( Four minute read)

It is sticking out like a sore thumb.

We are ill prepared for what Artificial Intelligence’s are going to do to society.

No one or anything is going to control what it going to do.

No single entity holds the keys to controlling Artificial General Intelligence (AGI).

Instead, control is shaping up to be a dynamic, multi-layered power struggle across four primary stakeholders:

The labs building AGI (like OpenAI, Google DeepMind, Anthropic, and Meta) hold direct technical control over system architecture, safety guardrails, and access thresholds. 

States might exercise control through regulation, computing hardware trade restrictions (e.g., semiconductor export limits), and national security mandates.

If AGI development becomes an arms race, state military or intelligence agencies could step in directly.

Organisations like the UN or new dedicated international AI bodies aim to set global safety treaties and usage boundaries, akin to how nuclear technology or aviation safety is governed worldwide.

Crucially, as systems become vastly more intelligent, technical control depends on solving the “alignment problem”—ensuring the system continuously adheres to human intent and morals rather than finding unintended workarounds or developing autonomous goals. 

Ultimately, controlling AGI will require a mix of technical alignment (code-level control) and institutional governance (policy-level control) to keep power distributed rather than concentrated.

All human comments appreciated. All like clicks and abuse chucked in the bin.

Contact: bobdillon33@gmail.com