Gap Analysis

Pages are scored by: importance × (1 + quality/100) - insightCount × 20. Higher scores indicate pages that likely contain valuable insights that haven't been extracted yet.

197Total Gaps

153No Insights

111High Priority

197Showing

All categoriescapabilitiescruxesdebatesfuture-projectionshistoryintelligence-paradigmsmetricsmodelsorganizationsotherpeopleresponsesrisksworldviews

Min importance:0

Score	Page	Imp	Qual	Insights	Category	Gap Reason
166	Intervention Portfolio 3,132 words	87	91	None	responses	High importance page with no insights
162	Scheming & Deception Detection 3,373 words	85	91	None	responses	High importance page with no insights
157	Capability Elicitation 3,467 words	82	91	None	responses	High importance page with no insights
157	AI Safety Cases 4,104 words	82	91	None	responses	High importance page with no insights
151	Sleeper Agents: Training Deceptive LLMs 1,932 words	85	78	None	risks	High importance page with no insights
149	Model Organisms of Misalignment 2,797 words	85	75	None	models	High importance page with no insights
149	AI Control 2,975 words	85	75	None	responses	High importance page with no insights
149	Sandboxing / Containment 4,346 words	78	91	None	responses	High importance page with no insights
149	Structured Access / API-Only 3,565 words	78	91	None	responses	High importance page with no insights
149	Tool-Use Restrictions 3,976 words	78	91	None	responses	High importance page with no insights
147	Leading the Future super PAC 2,558 words	85	73	None	organizations	High importance page with no insights
147	Eliciting Latent Knowledge (ELK) 2,604 words	77	91	None	responses	High importance page with no insights
147	Weak-to-Strong Generalization 2,976 words	77	91	None	responses	High importance page with no insights
147	Coordination Technologies 2,949 words	77	91	None	responses	High importance page with no insights
147	California SB 53 2,935 words	85	73	None	responses	High importance page with no insights
141	Evals & Red-teaming 2,625 words	82	72	None	responses	High importance page with no insights
138	Dangerous Capability Evaluations 3,624 words	84	64	None	responses	High importance page with no insights
138	Sparse Autoencoders (SAEs) 3,216 words	72	91	None	responses	High importance page with no insights
138	AI-Human Hybrid Systems 2,504 words	72	91	None	responses	High importance page with no insights
136	Sleeper Agent Detection 4,347 words	82	66	None	responses	High importance page with no insights
136	Mechanistic Interpretability 3,789 words	82	66	None	responses	High importance page with no insights
135	RLHF / Constitutional AI 3,034 words	83	63	None	responses	High importance page with no insights
134	Carlsmith's Six-Premise Argument 2,287 words	82	64	None	models	High importance page with no insights
134	Pause / Moratorium 2,133 words	78	72	None	responses	High importance page with no insights
130	US AI Safety Institute 4,807 words	68	91	None	organizations	High quality page under-extracted
130	New York RAISE Act 3,370 words	75	73	None	responses	High importance page with no insights
129	Alignment Evaluations 3,781 words	78	65	None	responses	High importance page with no insights
129	Evals-Based Deployment Gates 4,169 words	78	66	None	responses	High importance page with no insights
129	Scalable Oversight 1,328 words	77	68	None	responses	High importance page with no insights
128	Multi-Agent Safety 3,679 words	76	68	None	responses	High importance page with no insights
126	Responsible Scaling Policies 3,604 words	78	62	None	responses	High importance page with no insights
124	Palisade Research 2,274 words	75	65	None	organizations	High importance page with no insights
124	David Sacks 2,835 words	75	65	None	people	High importance page with no insights
124	Mechanistic Interpretability 3,678 words	78	59	None	responses	High importance page with no insights
124	Representation Engineering 1,847 words	72	72	None	responses	High importance page with no insights
124	Corrigibility Research 2,459 words	78	59	None	responses	High importance page with no insights
123	Third-Party Model Auditing 3,780 words	75	64	None	responses	High importance page with no insights
122	NIST and AI Safety 3,362 words	75	63	None	organizations	High importance page with no insights
122	Red Teaming 1,465 words	74	65	None	responses	High importance page with no insights
122	AI Safety via Debate 1,742 words	72	70	None	responses	High importance page with no insights
122	Constitutional AI 1,489 words	72	70	None	responses	High importance page with no insights
122	AI Welfare and Digital Minds 2,925 words	75	63	None	risks	High importance page with no insights
121	Circuit Breakers / Inference Interventions 3,346 words	74	64	None	responses	High importance page with no insights
120	Bletchley Declaration 2,510 words	75	60	None	responses	High importance page with no insights
119	Frontier Model Forum 3,523 words	75	58	None	organizations	High importance page with no insights
119	Marc Andreessen 3,872 words	75	58	None	people	High importance page with no insights
119	Formal Verification 2,162 words	72	65	None	responses	High importance page with no insights
119	Provably Safe AI (davidad agenda) 2,273 words	72	65	None	responses	High importance page with no insights
119	Capability Unlearning / Removal 1,744 words	72	65	None	responses	High importance page with no insights
119	Process Supervision 1,753 words	72	65	None	responses	High importance page with no insights
119	Compute Governance: AI Chips Export Controls Policy 2,631 words	75	58	None	responses	High importance page with no insights
119	Concentration of Power 1,216 words	72	65	None	risks	High importance page with no insights
118	Provable / Guaranteed Safe AI 2,502 words	72	64	None	intelligence-paradigms	High importance page with no insights
118	Long-Timelines Technical Worldview 4,668 words	62	91	None	worldviews	High quality page under-extracted
118	Optimistic Alignment Worldview 4,448 words	62	91	None	worldviews	High quality page under-extracted
117	Preference Optimization Methods 2,816 words	72	62	None	responses	High importance page with no insights
117	Refusal Training 2,886 words	72	63	None	responses	High importance page with no insights
116	Centre for Effective Altruism 2,168 words	65	78	None	organizations	Moderate importance page with no insights
116	Redwood Research 2,095 words	65	78	None	organizations	Moderate importance page with no insights
115	Dense Transformers 3,372 words	73	58	None	intelligence-paradigms	High importance page with no insights
114	Goal Misgeneralization Research 2,062 words	72	58	None	responses	High importance page with no insights
113	Heavy Scaffolding / Agentic Systems 2,795 words	72	57	None	intelligence-paradigms	High importance page with no insights
112	Authoritarian Tools Diffusion Model 6,981 words	68	65	None	models	Moderate importance page with no insights
109	Safe Superintelligence Inc (SSI) 2,855 words	75	45	None	organizations	High importance page with no insights
108	AI Alignment 3,617 words	88	91	3	responses	-
107	Council of Europe Framework Convention on Artificial Intelligence 3,050 words	65	65	None	responses	Moderate importance page with no insights
104	MATS ML Alignment Theory Scholars program 2,732 words	65	60	None	organizations	Moderate importance page with no insights
104	Output Filtering 2,650 words	64	63	None	responses	Moderate importance page with no insights
103	Agent Foundations 2,223 words	65	59	None	responses	Moderate importance page with no insights
103	Deepfake Detection 2,953 words	54	91	None	responses	High quality page under-extracted
102	Cooperative IRL (CIRL) 2,023 words	62	65	None	responses	Moderate importance page with no insights
102	Epistemic Infrastructure 2,751 words	64	59	None	responses	Moderate importance page with no insights
101	Content Authentication & Provenance 2,482 words	64	58	None	responses	Moderate importance page with no insights
101	AI-Assisted Deliberation Platforms 3,550 words	62	63	None	responses	Moderate importance page with no insights
100	Intervention Effectiveness Matrix 4,226 words	92	74	3	models	-
99	XPT (Existential Risk Persuasion Tournament) 1,974 words	64	54	None	responses	Moderate importance page with no insights
99	Erosion of Human Agency 1,865 words	52	91	None	risks	High quality page under-extracted
98	Expertise Atrophy Cascade Model 4,182 words	63	56	None	models	Moderate importance page with no insights
98	Adversarial Training 1,881 words	62	58	None	responses	Moderate importance page with no insights
97	Collective Intelligence / Coordination 2,745 words	62	56	None	intelligence-paradigms	Moderate importance page with no insights
97	Prediction Markets 1,505 words	62	56	None	responses	Moderate importance page with no insights
97	Epistemic Collapse 956 words	65	49	None	risks	Moderate importance page with no insights
96	Neuro-Symbolic Hybrid Systems 2,855 words	62	55	None	intelligence-paradigms	Moderate importance page with no insights
96	Coefficient Giving 3,618 words	62	55	None	organizations	Moderate importance page with no insights
96	Probing / Linear Probes 2,665 words	62	55	None	responses	Moderate importance page with no insights
96	Cooperative AI 2,069 words	62	55	None	responses	Moderate importance page with no insights
96	Reward Modeling 1,931 words	62	55	None	responses	Moderate importance page with no insights
96	Expertise Atrophy 972 words	58	65	None	risks	-
96	Preference Manipulation 1,021 words	62	55	None	risks	Moderate importance page with no insights
96	Trust Decline 1,583 words	62	55	None	risks	Moderate importance page with no insights
95	Government Regulation vs Industry Self-Governance 1,697 words	62	54	None	debates	Moderate importance page with no insights
95	Light Scaffolding 2,045 words	62	53	None	intelligence-paradigms	Moderate importance page with no insights
95	Novel / Unknown Approaches 3,347 words	62	53	None	intelligence-paradigms	Moderate importance page with no insights
95	World Models + Planning 2,228 words	62	54	None	intelligence-paradigms	Moderate importance page with no insights
95	ForecastBench 1,900 words	62	53	None	responses	Moderate importance page with no insights
94	AI-Augmented Forecasting 2,558 words	61	54	None	responses	Moderate importance page with no insights
94	Instrumental Convergence 5,044 words	82	64	2	risks	-
92	Intervention Timing Windows 4,408 words	87	75	3	models	-
92	Model Specifications 2,740 words	61	50	None	responses	Moderate importance page with no insights
91	Surveillance Chilling Effects Model 2,298 words	62	46	None	models	Moderate importance page with no insights
90	Sparse / MoE Transformers 2,200 words	62	45	None	intelligence-paradigms	Moderate importance page with no insights
90	Pause AI 2,355 words	60	-	None	organizations	Moderate importance page with no insights
89	Deceptive Alignment 2,097 words	85	75	3	risks	-
84	Forecasting Research Institute 3,860 words	54	55	None	organizations	-
83	State-Space Models / Mamba 3,478 words	54	54	None	intelligence-paradigms	-
83	The Sequences by Eliezer Yudkowsky 2,204 words	50	65	None	organizations	-
83	Secure AI Project 1,691 words	55	-	None	organizations	-
81	Trust Erosion Dynamics Model 2,509 words	52	56	None	models	-
81	Long-Term Future Fund (LTFF) 4,782 words	52	56	None	organizations	-
80	Solution Cruxes 3,594 words	82	71	3	cruxes	-
80	OpenAI 2,036 words	55	46	None	organizations	-
79	Public Opinion & Awareness 2,560 words	52	52	None	metrics	-
79	SecureBio 1,394 words	48	65	None	organizations	-
78	Corrigibility Failure Pathways 1,937 words	82	68	3	models	-
78	Deepfakes 1,487 words	52	50	None	risks	-
77	Bioweapons 10,641 words	82	91	4	risks	-
76	Technical AI Safety Research 3,860 words	82	66	3	responses	-
76	AI Governance and Policy 2,999 words	82	66	3	responses	-
75	AI Futures Project 2,383 words	50	-	None	organizations	-
75	CSER (Centre for the Study of Existential Risk) 2,315 words	50	-	None	organizations	-
74	Aligned AGI - The Good Ending 4,983 words	48	54	None	future-projections	-
74	Responsible Scaling Policies 4,586 words	82	64	3	responses	-
72	When Will AGI Arrive? 1,040 words	54	33	None	debates	-
72	Meta AI (FAIR) 4,370 words	49	47	None	organizations	-
71	Warning Signs Model 3,495 words	78	68	3	models	-
71	Safety Research Allocation Model 1,754 words	78	68	3	models	-
71	Power-Seeking Emergence Conditions Model 2,267 words	78	68	3	models	-
71	Reward Hacking 4,014 words	79	91	4	risks	-
70	AI Uplift Assessment Model 4,452 words	76	71	3	models	-
70	Survival and Flourishing Fund (SFF) 4,854 words	44	59	None	organizations	-
70	Texas TRAIGA Responsible AI Governance Act 2,676 words	45	55	None	responses	-
69	Voluntary Industry Commitments 4,661 words	78	91	4	responses	-
68	Global Partnership on Artificial Intelligence (GPAI) 2,977 words	45	50	None	organizations	-
67	Critical Uncertainties Model 2,538 words	85	73	4	models	-
67	Reality Fragmentation 767 words	52	28	None	risks	-
65	Neuromorphic Hardware 4,475 words	42	55	None	intelligence-paradigms	-
65	X Community Notes 1,789 words	42	54	None	responses	-
65	Treacherous Turn 4,016 words	87	67	4	risks	-
64	The Case FOR AI Existential Risk 6,597 words	87	66	4	debates	-
64	Minimal Scaffolding 2,515 words	42	52	None	intelligence-paradigms	-
64	Jaan Tallinn 5,386 words	42	53	None	people	-
63	Deep Learning Revolution (2012-2020) 3,066 words	44	44	None	history	-
63	Epoch AI 4,591 words	42	51	None	organizations	-
63	Metaculus 4,604 words	42	50	None	organizations	-
63	Scheming 5,170 words	82	74	4	risks	-
62	Economic & Labor Metrics 2,934 words	42	48	None	metrics	-
62	Cyber Psychosis Cascade Model 2,601 words	44	41	None	models	-
62	Fraud Sophistication Curve Model 3,551 words	42	47	None	models	-
62	AI-Powered Fraud 1,345 words	42	47	None	risks	-
61	Future of Life Institute (FLI) 6,007 words	42	46	None	organizations	-
61	80,000 Hours 3,837 words	42	45	None	organizations	-
61	AI Evaluation 1,739 words	82	72	4	responses	-
61	Economic Disruption 1,783 words	43	42	None	risks	-
61	Racing Dynamics 2,733 words	82	72	4	risks	-
60	Is Scaling All You Need? 1,559 words	42	42	None	debates	-
60	Mainstream Era (2020-Present) 4,284 words	42	42	None	history	-
60	AI Proliferation Risk Model 1,859 words	72	66	3	models	-
60	Disinformation Detection Arms Race Model 2,716 words	42	42	None	models	-
60	Public Opinion Evolution Model 2,851 words	42	44	None	models	-
60	Capabilities-to-Safety Pipeline Model 2,526 words	82	71	4	models	-
60	Post-Incident Recovery Model 1,938 words	42	42	None	models	-
60	Microsoft AI 4,719 words	42	44	None	organizations	-
60	CAIS (Center for AI Safety) 837 words	42	42	None	organizations	-
60	Policy Effectiveness Assessment 4,007 words	82	71	4	responses	-
60	Historical Revisionism 1,277 words	42	43	None	risks	-
60	Legal Evidence Crisis 1,119 words	42	43	None	risks	-
59	Biological / Organoid Computing 2,656 words	38	54	None	intelligence-paradigms	-
59	Racing Dynamics Impact Model 1,659 words	84	66	4	models	-
59	Defense in Depth Model 1,636 words	82	70	4	models	-
59	Goal Misgeneralization Probability Model 1,747 words	82	69	4	models	-
59	Alignment Robustness Trajectory 2,282 words	82	69	4	models	-
59	Demis Hassabis 3,170 words	41	45	None	people	-
59	AI Forecasting Benchmark Tournament 1,700 words	42	41	None	responses	-
58	Risk Interaction Network 1,913 words	82	68	4	models	-
58	Preference Manipulation Drift Model 1,974 words	42	37	None	models	-
58	Cyber Psychosis & AI-Induced Psychological Harm 938 words	42	37	None	risks	-
57	Sharp Left Turn 4,309 words	81	69	4	risks	-
57	Multipolar Trap 3,880 words	82	91	5	risks	-
55	Reasoning and Planning 4,895 words	82	65	4	capabilities	-
55	Scheming Likelihood Assessment 1,522 words	83	63	4	models	-
55	Reality Fragmentation Network Model 1,804 words	42	31	None	models	-
54	Institutional Adaptation Speed Model 3,216 words	82	63	4	models	-
54	Longview Philanthropy 3,481 words	37	45	None	organizations	-
54	Manifund 3,845 words	36	50	None	organizations	-
54	AI-Assisted Alignment 1,985 words	82	63	4	responses	-
54	Squiggle 1,850 words	38	41	None	responses	-
53	Hardware-Enabled Governance 3,562 words	78	70	4	responses	-
52	Autonomous Coding 2,489 words	81	63	4	capabilities	-
52	Mesa-Optimization Risk Analysis 1,716 words	78	69	4	models	-
52	CHAI (Center for Human-Compatible AI) 1,248 words	38	37	None	organizations	-
52	Dustin Moskovitz 4,012 words	35	49	None	people	-
52	Compute Monitoring 4,511 words	78	69	4	responses	-
52	AI Safety Institutes 4,247 words	78	69	4	responses	-
52	Parameter Table 29 words	42	24	None	other	-
51	Large Language Models 2,828 words	82	60	4	capabilities	-
51	Future of Humanity Institute (FHI) 4,200 words	34	51	None	organizations	-
51	Corporate Responses 1,405 words	78	68	4	responses	-