operate advanced and secure dark pool market exhanges that can facilitate block trades through high-frequency algorithms for major institutional investors, all away from public view. For these elite investors, ensuring organized secrecy and informational asymmetry is exactly the point, as they don’t want to alert other investors and the public at large about the prices involved in their deals, prices that in some cases can differ sharply from those in public exchanges like the NYSE or NASDAQ. High-frequency trading also makes it far easier to commit spoofing, the practice of placing vast numbers of fake orders that are canceled before execution, all as a way of generating artificial demand for a particular stock or asset. In 2020, JP Morgan Chase paid a fine of almost $1 billion to the federal government for carrying out widespread order spoofing in precious metals futures contracts. Although JP Morgan happened to get caught in this one case, the vast majority of spoofing incidents are never publicized or prosecuted, precisely because high-frequency trading allows sophisticated investors to game the system and get away with it.
Furthermore, most of the examples given by Cowen and Tabarrok are focused on informational disparities at the point of market exchange. But so many of the major informational asymmetries in modern capitalism are actually found upstream in the production process. For example, a massive global auto parts cartel in the 2000s had substantial downstream effects on car prices, but consumers knew absolutely nothing about it. The cartel fixed and rigged bidding prices for over 200 different types of car parts sold to manufacturing companies like Toyota and Honda. A paper in 2019 cited estimates that the cartel’s activities affected international commerce worth anywhere from $3.2 trillion to $5 trillion, with injuries and overcharges sitting somewhere between $600 billion and $1 trillion. As of 2019, over $20 billion in penalties had been imposed on over 300 companies grouped into roughly 70 cartels worldwide. Likewise, the Lysine Cartel in the 1990s impacted prices for consumer food products around the world. Lysine is a critical amino acid that goes into animal feed. In the early 1990s, the American corporate giant ADM entered into a cartel agreement with other major food companies in East Asia and elsewhere. Under the agreement, the cartel tightly controlled global lysine prices and shipping volumes, effectively carving up the world market among the various cartel participants.
Simply put, even if Cowen and Tabarrok are correct that informational asymmetries are declining at the point of market exchange, that would prove nothing about asymmetries at different points along the economic chain, especially for industrial production, logistics, and distribution. The evidence clearly shows that these asymmetries still exist and have many complex downstream effects on prices for consumer goods and services. Asymmetric information is still here and still routinely plagues markets around the world. What we need now is a fight against monopoly power, not distractions from wayward libertarians who don’t know how the world works.
GEA & Monkeys’ thanks to the @Substack writer of say 5 April 2026 so let me do an awesome resource-profligate Artificial-intelligence Tell-search on the dude..
Author of The Physics of Capitalism. PhD in Physics. Complex systems theorist exploring the nexus of economics, politics, energy networks, and ecological dynamics. Find out more about my work at www.eraldkolasi.com.
