On the Title: Duex Ex Machina
The title Duex Ex Machina is a reference to a phrase used in the TV series Person of Interest (2014).
In the series, it refers to a form of General AI (GAI) — a machine endowed with near-omnipotent intelligence — and is translated into Japanese as “a god made by machines.”
The phrase itself originates from ancient Greek theater.
Deus ex machina described a god who was lowered onto the stage by mechanical means to resolve an otherwise unsolvable situation.
It represented a break from causality, an abrupt and almost unfair resolution — what might be called a “plot twist” or “miraculous leap” today.
Can AI Become a “Machine-Made God”?
Since the term Artificial Intelligence was coined in 1956, countless techniques and theories have been proposed.
In fiction, General AI is often portrayed as already realized.
In reality, however, we are far from achieving it — we do not even have a clearly defined end goal.
What we commonly call “AI” today is not a universal intelligence.
It is a collection of highly constrained, task-specific systems — powerful, but fundamentally limited.
The idea of a machine-made omniscient intelligence remains aspirational, not factual.
Human Limits and Externalized Knowledge
To build something resembling a general intelligence, one would need to stand on the accumulated knowledge of countless predecessors, and then combine it with original hypotheses and new perspectives.
Yet human memory itself is limited.
Recent cognitive science suggests that the human brain’s effective capacity may be equivalent to only about 1 GB of memory.
Much of what we believe we “know” is actually an illusion — a phenomenon often referred to as the illusion of explanatory depth.
As a result, both humans and AI systems are forced to rely on externalized knowledge:
how information is stored, structured, retrieved, and reused becomes a central design problem.
Purpose of This Blog
This blog exists to systematically record the technical foundations required to approach a machine-made general intelligence, including:
-
Programming languages
-
Distributed systems and ICT
-
Artificial intelligence and machine learning
-
Databases and data modeling
-
Decision systems and DX
-
Multi-agent architectures
Rather than presenting polished conclusions, the focus is on thinking, experimentation, and design traces — how ideas evolve through practice.
At the same time, this blog is not meant to be purely academic.
It also serves as a space to share insights casually, through everyday observations, side notes, and conversations.
Relationship to GitHub
In parallel with this blog, the same author publishes design notes and system architectures on GitHub under the name:
masao-watanabe-ai
GitHub:
https://github.com/masao-watanabe-ai
The GitHub repositories focus on more structured representations of ideas, including:
-
AI system architectures
-
Decision pipelines
-
Multi-agent orchestration
-
Design principles to prevent PoC failure in real organizations
Closing Note
This blog and the GitHub repositories together represent an ongoing exploration of:
AI systems, decision design, and multi-agent architectures,
grounded in real-world implementation and reflection.
They are not written to glorify a mythical “machine god,”
but to treat it as a design problem — something to be decomposed, questioned, and incrementally constructed.