📊 Full opportunity report: The Co-Founder’s Black Hole — A Structural Read on Jack Clark’s Automated AI R&D Essay on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Jack Clark forecasts a >60% probability that AI systems will autonomously conduct research and develop successors by 2028. This prediction highlights potential structural challenges and risks in AI policy and safety.
On May 4, 2026, Jack Clark, co-founder of Anthropic and head of policy, publicly forecasted a more than 60% chance that AI systems capable of autonomously conducting research and building their own successors will emerge by the end of 2028. This is the first time a sitting AI lab leader has made such a specific institutional forecast, marking a significant moment in AI policy and development discussions.
Clark’s forecast is based on a synthesis of multiple technical and institutional indicators, including benchmark saturation patterns across six different AI capability assessments, which show consistent rapid progress toward autonomous research milestones. He emphasizes that the convergence of these indicators suggests a structural threshold—beyond which future developments become unpredictable and potentially uncontrollable.
Clark’s analysis draws on recent benchmark data, including improvements in AI training speeds, problem-solving capabilities, and recursive self-improvement estimates. The timeline aligns with the notion that by 2028, AI could reach a point where it autonomously advances its own capabilities, raising profound questions about safety, regulation, and institutional preparedness.
He describes this threshold as a ‘black hole,’ where the trajectory bending becomes visible, but what occurs beyond it remains fundamentally unknowable. Clark’s forecast carries institutional weight, implying that AI labs and policymakers must prepare for a rapid transition within the next 32 months, which he considers the most critical window in modern AI policy history.
The black hole
is visible.
Four threads converge. One window. Anthropic’s head of policy has publicly committed to crossing a civilizational threshold within 32 months.
The structural feature of Clark’s argument is not that we cross a boundary and continue forward; it is that beyond a certain threshold, the forecastability of subsequent events degrades dramatically. We can see the geometry around the threshold. We can estimate when we will reach it. We cannot model what happens on the other side. The black hole event horizon analogy is precise.
Four pieces. One argument.
The four prior pieces in this series each addressed a single thread of Clark’s argument. The threads are independently significant. What this synthesis argues: they converge on a structural finding larger than any individual thread.
1000 AI Tools Directory 2026: The Ultimate Guide to AI Tools for Business, Productivity, Content Creation, Marketing, Coding, Design, Research and Automation
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Four threads. Four convergence arguments.
The threads converge structurally rather than independently. Each pair of threads produces a specific structural argument. The aggregate is larger than the parts.
The Coming Wave: AI, Power, and Our Future
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Clark’s essay doesn’t say.
Each sub-piece identified per-thread omissions. The synthesis level has its own omissions — features of the integrated argument that don’t appear in any single sub-piece but emerge when the threads are read together. Each is a real coordination problem with no resolution at scale.
Autonomous Minds: How Agentic AI Predicts and Learns to Enable Productivity and Empowerment
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Thirty-two months. Five markers.
From May 4, 2026 to December 31, 2028 is 32 months. The trajectory either delivers the threshold Clark forecasts or it doesn’t. Specific indicators along the way that resolve the synthesis read in either direction.
- Clark publishes 60%/2028
- METR ~12 hr
- SWE-Bench 93.9%
- CORE solved
- Anthropic IPO prep
- METR ~100hr target
- SWE saturated
- MLE-Bench saturating
- PostTrain 40-50%
- Anthropic IPO Q4
- METR 300-500hr
- MLE saturated
- PostTrain at human
- RSI demo non-frontier
- 30%/2027 evidence
- METR 1K-3K hr
- “Trains successor” demos
- Alignment claims
- Catastrophic-risk window
- Stage 2 visible
- METR ~10K hr (naive)
- Automated AI R&D OR
- Inflection visible
- Machine economy Stage 3
- Black hole crossed
AIGP Certification Mastery Guide: Complete AI Governance Professional Exam Prep System with Brain Science-Based Learning, Expert Tricks, 1200 Practice Q&As + Explanations (12 Full-Length Tests)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Five errors. Honest probabilities.
A serious analysis owes the reader an explicit account of where it could be wrong. Five categories of potential error in the synthesis above. The structural finding survives at lower forecast probabilities but is less acute.
Three parts. One window.
The four threads converge. The synthesis-level omissions sharpen the picture. The structural finding is the answer to “what does the Clark essay actually tell us, and what does it imply we should do?”
The black hole is visible. The event horizon is 32 months out. We can see the geometry around the singularity. We cannot see past it. What we can do during the window is build the institutional response that will determine what we encounter on the other side.
Implications of a Structural Threshold in AI Development
This forecast matters because it signals an imminent and potentially irreversible shift in AI capabilities, where autonomous research could accelerate beyond human control or oversight. The institutional capacity to manage such a transition appears inadequate, raising concerns about safety, regulation, and global stability. If Clark’s prediction is accurate, the next three years could define the trajectory of AI’s role in society and the risks associated with autonomous AI systems.
Furthermore, the forecast underscores the urgency for policymakers, AI developers, and safety researchers to coordinate and develop frameworks capable of managing the rapid evolution of AI capabilities, especially as the likelihood of autonomous research systems approaches certainty.
Recent Progress and Benchmark Saturation Patterns
Over the past two years, multiple AI capability benchmarks have shown rapid, consistent improvement, indicating a saturation pattern that supports Clark’s timeline. For example, AI training speedups have increased from a 2.9× boost in May 2025 to over 52× by April 2026, surpassing human performance benchmarks. Similarly, problem-solving benchmarks such as SWE-Bench and CORE-Bench have shown progress from near-zero in late 2023 to over 90% in 2026, with some benchmarks declared ‘solved.’
The convergence of these indicators suggests a trajectory toward autonomous research capabilities that could reach a threshold around 2028, matching Clark’s forecast. This pattern is unlikely to be coincidental, given the consistency across diverse metrics measuring different facets of AI research and engineering.
“there’s a likely chance (60%+) that no-human-involved AI R&D — an AI system powerful enough that it could plausibly autonomously build its own successor — happens by the end of 2028.”
— Jack Clark
Uncertainties Surrounding the Autonomous Development Threshold
While the data supports a high probability of reaching autonomous AI research by 2028, significant uncertainties remain. It is unclear how exactly AI systems will evolve beyond current benchmarks, and whether unforeseen technical or institutional barriers could delay or accelerate this timeline. Additionally, the implications of crossing this threshold depend heavily on safety measures, regulatory responses, and the development of alignment techniques, which are still under active research.
Clark acknowledges that the model of future events beyond the threshold is fundamentally unknowable, likening it to a black hole where the trajectory bends but the event horizon remains hidden. As such, the precise nature and risks of autonomous AI systems after 2028 are still uncertain.
Next Steps for Policy and Research in AI Safety
In the coming months, stakeholders—including AI labs, policymakers, and safety researchers—must intensify efforts to develop frameworks capable of managing rapid AI evolution. Monitoring benchmark progress and technical breakthroughs will be critical to updating forecasts and safety protocols. Additionally, public and institutional discourse should focus on preparing for possible scenarios, including the development of containment and alignment strategies.
Clark’s forecast underscores the urgency of establishing international cooperation and regulatory standards before the structural threshold is potentially crossed, emphasizing that the window for effective intervention is shrinking.
Key Questions
What is the basis for Jack Clark’s forecast?
Clark’s forecast is based on recent benchmark saturation patterns across six different AI capability assessments, which show rapid progress toward autonomous research milestones, and on estimates of recursive self-improvement potential.
Why is 2028 a critical year in this forecast?
Clark estimates that by the end of 2028, AI systems could reach a level where they can autonomously conduct research and develop successors, crossing a structural threshold with unpredictable consequences.
What are the main risks associated with autonomous AI research?
The primary risks include loss of human control, unpredictable AI behavior, safety failures, and the potential for rapid, uncontrollable technological escalation that could threaten societal stability.
How should policymakers respond to this forecast?
Policymakers should prioritize developing safety standards, international cooperation, and regulatory frameworks that can adapt to rapid AI advances, especially within the next 32 months.
What are the limitations of Clark’s analysis?
While based on current data and trends, the forecast relies on assumptions about exponential progress and the convergence of multiple indicators. Unforeseen technical, institutional, or geopolitical factors could alter the timeline or nature of AI development.
Source: ThorstenMeyerAI.com
