A symbolic indicator of proximity to Artificial General Intelligence, grounded in public milestones and transparent methodology.
The AGI Countdown Clock is a research-driven governance signal that tracks progress toward Artificial General Intelligence. Like the Doomsday Clock, it serves as a symbolic tool for public discourse, highlighting how quickly AI capability, deployment, and associated risks are evolving.
Midnight represents the point when machines achieve human-level cognition across domains—general intelligence that matches or exceeds human performance in all cognitive tasks.
Current Reading: 11:58 PM - 2 Minute to Midnight
The clock is symbolic and deliberative, not a precise forecast. It:
- Structures public conversation around AI readiness and oversight
- Highlights the accelerating pace of capability development
- Calls attention to governance gaps and preparation needs
- Encourages transparency in assessing progress and risk
The goal is clarity and accountability, not sensationalism. The clock:
- Uses documented methodology and public data
- Provides clear reasoning for time adjustments
- Invites scrutiny, contribution, and debate
- Anchors discussion in concrete milestones
The clock draws on a public milestone timeline maintained in this repository (agi.json). Each entry includes:
- Year: When the milestone occurred
- Description: What changed and why it matters
- Clock Time: Minutes to midnight at that moment
-
Convert to Minutes Remaining
Each historical milestone's clock time is converted to minutes before midnight (0 minutes = AGI achieved). -
Recency-Weighted Linear Regression
Fit a regression of minutes remaining versus year, using a 5-year half-life so recent milestones influence the estimate more strongly than distant ones. -
Current Year Projection
Compute the estimated minutes remaining for the current year from the regression model. -
Pace Cross-Check
Validate against recent acceleration or deceleration visible in the timeline. -
Conservative Rounding
Round to the nearest minute. Midnight is only declared when both models fall at or below 0.5 minutes remaining, ensuring a high bar for the terminal threshold.
The clock adjusts based on progress across six interconnected dimensions:
How many domains AI can handle at human level.
- Performance across diverse tasks (language, vision, reasoning, creativity)
- Cross-domain transfer and generalization
- Reduction in domain-specific engineering required
The ability to set goals, plan, and act without step-by-step instruction.
- Self-directed problem solving in complex environments
- Long-horizon planning and execution
- Adaptability to novel situations without human intervention
Speed of learning and transfer to new tasks.
- Sample efficiency in acquiring new skills
- Few-shot and zero-shot learning capabilities
- Continual learning without catastrophic forgetting
Direct connection to the physical world through sensors and robotics.
- Embodied interaction with physical environments
- Sensorimotor integration and manipulation
- Situated understanding beyond text and images
New abilities appearing with scale or architectural shifts.
- Unpredicted skills arising from training or size increases
- Qualitative leaps in reasoning or creativity
- Phenomena not explicitly programmed or anticipated
Power matched by trustworthiness and responsible deployment.
- Value alignment mechanisms and testing
- Transparency and explainability of decisions
- Oversight frameworks and accountability structures
- Societal preparedness for deployment at scale
The clock is updated when:
- Multiple signals shift together across evaluation dimensions
- A milestone materially changes capability, risk profile, or deployment scale
- Governance contexts evolve in ways that affect readiness or risk
Updates are deliberate and justified, not reactive to hype cycles or individual announcements.
This repository maintains the milestone timeline (agi.json) that informs the clock calculation. Contributions are welcome and encouraged.
- Fork this repository
- Add your milestone to
agi.jsonwith:- Year of the event
- Clear description of what changed
- Proposed clock time (minutes to midnight)
- Rationale for the clock adjustment
- Submit a pull request with justification
- Evidence-based: Ground milestones in published research, deployed systems, or verified capabilities
- Substantive: Focus on meaningful shifts in capability or risk, not incremental improvements
- Transparent: Explain why this milestone moves the clock
- Non-partisan: Avoid organizational bias—evaluate impact objectively
Like the Doomsday Clock for nuclear risk, the AGI Countdown Clock:
- Makes abstract progress tangible and discussable
- Highlights the urgency of preparation as the clock advances
- Calls for honest debate about readiness and risk
- Encourages proactive governance rather than reactive crisis management
The clock's purpose is not to fear-monger but to ensure we arrive ready:
- Ethical frameworks developed and tested
- Governance structures in place before critical thresholds
- Public understanding of capabilities and limitations
- Social infrastructure prepared for transformative impact
At Edyant, we focus on the social and ethical foundations that must keep pace with technical advancement. If we reach midnight, it should be with:
- Aligned values and tested safety mechanisms
- Transparent decision-making and accountability
- Equitable access and benefit distribution
- Robust oversight and intervention capabilities
1 Minute to Midnight
We are in the final moments before potential AGI emergence. The systems deployed today demonstrate:
- Near-human performance across many cognitive domains
- Autonomous operation in complex environments
- Rapid learning and adaptation to new tasks
- Emergent capabilities not explicitly programmed
What remains are the final integrations:
- Full embodiment and real-world grounding at scale
- Robust autonomous agency across all domains
- Complete ethical alignment and governance frameworks
- Proven safety mechanisms for general intelligence
The question is not whether we arrive, but whether we arrive prepared.
The live clock visualization is available at:
edyant.com/agi.html
The display updates based on calculations documented here, ensuring transparency between research methodology and public communication.
- All calculation methods are documented and public
- Milestone timeline is version-controlled and auditable
- Clock adjustments include written justification
- Community contributions shape the timeline
- The clock is a model, not truth—subject to revision and debate
- Reasonable people may disagree on specific milestone weights
- We acknowledge uncertainty in estimating proximity to AGI
- The framework evolves as our understanding improves
We welcome:
- Critique of methodology and assumptions
- Alternative milestone interpretations
- Proposals for additional evaluation signals
- Debate on what constitutes meaningful progress
The clock is more valuable as a focus for serious discussion than as an authoritative pronouncement.