đź’» Part 2: How Digital Fragments of Optimization Become Abstract Agents


Details
đź’» Part 2: How Digital Fragments of Optimization Become Abstract Agents
A Deep Dive in the “Fragments to Agents” Series
In Part 1, we explored how biological and cultural systems turn scattered parts into living, social, and economic agents.
In Part 2, we shift focus to the digital and abstract realms, where fragments of optimization and computation give rise to systems that act—sometimes in ways even their designers didn’t anticipate.
This session features a substantial presentation followed by deep group discussion. We’ll unpack the central question:
How do fragments of digital optimization become abstract agents?
From machine learning systems and feedback loops to digital marketplaces and algorithmic trading—how do we draw the line between a tool and an agent? And what happens when optimization processes become powerful enough to behave as if they have goals?
We’ll explore topics and questions such as:
- When does optimization become agency?
- What distinguishes a system doing work from one with intent?
- What does mesa-optimization teach us about AI and control?
- How do abstract systems—like social networks or algorithms—begin to influence the real world with agent-like persistence?
This session draws from thinkers such as Alan Turing, Norbert Wiener, Stuart Russell, Melanie Mitchell, John von Neumann, and others in the space of AI alignment, computational emergence, and abstract systems.
This event is open to all groups, but the core emphasis will be on questions most relevant to our AI, computation, and systems-focused members.
Other lenses—including biological, philosophical, and cultural—will return in future sessions.
Format:
- Substantial presentation on core ideas
- Open, in-depth discussion and Q&A
- Optional follow-up materials and content links
Whether you’re curious about optimization, autonomy, agency, or existential risk—this session offers a deep look into one of the most urgent and fascinating transitions in the modern world.


đź’» Part 2: How Digital Fragments of Optimization Become Abstract Agents