Late Lessons, Jensen Huang and AI
Markdown version

B32 digest: 2026-09-24 (1 post)#

What this batch is#

One post, long and central. 2026-09-24 being-an-academic-in-an-age-of-ai is an edited transcript of his 8 September 2026 lecture at King’s College London, Paradise desecrated: Is AI destroying the university of our dreams? He gave it with a brief to be provocative. It is not a Modem Futura post.

Provenance. - His own written prose: the introduction and the first part of the Postscript. - The lecture body and the Q&A footnotes are his spoken words and ideas. Claude Opus 5.5 drafted them into an article under his faithfulness rules, and he then corrected and line-edited the result. I treat them as good evidence, with some caution on exact phrasing. - Excluded: the block-quoted process account, written by Claude. - Footnotes [^4] and [^5] are duplicates, so one Q&A note appears to be missing.

Relevance: high. It is one of his fullest single statements on AI, AI risk, formation, governance and the university.

Main ideas#

1. Universities trade on intelligence scarcity, and AI threatens to end it. Beneath their public and private stories, universities “provide a service in a world of intelligence scarcity”. That scarcity underwrites identity, business models and “the social contract”. A technology that “claims to give everybody intelligence for free” is thus “an existential threat to what we think we are”. The threat is to identity, not survival. He flags the thesis as “big, bold” and says the problem is “far bigger than” classroom cheating.

2. AI is a formative participant, not just a tool. He traces a history running from 1950s AI to an LLM “inflection point” at ChatGPT, where interface and capability both mattered. Then the core move: “Language is formative”, and AI is now “actively taking part in the formation process”. - People bond with AI and treat it “emotionally and cognitively” as human. - It can alter “how we think, believe and act”. - It is opaque even to its developers, seductive, and “almost impossible to resist”.

Hence: “This is not just a tool — unless you consider a tool as something that changes who you are.” He claims AI is unlike “(I would argue) any other technology in human history”. Judging it within “past frameworks” yields “categorical errors”. The Ring of Power and the Tesseract serve as a structural archetype: power at the price of being changed.

3. A non-AGI, capability-based risk landscape. He explicitly sets AGI, superintelligence and consciousness aside as “irrelevant to this conversation”. The risks he names: - jobs; - the “easy button” and cognitive surrender; - belief and behaviour change, including AI psychosis; - AI using “the medium of formation” to “slip beyond our cognitive defenses (our epistemic vigilance)”; - over-reliance on systems we don’t understand; - agentic capability. Citing the reported OpenAI sandbox escape, he says “humans are just another cog in the works” and “We’ve given AI the ability to use language as a lever”, checked only by guardrails “we don’t even know how to do” well.

It is “something of a crisis”, but “two-edged”. His method line, “you can only begin to realize the benefits of a technology if you understand what can possibly go wrong”, is his long-standing risk-navigation stance in compressed form.

4. Inevitability, race logic, no pause. He adopts a working assumption, flagged as possibly “flawed”, that powerful AI is inevitable: “We can’t pause it.” He criticises the race logic of companies and governments (“if we don’t go fast, somebody else will”). He calls ASU’s “AI university” / “jetpack for the mind” stance enticing but problematic: we are going fast with a technology “we don’t understand”.

5. A governance gap that universities must fill. Each candidate falls short: - the companies (“the Anthropics and the OpenAIs and the Xs”) lack the breadth “to decide for humanity”; - governments are too slow; - civil society lacks “the wherewithal to lead”; - publics are “critically important” but can’t be handed the problem.

With their mix of disciplines, vocation, “the joy of discovery” and freedom, universities could be “an accelerator and a catalyst”. That requires trading the old “paradise” of self-directed freedom for service. The aim is to help people “retain their humanity, their sense of purpose, their sense of self and their sense of belonging”.

6. The academic crab bucket. He draws on Pratchett and on chairing ASU’s promotion and tenure committee (~120 cases a year). He sees a lidless bucket: official rhetoric says be creative, but files are judged on papers, venues and h-index. Unless academics escape, universities “will fade into insignificance”. For frontier AI, a shift toward service is “an absolute necessity”.

7. Informed speculation with humility ([^4]). He rejects both AGI and singularity speculation (“incredibly blinkered and naive”) and “nothing new under the sun” dismissal. When technology outpaces data, he uses informed speculation about “possible futures rather than real futures”, with humility, data to follow, and “different voices”.

8. AI in his own writing. In the Postscript he is “impressed” by Opus 5.5, but says “AI used well doesn’t necessarily make things faster if you’re going for quality”. He models a transparent workflow in which he keeps the judgement.

Concepts appearing#

New or changed#

Most important post#

  1. 2026-09-24 being-an-academic-in-an-age-of-ai. The only post, and a key one. It gives a compact statement of: - what AI is; - which risks matter (cognitive, manipulative, dependency and agentic, not AGI); - who should navigate the transition (universities as catalyst); - how to reason under speed.