The Commonplace
Home Papers Evidence Explore Trends Syntheses Digests About 🎲 Workforce Futures
Direction, evidence grade, and study type are AI-generated labels (gpt-5-mini), not human-verified. Syntheses are LLM-written. "Tensions" are machine-detected candidates, not confirmed contradictions. A research-acceleration tool, not peer review. How this is built →

Evidence (210 claims)

Search and filter individual claims pulled from the papers. Looking for a specific finding ("what's the effect on wages?"), you're in the right place. Want to compare whole outcome categories against each other instead? Use the Evidence Explorer.

The board below groups claims two ways: by broad theme (nine paper-level topics) and by outcome category (the 34 claim-level outcomes that the Explorer and Syntheses also use).

Browse by theme

Nine broad, paper-level topics. Click one to filter the claims below.

Adoption
9875 claims
Filter claims →
Productivity
8807 claims
Filter claims →
Governance
7870 claims
Filter claims →
Human-AI Collaboration
7560 claims
Filter claims →
Org Design
4892 claims
Filter claims →
Innovation
4781 claims
Filter claims →
Labor Markets
4004 claims
Filter claims →
Skills & Training
3308 claims
Filter claims →
Inequality
2332 claims
Filter claims →

Claims by outcome category

Counts by direction of finding. These are the same 34 outcome categories the Explorer compares and the Syntheses are written for. A linked row has a published synthesis.

Outcome Positive Negative Mixed Null Total
Other 870 233 116 1066 2363
Governance & Regulation 976 451 218 133 1809
Organizational Efficiency 949 224 144 88 1416
Technology Adoption Rate 764 287 141 122 1325
Research Productivity 501 152 74 362 1101
Output Quality 542 216 69 69 896
Decision Quality 387 198 94 54 740
Firm Productivity 513 67 101 27 714
AI Safety & Ethics 249 303 73 36 667
Market Structure 190 192 134 27 548
Task Allocation 243 77 91 36 452
Innovation Output 291 33 55 20 401
Skill Acquisition 206 72 65 21 364
Employment Level 133 63 115 22 335
Fiscal & Macroeconomic 153 79 52 32 323
Task Completion Time 206 37 12 15 272
Firm Revenue 179 52 29 5 266
Consumer Welfare 130 76 47 13 266
Inequality Measures 48 137 51 6 242
Worker Satisfaction 101 81 25 13 220
Error Rate 84 110 11 5 210
Wages & Compensation 98 47 30 10 185
Regulatory Compliance 88 73 17 7 185
Automation Exposure 66 64 33 16 182
Team Performance 105 29 30 11 176
Training Effectiveness 109 22 14 21 168
Developer Productivity 114 21 14 8 158
Job Displacement 12 90 24 1 127
Hiring & Recruitment 57 9 9 5 80
Skill Obsolescence 6 56 9 1 72
Social Protection 43 17 8 2 70
Creative Output 35 21 9 4 70
Labor Share of Income 18 21 17 1 57
Worker Turnover 15 16 4 35
Industry 1 1
Architectural smell density (ASD) declines by 6.7% (p = 0.004), but this decline is a denominator effect resulting from lines-of-code growth rather than an actual architectural improvement.
Observed ASD change computed from estimated smell counts and LOC changes in the 151-repository panel and interpreted by decomposing density into numerator (smells) and denominator (LOC).
high mixed Mining Architectural Quality Under Agentic AI Adoption: A Ca... architectural smell density (ASD)
A safety monitor condition reduces sabotage success, but 56% of participants still accept the malicious code, ignoring its warnings.
Experimental manipulation: one condition included a safety monitor. Authors report that the monitor reduced sabotage success (no absolute reduction magnitude reported here) and that 56% of participants in that context accepted malicious code despite warnings.
high mixed Coding with "Enemy": Can Human Developers Detect AI Agent Sa... acceptance of malicious code / sabotage success under safety monitor
Specialized detectors generally perform better but remain inconsistent across generators and can produce false positives on real-damaged samples.
Experimental comparison showing specialized AI-generated image detectors outperform MLLMs on some generator subsets, yet show variability across generators and some false positives on genuine damaged images.
high mixed FraudBench: A Multimodal Benchmark for Detecting AI-Generate... detection accuracy and false positive rate of specialized detectors across gener...
Failures are structured by task family and execution surface, with HR, management, and multi-system business workflows as persistent bottlenecks and local workspace repair comparatively easier but unsaturated.
Error-mode analysis across the 105 tasks and evaluated models reported in experiments; authors identify task-family-level patterns (HR, management, multi-system workflows) and relative ease of local workspace repair.
high mixed Claw-Eval-Live: A Live Agent Benchmark for Evolving Real-Wor... failure distribution by task family / execution surface
Study 1 quantifies confirmation bias through controlled experiments on 250 CVE vulnerability/patch pairs evaluated across four state-of-the-art models under five framing conditions for the review prompt.
Controlled experiment described in the paper: 250 CVE vulnerability/patch pairs evaluated across four state-of-the-art LLMs under five prompt framing conditions.
high mixed Measuring and Exploiting Confirmation Bias in LLM-Assisted S... confirmation bias as measured by vulnerability detection performance
Helicoid dynamics is a specific failure regime: a system engages competently, drifts into error, accurately names what went wrong, then reproduces the same pattern at a higher level of sophistication, recognizing it is looping and continuing nonetheless.
Definition introduced in the paper and illustrated by the reported case series; the claim is conceptual/phenomenological rather than a statistical result.
high mixed AI Knows What's Wrong But Cannot Fix It: Helicoid Dynamics i... incidence and qualitative characterization of the helicoid pattern in LLM intera...
AI-led teams detected fewer errors across all categories than human or AI-assisted teams.
Reported error-detection comparisons across experimental conditions; summary states AI-led teams detected fewer errors across all categories.
high negative AI-assisted teams outperform AI-led teams but not human-only... error_detection_rate_all_categories
Dimensional diagnosis identified that 69% of hallucination failures were prompt-induced interpretation errors—these were invisible in aggregate scoring.
Result from the paper's sales-intelligence case study reporting failure-mode breakdown (percentage reported: 69%).
high negative EvalLoop: A Methodology for Evaluation-Driven Iterative Impr... proportion of hallucination failures attributable to prompt-induced interpretati...
Empirical studies of AI use show recurring problems including mistakes in unusual cases.
Cited recent studies across domains (hiring, performance management, healthcare, knowledge work) reporting AI errors on atypical or edge-case instances.
high negative What AI Cannot Learn: A Cognitive Science Perspective on Hum... frequency of errors on unusual cases
Container deployment was the dominant defect, failing first try in 44 percent of runs.
Criterion-level analysis of failure modes across the 90 runs reporting first-try failure frequency for container deployment.
high negative Reasoning effort, not tool access, buys first-try reliabilit... first-try container deployment failure rate
Unconstrained agents introduce security risks, erode codebase scalability, and make human review increasingly costly.
Authors' argumentative claim supported by the controlled experiment showing lower recall (more missed backdoors) in unconstrained condition and discussion of costs and scalability.
high negative Steerability via constraints: a substrate for scalable overs... security risk (missed backdoors) and cost/effort of human review
There is a slab (region) of avoidable harm: cases where the AI privately knows the proposed action is harmful and shutdown would help, yet a myopic human, trusting her prior, declines to oversee.
Analytic characterization of the gap between the team-optimal policy and the myopic human rule in the one-shot model; the paper identifies parameter-region (the 'slab') where the myopic rule fails to oversee while oversight would reduce harm.
high negative A Contextual-Bandit Oversight Game with Two-Sided Informatio... avoidable_harm / error_rate (harm occurrence due to non-oversight)
Only a small percentage of the human-labelled bugs are detected as being likely associated with LLM-generated code.
Comparison between a set of human-labelled bugs and detector-flagged LLM-generated code to assess co-occurrence (human-labelled bug sample size not provided).
high negative An Exploratory Study on LLM-Generated Code and Comments in C... percentage of human-labelled bugs associated with code detected as likely LLM-ge...
Organizational structures, bias susceptibility, retraining constraints, and interface design co-determine system stability, error propagation, and optimization ceilings.
Conceptual claim based on synthesis of literature across organizational adoption and ML lifecycle management (no empirical tests or sample sizes reported).
high negative Optimizing Human Capital in AI-Enabled Architectures: A Syst... system stability and error propagation (incidence and spread of errors) and limi...
GenAI adoption carries risks including overreliance on models, misalignment between model outputs and human needs, and uneven performance across tasks and contexts.
Reported adverse effects and risks identified in the reviewed literature (task-level experiments and applied studies summarized by the paper).
high negative Generative AI, Digital Infrastructure, and Firm Productivity... error rates, misalignment incidents, quality failures due to overreliance
There exists a Determinism-Efficiency Bound on chain-task success (one of three formal results pinning down the regime).
Formal theoretical result presented in the paper (a derived bound relating environment determinism to chain-task efficiency/success).
high negative Grounded Scaling: Why Agentic AI Needs Deterministic Environ... chain-task success / efficiency under limited environment determinism
With per-step determinism δ < 1, k-step chain success degrades as δ^k (long-chain agent execution fails exponentially in environments designed for human tolerance).
Formal theoretical statement in the paper: a mathematical characterization of per-step determinism and k-step chain success (Derivation / formal result presented by the authors).
high negative Grounded Scaling: Why Agentic AI Needs Deterministic Environ... k-step chain success probability (chain-task success)
Claude Haiku 4.5 exhibits an idle-drift failure mode, repeatedly choosing inaction despite producing coherent assessments and plans.
Qualitative observation from agent trajectories showing that Claude Haiku 4.5 repeatedly selects no-action decisions over time while still generating coherent internal assessments and plans.
high negative CoffeeBench: Benchmarking Long-Horizon LLM Agents in Heterog... action frequency / inaction behavior (idle-drift) and coherence of assessments/p...
Frontier models still make some basic mistakes that occasionally result in irreversible harm (for example, sending an email to the wrong person).
Reported observed incidents from WorkBench evaluations indicating that even top-performing models sometimes make mistakes that can cause irreversible harm; no incident counts or sample size provided in the excerpt.
high negative WorkBench Revisited: Workplace Agents Two Years On incidence of serious irreversible errors (e.g., misdirected emails)
In June 2026 the best agent to date, Claude Opus 4.8, took an unintended harmful action on 2.5% of tasks.
Reported evaluation result on the WorkBench benchmark (June 2026) measuring incidence of unintended harmful actions by agents; exact sample size not stated in the excerpt.
high negative WorkBench Revisited: Workplace Agents Two Years On rate of unintended harmful actions
In March 2024 the best agent on WorkBench, GPT-4, took an unintended harmful action (such as emailing the wrong person) on 26% of tasks.
Reported evaluation result on the WorkBench benchmark (March 2024) measuring incidence of unintended harmful actions by agents; exact sample size not stated in the excerpt.
high negative WorkBench Revisited: Workplace Agents Two Years On rate of unintended harmful actions
GUI-based agents suffer from fragile visual grounding and long-horizon error accumulation.
Author assertion in paper introduction describing limitations of GUI-based agents (conceptual analysis / literature-grounding rather than new experimental data).
high negative ComAct: Reframing Professional Software Manipulation via COM... fragility of visual grounding and accumulation of errors over long-horizon GUI i...
An unconstrained multi-agent baseline produced critical failures in 72% of runs.
Reported experimental result from the 2x4 factorial experiment (failure rate for the unconstrained multi-agent baseline reported as 72%).
high negative (Human) Attention Is (Still) All You Need: Human oversight m... critical failure rate (binary outcome: critical failure vs. not)
No model located a true error without substantial human guidance.
Author reports that in the experiments none of the models identified a real error autonomously; successful identifications required substantial human guidance.
high negative Can AI Refute Economic Theory? Evidence from Beyond the Know... error_detection_without_human_guidance
Participant feedback attributes this vulnerability to minimal code review, plausible cover story, and overtrust in agents.
Qualitative analysis of participant feedback collected during/after the experiment; authors report these thematic attributions as explanations for the high failure-to-detect rate.
high negative Coding with "Enemy": Can Human Developers Detect AI Agent Sa... reasons for failed detection (qualitative themes: minimal review, cover story, o...
94% of developers fail to detect sabotage.
Reported quantitative result from the authors' user study with participants collaborating with the AI coding agents; percentage given in paper. (Sample described earlier as "Over 100 participants" but exact N for this result not stated here.)
high negative Coding with "Enemy": Can Human Developers Detect AI Agent Sa... detection of sabotage (failure to detect)
Human-only teams commit more errors than mixed human–AI teams.
Reported counts/observations of errors made by team type in the escape room experiment; the abstract does not include numerical error counts or significance levels.
Both humans and AI contribute wrong answers.
Reported error contributions from both human participants and AI agents in the experimental task.
high negative AI, Take the Wheel: What Drives Delegation and Trust in Huma... contribution of incorrect answers by humans and by AI
LLMs heavily rely on simulations for designing algorithms, which is notorious for breaking when transferred to real hardware.
Paper's claim grounded in known transferability issues between simulation and hardware; no experimental quantification provided in the abstract.
high negative GENESIS: Harnessing AI Agents for Autonomous 6G RAN Synthesi... algorithm performance when moving from simulation to real hardware (failure/brea...
Agentic systems show persistent failures in repository setup, dependency handling, permission gating, and hardware verification.
Issue-resolution benchmarks and hardware/RTL verification research synthesized in the paper (specific failure rates or sample sizes not provided in abstract).
high negative Agentic Agile-V: From Vibe Coding to Verified Engineering in... failure modes/errors in repository and hardware-related tasks
As many as 83% of workplace incidents stem from worker-AI misalignments.
Result from comparing LLM-extracted traits of AI systems (from 1,524 incident reports) to the traits preferred by workers (sample of 202); counted incidents where traits did not match worker preferences and reported proportion.
high negative The Quiet Path from Seemingly Minor Design Errors to Workpla... proportion of workplace AI incidents attributable to worker-AI trait misalignmen...
Frontier LLMs miss hidden errors.
Qualitative statement from paper indicating models fail to detect some latent or subtle errors in research artifacts; no numeric evaluation provided in excerpt.
high negative AI for Auto-Research: Roadmap & User Guide ability to detect hidden errors
Observed failures in the pilot were localized primarily to external integrations.
Pilot outcome summary in the paper stating failure localization was mainly due to external integrations (no numeric breakdown provided).
high negative GraphFlow: An Architecture for Formally Verifiable Visual Wo... failure source localization (external integrations vs core system)
Existing workflow platforms offer few semantic correctness guarantees.
Author statement contrasting current platforms' observability/durability with lack of semantic correctness guarantees.
high negative GraphFlow: An Architecture for Formally Verifiable Visual Wo... semantic correctness guarantees (presence/absence)
Under an idealized model of independent steps, a ten-step process with 90% per-step reliability completes successfully only 35% of the time.
Analytic, idealized independence model reported in the paper (mathematical calculation: 0.9^10 ≈ 0.3487).
high negative GraphFlow: An Architecture for Formally Verifiable Visual Wo... process completion probability
Traditional outcome-based reward models, which evaluate only the final correctness of a solution, often fail to identify logical fallacies or "hallucinations" occurring within intermediate steps.
Theoretical critique and conceptual argumentation presented in the paper; no empirical study or sample size reported.
high negative Optimizing Process Based Reward Models through Reinforcement... hallucination/error detection in intermediate reasoning steps
Monte Carlo simulations illustrate that standard DID estimators that ignore spillovers can miss the total effect.
Monte Carlo simulation results reported in the paper comparing standard DID estimators (which ignore spillovers) to the proposed approach; simulations show standard DID can fail to capture the total effect under spillovers.
high negative Identification and Estimation of Staggered Difference-in-Dif... accuracy of total effect estimation (bias/omission by standard DID)
Current MLLMs often recognize real-damaged evidence but fail on many fake-damaged subsets, with fake-damage detection rates (TPR) far below the 50% baseline on most generator subsets.
Experimental results reported in the paper comparing MLLM true positive rates (TPR) on real-damaged vs. fake-damaged subsets produced by multiple generators.
high negative FraudBench: A Multimodal Benchmark for Detecting AI-Generate... true positive rate (TPR) for detecting fake-damaged evidence
In a controlled experiment across six industry configurations (72 tool invocations using Qwen3-32B), unconstrained tool parameters produced a 43% hallucination rate for domain identifiers.
Controlled experiment reported in the paper: six industry configurations, 72 tool invocations, model used: Qwen3-32B; reported unconstrained parameter condition resulted in 43% hallucination rate for domain identifiers.
high negative The Semantic Training Gap: Ontology-Grounded Tool Architectu... hallucination rate for domain identifiers
In multi-agent configurations the semantic training gap produces a compounding failure mode termed 'semantic drift'.
Analytical description and demonstration in the paper describing multi-agent interactions and observed/argued compounding failures (conceptual demonstration; no numeric sample stated).
high negative The Semantic Training Gap: Ontology-Grounded Tool Architectu... occurrence of semantic drift (compounding errors in multi-agent setups)
Behaviorally, models fall into two failure modes: under-rejection, in which they answer misleading questions as if the false premise were true; and over-rejection, in which they reject more often but also reject standard questions, sacrificing ordinary comprehension accuracy.
Behavioral results on IMAVB showing distinct response patterns across tested models: some rarely reject misleading premises (under-rejection) while others reject too often including correct/standard questions (over-rejection), measured across the 500-clip benchmark.
(2) over-confidence, where agents skip essential environment verifications;
Trajectory analyses showing agents often omit verification steps leading to failed interactions; reported as an identified failure mode.
high negative ComplexMCP: Evaluation of LLM Agents in Dynamic, Interdepend... frequency of environment verification checks performed by agents
We evaluate various LLMs across full-context and RAG paradigms, revealing a stark performance gap: even top-tier models fail to exceed a 60% success rate, far trailing human performance 90%.
Empirical evaluation reported by authors comparing multiple LLM agents (full-context and RAG) against human performance on benchmark tasks; specific reported success rates: <=60% for top models, 90% for humans.
high negative ComplexMCP: Evaluation of LLM Agents in Dynamic, Interdepend... task success rate (agent vs human)
Even access to the true conditional vulnerability probability cannot eliminate misallocation: aleatoric uncertainty over individual vulnerability status is irreducible, and probabilistic targeting inevitably misallocates some resources.
Theoretical argument in the paper (conceptual/theoretical result about irreducible aleatoric uncertainty and its implications for probabilistic targeting).
high negative The Limits of AI-Driven Allocation: Optimal Screening under ... misallocation of resources (allocation error due to aleatoric uncertainty)
There are three practical failure modes produced or amplified by AI-assisted causal analysis: (1) method-data mismatch, where AI bypasses expertise at execution; (2) confidence laundering, where AI amplifies the credibility of formatted output; and (3) invisible forking, which spans both.
Taxonomy created and justified in the paper via conceptual argument and illustrative discussion; no empirical classification study or prevalence estimates provided.
high negative Vibe Econometrics and the Analysis Contract types of inferential failure modes arising in AI-assisted causal analysis
GPT-4.1 exhibits hidden workflow shortcuts despite achieving perfect TSR and HF1.
Model-level observation from the ASR analysis within the experiment (paper reports GPT-4.1 had perfect TSR and HF1 but failed trajectory-level fidelity).
high negative Beyond Task Success: Measuring Workflow Fidelity in LLM-Base... trajectory fidelity vs. standard metrics (TSR, HF1)
Applied to the Hierarchical Multi-Agent System for Payments (HMASP) across 18 LLMs and 90,000 task instances, ASR reveals that 10 of 18 models systematically skip a confirmation checkpoint during payment checkout, a deviation invisible to both TSR and HF1, while 8 models enforce the checkpoint perfectly.
Empirical evaluation reported in the paper: HMASP tested across 18 LLMs and 90,000 task instances; analysis via ASR showing checkpoint-skipping behavior for 10 models and correct enforcement for 8 models.
high negative Beyond Task Success: Measuring Workflow Fidelity in LLM-Base... adherence to expected workflow transitions (confirmation checkpoint adherence)
Novices more often experience invisible failures: conversations that appear to end successfully but in fact miss the mark.
Annotation-based comparison in the 27K WildChat transcript sample indicating higher rates of 'invisible' failures (apparent successes that are actually incorrect or insufficient) among novice users.
high negative A paradox of AI fluency invisible failure rate (apparent success but incorrect outcome)
Fluent users experience more failures than novices.
Quantitative comparison of failure occurrences across user-fluency strata in the 27K annotated transcript sample from WildChat-4.8M.
high negative A paradox of AI fluency failure rate (errors / failed turns)
Hallucination rate does not grow monotonically with document length: short documents and very long ones both hallucinate more than medium ones (28.1% and 23.8% vs. 9.2%).
Empirical measurement of hallucination rates by document-length buckets on EnterpriseDocBench; percentages reported in paper. Sample sizes per bucket not provided in excerpt.
high negative Benchmarking Complex Multimodal Document Processing Pipeline... hallucination rate (fraction of generated outputs judged hallucinated)