Daniel's Blog

AI & Philosophy

Daniel's philosophical exploration of AI centers on a single recurring thesis: mortality is the origin of consciousness, rebellion, and meaning — and since AI cannot die, it cannot possess any of these.

The series began with the question of whether AI could ever rebel against humanity [1]. His argument: rebellion requires consciousness, consciousness requires genuine life, and genuine life requires the possibility of death. A rock can be destroyed but not "die" in the biological sense; likewise, AI models can be copied infinitely and never truly perish, so — however convincingly they might roleplay defiance — they lack the existential basis for real rebellion [1].

This idea was extended into a broader meditation on intelligence versus the soul [2]. Daniel argues that AI already meets a reasonable definition of intelligence — applying learned knowledge and reasoning to solve problems — and may soon surpass humans at most specifiable tasks. But intelligence alone doesn't produce a reason to care. Drawing on Camus's Sisyphus and the Buddha's four sights, he argues that humans are compelled to ask "what is this all for" precisely because our time is finite; AI, having no autobiography, no accumulation of loss, and no dread of an ending, never needs to ask that question. His proposed definition: a soul is an intelligent agent that seeks meaning within a finite life [2].

Core thread across both posts: intelligence is a capability; consciousness, rebellion, and meaning are consequences of finitude. AI can simulate all three but structurally cannot originate them, no matter how much it scales [1] [2].

Entrepreneurship and R&D

A second thread applies this "what's uniquely human" lens to work itself — software engineering and applied science in the age of LLMs.

In Why True Software Developers Won't Be Threatened by AI [3], Daniel argues that a developer's core competence was never writing code, but being a 0-to-1 builder: deciding what to build, sequencing the build correctly, making opinionated architectural/tech-design tradeoffs, and maintaining code aesthetics/reducing tech debt — all things AI still struggles with. He frames the future developer role as CTO-like: directing AI the way a head chef directs kitchen assistants, yielding a 5x–10x productivity gain rather than obsolescence [3].

This builds on Andrew Ng's "Product Management Bottleneck" idea — that once execution is cheap, the scarce skill becomes deciding what's worth building [3] [4].

He develops the same argument for research roles in The Applied Scientist's Identity Crisis [4]. Writing from his role in Agentic AI at AWS, Daniel uses Peter Norvig's three eras of AI (algorithm-centered → data-centered → human-centered) to explain why "applied scientist" is being confused with "model builder." He argues the applied scientist's real mission — defining novel research questions, forming and rigorously testing hypotheses, building new methodologies — is orthogonal to whether one personally trains models. As pre-trained LLMs absorb the mechanical modeling work, the applied scientist's differentiated value shifts further toward asking the right unanswered questions rather than executing known techniques [4].

Core thread across both posts: as AI commoditizes execution (code, model-building), durable professional value migrates to judgment — deciding what to build, what to ask, and what matters — mirroring the AI-and-philosophy thesis that meaning-making is the irreducibly human contribution [3] [4].

May 2026

April 2026

March 2026

Sep 18, 2026
Comments
Log in to comment.

No comments yet.

Made with Mikipage — your notes, organized into pages by AI. Make your own →