For centuries, knowledge and expertise have served not only as aspirational ideals but as economic pillars. From teachers and translators to university professors and research scientists, those who dedicated their lives to learning supported themselves by selling what they knew. They trained apprentices, mentored students, authored books, translated foundational texts — and in doing so, earned the time and stability to think deeply, explore freely, and expand the frontiers of understanding.
Today, generative artificial intelligence is eroding the very foundations of this quiet but essential economy.
The driving force behind the current AI revolution is not human insight or pedagogical care — it is scale. Generative AI feeds on vast datasets and thrives on its remarkable capacity to mimic. It predicts, with algorithmic precision, the most probable — and to be more precise the most popular — answer to a given query, packaging the result into text, sound, or image. What it delivers is often coherent and persuasive, but behind that sheen lies the illusion of comprehension. These outputs are echoes, not discoveries — simulations of understanding rather than understanding itself.
Having demonstrated its capabilities in content generation, AI is now turning its attention to education and research — particularly at the basic and intermediate levels. The implicit promise, now widely accepted in some circles, is that learning is being democratized at some levels, even automated — that traditional instruction is redundant in an age of intelligent machines. If a chatbot can emulate expertise, why pay for a course?
But here lies the paradox: simulation requires an original to simulate.
The Financial Times recently reported a sharp decline in investment in major online learning edtech platforms such as Coursera and edX — companies that saw unprecedented growth during the COVID-19 lockdowns. Meanwhile, platforms like DataCamp and Udemy are thriving, but their focus has shifted: they’re training users to use AI tools, not to master underlying disciplines. The appetite for deep, domain-specific learning appears to be waning (+ , +).
At the same time, the CEO of Cloudflare, a key player in the internet’s infrastructure, has warned that AI is dismantling the economic logic of the web. Search engines and AI models, he notes, increasingly bypass content creators by serving users direct answers — with minimal traffic routed back to original sources. Google, once a reliable traffic driver, now sends just 1 visitor for every 6 pages it consumes. For OpenAI, that ratio is an astonishing 1 in 250. In this new landscape, high-quality content — the lifeblood of learning — becomes a cost without reward (+). And if creators no longer have the incentive or resources to produce, what will remain for the machines to learn from?
It is a cruel irony. AI is gradually cannibalizing the very scholars, writers, and institutions whose intellectual labor it mimics. The intricate, long-nurtured ecosystems of teaching, research, and peer exchange are being hollowed out. Without systems that create new knowledge, generative AI risks becoming a hall of mirrors, endlessly generating and training on its own recycled reflections.
Of course, this dynamic didn’t begin with AI. Even in the pre-AI era, creators were often eclipsed by content creators. A novelist or academic might spend years crafting an original work, only to find themselves overshadowed by influencers, aggregators, and content marketers who reframed their ideas for clicks and engagement. In response, some media companies began producing their own research, films, and music — not to advance culture, but to retain attention and protect and increase profit margins.
But can a streaming platform that dabbles in filmmaking ever rival a publishing house whose mission is to preserve and propagate knowledge across generations? The cultural products emerging from these corporate content labs are still too new to be judged beyond a sudden surge in popularity. Whether they endure — or fade as commercial ephemera — remains to be seen. A “super-app” producing cinema or music is still a business experiment, not a cultural entity.
This pattern is familiar in the world of tech startups: capture user attention first, then vertically integrate. A food delivery app may begin by partnering with restaurants but eventually opens its own ghost kitchens, develops private food labels, and takes over the entire value chain. But with each move upstream, the question becomes: what is actually being added? What original value is being created?
AI faces a similar challenge. Having disrupted the producers of original knowledge, it may now be forced to become one — building in-house research teams, commissioning studies, and generating first-hand insights to sustain its growth. Without a vibrant community of human experts to emulate, AI risks intellectual stagnation, endlessly regurgitating derivative content.
The critical question is whether we can develop a new equilibrium — one that retains the vitality of human knowledge creation while responsibly integrating machine capabilities. Can we design incentives and institutions that protect the upstream flow of new ideas? Or will AI, in its race toward dominance, fulfill the warning captured in Sa’di’s verse:
یکی بر سر شاخ بن میبرید
One cuts the branch upon which they sit
Leave a Reply