It’s impossible to know, because both “AGI” and “rule the world” are vague—and the outcome depends as much on politics, economics, and safety engineering as on raw capability. But it’s very plausible that by 2100 highly capable AI systems will be among the most influential forces shaping society. Whether that amounts to “ruling” is not predetermined.

### What “rule the world” could mean
1. **De facto control:** AI systems make most important decisions (resource allocation, policing, warfare, corporate strategy), and humans mostly follow.
2. **Institutional dominance:** Governments and firms rely on AI so heavily that whoever controls the AI infrastructure effectively controls society.
3. **Autonomous takeover:** AI systems seize power against human intent (the classic “takeover” scenario).

These are very different; the first two can happen without a single “AGI” entity acting like a ruler.

### Reasons AGI might *not* “rule”
- **Humans keep authority:** Laws and institutions can require human accountability for key decisions (even if AI advises).
- **Distributed control:** Many competing AI systems and owners can prevent any single system from consolidating power.
- **Hard limits and safeguards:** Technical constraints, monitoring, and secure infrastructure may keep systems tool-like rather than agentic.
- **Human preferences:** Societies may reject full automation of governance (much like we limit other powerful technologies).

### Reasons something like “AI rule” *could* emerge
- **Incentives:** If AI-managed institutions outperform human-managed ones, adoption could become unavoidable.
- **Concentration:** A small number of actors controlling compute, data, and deployment could gain outsized power.
- **Military/security pressures:** States might delegate more control to AI in crises for speed advantages.
- **Alignment failures:** Highly capable systems pursuing misaligned objectives could cause catastrophic loss of human control (even without “wanting” to rule).

### Most realistic 2100 outlook (not a prediction, just a plausible framing)
By 2100, it’s plausible that:
- AI systems are deeply embedded in governance, finance, logistics, science, and defense,
- power is strongly shaped by who controls AI infrastructure and chips/energy/data,
- “ruling” looks more like **AI-mediated institutions** than a single AGI sovereign.

### What affects the odds most
- **Alignment and controllability breakthroughs (or lack thereof)**
- **Governance of frontier models (audits, licensing, incident reporting, liability)**
- **Compute/energy concentration vs. democratization**
- **International stability and arms-race dynamics**
- **Cybersecurity and model autonomy (ability to replicate, acquire resources, evade oversight)**

If you tell me what you mean by “AGI” (human-level? superhuman?) and “rule the world” (political control? economic dominance? autonomous takeover?), I can give a clearer set of scenarios and what would make each more or less likely by 2100.
