Keyboard shortcuts

Press or to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Chapter 6: The Six Hypotheses

The Critical Questions That Define Our Future

Six binary hypotheses capture the fundamental uncertainties of the AI revolution. Each represents a fork in the road where humanity must choose—or have chosen for us—between dramatically different paths.

H1: AI Progress Trajectory

The Question

Will AI capabilities continue their exponential growth, or will we hit fundamental barriers?

Option A: Accelerating Progress (91.1% probability)

Evidence Supporting:

  • GPT-3 to GPT-4: 10x parameter increase in 2 years
  • Compute availability growing 10x every 18 months
  • $200B+ annual investment accelerating
  • Breakthrough demonstrations monthly
  • No fundamental barriers identified

What This Means:

  • Human-level performance in most cognitive tasks by 2035
  • Continuous capability surprises
  • Rapid deployment across industries
  • Accelerating societal transformation

Option B: Fundamental Barriers (8.9% probability)

Evidence Supporting:

  • Scaling laws may plateau
  • Energy constraints emerging
  • Data limitations possible
  • Regulatory restrictions growing

What This Means:

  • AI remains powerful but limited
  • Incremental improvements only
  • More time for adaptation
  • Current paradigms persist

H2: Intelligence Architecture

The Question

Will we achieve Artificial General Intelligence (AGI) or remain with narrow AI systems?

Option A: AGI Emerges (44.3% probability)

Evidence Supporting:

  • Emergent abilities in large models
  • Transfer learning improving
  • Multimodal integration advancing
  • Reasoning capabilities expanding

What This Means:

  • Single systems handle diverse tasks
  • Human-level general intelligence
  • Unprecedented capabilities
  • Existential questions arise

Option B: Narrow AI Persists (55.7% probability)

Evidence Supporting:

  • Current systems still brittle
  • True understanding absent
  • Combinatorial explosion remains
  • Domain-specific solutions dominate

What This Means:

  • Specialized systems for each domain
  • Human expertise remains valuable
  • More predictable development
  • Easier safety management

H3: Employment Dynamics

The Question

Will AI complement human workers or displace them faster than new jobs emerge?

Option A: Complementary Enhancement (25.1% probability)

Evidence Supporting:

  • Historical precedent of technology creating jobs
  • New role categories emerging
  • Human skills remain unique
  • Augmentation tools proliferating

What This Means:

  • Humans and AI work together
  • Productivity dramatically increases
  • New job categories emerge
  • Skills evolution manageable

Option B: Mass Displacement (74.9% probability)

Evidence Supporting:

  • Automation scope unprecedented
  • Cognitive tasks now vulnerable
  • Speed exceeds retraining capacity
  • Network effects concentrate gains

What This Means:

  • 21.4% net job losses by 2050
  • Structural unemployment rises
  • Social safety nets stressed
  • Economic restructuring required

H4: Safety and Control

The Question

Can we develop AI safely with proper alignment, or will significant risks materialize?

Option A: Safe Development (59.7% probability)

Evidence Supporting:

  • Alignment research progressing
  • Safety culture strengthening
  • Regulatory frameworks emerging
  • Technical solutions advancing

What This Means:

  • AI remains under human control
  • Risks identified and mitigated
  • Beneficial outcomes dominate
  • Trust in AI systems grows

Option B: Significant Risks (40.3% probability)

Evidence Supporting:

  • Control problem unsolved
  • Misalignment examples accumulating
  • Dual-use concerns growing
  • Accident potential high

What This Means:

  • Major incidents likely
  • Existential risks possible
  • Public backlash probable
  • Restrictive regulation follows

H5: Development Paradigm

The Question

Will AI development remain distributed or centralize among few powerful entities?

Option A: Distributed Development (22.1% probability)

Evidence Supporting:

  • Open source movement strong
  • Academic research continues
  • Startup ecosystem vibrant
  • International competition

What This Means:

  • Innovation from many sources
  • Competitive markets preserved
  • Democratic access possible
  • Resilient ecosystem

Option B: Centralized Control (77.9% probability)

Evidence Supporting:

  • Compute costs escalating
  • Data moats expanding
  • Network effects dominant
  • Winner-take-all dynamics

What This Means:

  • 2-5 entities control AI
  • Monopolistic tendencies
  • Power concentration extreme
  • Democratic challenges arise

H6: Governance Evolution

The Question

Can democratic institutions adapt to govern AI, or will authoritarian control emerge?

Option A: Democratic Governance (36.1% probability)

Evidence Supporting:

  • Democratic resilience historically
  • Public awareness growing
  • Regulatory efforts underway
  • Civil society engaged

What This Means:

  • Human rights preserved
  • Transparent AI governance
  • Public participation maintained
  • Individual agency protected

Option B: Authoritarian Drift (63.9% probability)

Evidence Supporting:

  • Surveillance capabilities expanding
  • Emergency powers normalizing
  • Tech-state fusion occurring
  • Democratic norms eroding

What This Means:

  • AI enables total surveillance
  • Social control mechanisms
  • Individual freedom curtailed
  • Power permanently concentrated

The Interconnected Web

These hypotheses don’t exist in isolation. Their interactions create the complex dynamics of our future:

Critical Relationships

  1. Progress → Everything: H1A (rapid progress) influences all other outcomes
  2. AGI → Displacement: H2A makes H3B (job losses) highly probable
  3. Centralization → Authoritarianism: H5B enables H6B directly
  4. Displacement → Instability: H3B threatens H6A (democracy)

Feedback Loops

  • Authoritarian-Centralization: H6B reinforces H5B and vice versa
  • Safety-Trust: H4A builds confidence, enabling positive outcomes
  • Risk-Restriction: H4B triggers regulation, slowing progress

What the Probabilities Tell Us

High Confidence Predictions (>75%)

  • AI will advance rapidly (91.1%)
  • Jobs will be displaced (74.9%)
  • Development will centralize (77.9%)

Genuine Uncertainties (40-60%)

  • AGI achievement (44.3%)
  • Safety outcomes (59.7% safe)

Warning Signals (<40%)

  • Democratic preservation (36.1%)
  • Distributed development (22.1%)
  • Job complementarity (25.1%)

The Composite Picture

Combining these probabilities reveals our most likely future:

  • Rapid AI progress continues (H1A)
  • Uncertainty about AGI (H2 mixed)
  • Significant job displacement (H3B)
  • Reasonable safety measures (H4A)
  • Centralized development (H5B)
  • Democratic erosion (H6B)

This points toward our three scenario clusters, with Adaptive Integration most likely if we act wisely, but Fragmented Disruption probable if we don’t.


Next: Causal Network Model →
Previous: Research Design ←