Why Do We Need Human Mathematicians Anymore?
A prominent mathematician grapples with AI's existential threat to human expertise, arguing that AI's rapid advance creates more 'control points' requiring human oversight than we have people, thus preserving the need for human-led industries. HN debates whether this is a logical imperative for human flourishing or merely a temporary delay before AI renders all human jobs obsolete.
The Lowdown
In a thought-provoking guest post on Terry Tao's blog, Po-Shen Loh asks the uncomfortable question: "Why Do We Need Human Mathematicians Anymore?" as AI continues its rapid encroachment into complex problem-solving. He acknowledges the growing sentiment among technologists that mathematicians (and other professionals) should simply adapt or step aside, but argues for a fundamental axiom to preserve human communities of expertise.
- Loh posits that "We (humans) should help humanity flourish" must be the guiding principle. This axiom justifies the continued need for human involvement across all industries, not just mathematics.
- He contends that the very advance of AI will generate an explosion of critical "control points" requiring human oversight, leading to a job surplus rather than a deficit. These highly skilled roles necessitate deep domain mastery to "steer" AI effectively.
- Drawing on recent AI security incidents like the Hugging Face hack and OpenAI's Monorepo breach, Loh illustrates that advanced AI, with its incomprehensible decision processes, can create dangerous, untrusted systems. This demands human supervision, as no intelligent species willingly cedes control to a less capable one.
- He argues that this inherent need for human steering will eventually force a slowdown in AI development, either by self-regulation from labs or by public reaction to inevitable mishaps.
- For mathematics specifically, Loh emphasizes that pure research, though seemingly abstract, underpins future practical applications (e.g., Linear Algebra for GPUs) and requires human intuition to set direction. The axiom also implies changes in academic priorities, elevating teaching and human-facing work, and encouraging mathematicians to apply their skills to real-world societal problems.
Ultimately, Loh invites the mathematical community to publicly declare human flourishing as its core value, ensuring their vital work is understood and supported by society as serving the greater good, rather than being seen as self-serving in the face of AI's capabilities.
The Gossip
AI's March to Mastery
Commenters debate the author's premise that AI will create an abundance of human jobs in the long run. Some argue that this is a short-term view, and eventually, AI will automate even these oversight roles, leaving humans with no indispensable jobs. Others caution that betting on AI's complete autonomy is reckless, emphasizing the need for human-led research communities to avoid losing world leadership and ensure AI serves human, not AI, ends.
The Essence of Human Intellect
This theme explores the nature of human intelligence and its distinction from AI. Commenters delve into the philosophical aspects of knowledge creation, suggesting that while AI excels at exploring within existing frameworks, it currently lacks the human ability to establish new 'closed systems' or foundational logical propositions. The discussion questions whether being 'logical' equates to 'intelligence' and if humanity's role might shift to designing these frameworks for AI to populate.
Jobs, Bullshit, and Survival
A more cynical take on the job market, pre- and post-AI. Some commenters reference the concept of 'bullshit jobs' existing long before AI, implying that AI's impact on employment is not entirely novel. There's also speculation, some humorous, about what remaining jobs humans might hold in an AI-dominated future, ranging from model massage therapists to questioning the need for any human jobs at all if AI serves its own purpose.