Ten evidence-anchored signals. Each card cites the source and its confidence grade. Swipe through. At the end, see how your task map compares.
Card 1 of 11
MIT Sloan Management Review / Boston Consulting Group — MIT Sloan Management Review — Achieving AI at Scale (2024) · Confirmed
Cash Flow Management
MIT Sloan Management Review's annual survey of 3,000+ managers found that only 10% of organizations report significant financial value from AI deployment, despite widespread experimentation. For tasks like Cash Flow Management, the study indicates 15% effective capability in practice — not due to technical limitations but because change management and data infrastructure are bigger barriers than technology. Organizations achieving value focus on process-level rather than task-level deployment.
Evidence: 2024Roles: finance manager
Stanford University Human-Centered AI Institute — Stanford HAI AI Index Report 2025 (2025) · Confirmed
Data Entry & Administration
The Stanford HAI AI Index Report 2025 documents AI systems achieving expert-level performance on graduate-level science questions and professional coding tasks. For tasks like Data Entry & Administration, current AI capability is estimated at 64% of human expert performance, with strongest results on well-defined problems and weaker performance on open-ended reasoning.
Evidence: 2025Roles: all
Noy & Zhang (MIT) — Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence (2023) · Confirmed
Financial Reporting
Noy & Zhang found in a controlled experiment that AI assistance reduced professional writing task completion time by 40% and improved output quality by 18%. Tasks similar to Financial Reporting fall within the category of professional writing where these productivity gains were observed, suggesting 52% quality parity with human baseline on structured writing components.
MIT Sloan Management Review / Boston Consulting Group — MIT Sloan Management Review — Achieving AI at Scale (2024) · Confirmed
Pipeline Management
MIT Sloan Management Review's annual survey of 3,000+ managers found that only 10% of organizations report significant financial value from AI deployment, despite widespread experimentation. For tasks like Pipeline Management, the study indicates 24% effective capability in practice — not due to technical limitations but because change management and data infrastructure are bigger barriers than technology. Organizations achieving value focus on process-level rather than task-level deployment.
Evidence: 2024Roles: sales manager
Noy & Zhang (MIT) — Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence (2023) · Confirmed
Data Analysis & Reporting
Noy & Zhang found in a controlled experiment that AI assistance reduced professional writing task completion time by 40% and improved output quality by 18%. Tasks similar to Data Analysis & Reporting fall within the category of professional writing where these productivity gains were observed, suggesting 63% quality parity with human baseline on structured writing components.
System: ChatGPTEvidence: 2023Roles: business analyst
OpenAI Blog · Plausible
Code Review
Enterprise adoption for accelerating code experimentation indicates code generation capability is mature enough for production workflows, enabling developers to write code faster.
Evidence: 2026Roles: software engineer
Harvard Law School Center on the Legal Profession — Artificial Intelligence and the Legal Profession (2024) · Confirmed
Regulatory Monitoring
Harvard Law School research found that AI contract review tools achieve 85-95% accuracy on standard clause identification, exceeding junior associate performance on routine reviews. For tasks like Regulatory Monitoring, AI demonstrates approximately 49% quality on structured legal analysis, though novel legal arguments, jurisdictional nuances, and high-stakes advisory work remain human-dependent.
Evidence: 2024Roles: compliance officer
Harvard Law School Center on the Legal Profession — Artificial Intelligence and the Legal Profession (2024) · Confirmed
Compliance Training Delivery
Harvard Law School research found that AI contract review tools achieve 85-95% accuracy on standard clause identification, exceeding junior associate performance on routine reviews. For tasks like Compliance Training Delivery, AI demonstrates approximately 38% quality on structured legal analysis, though novel legal arguments, jurisdictional nuances, and high-stakes advisory work remain human-dependent.
Evidence: 2024Roles: compliance officer
Noy & Zhang (MIT) — Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence (2023) · Confirmed
Audit Preparation & Coordination
Noy & Zhang found in a controlled experiment that AI assistance reduced professional writing task completion time by 40% and improved output quality by 18%. Tasks similar to Audit Preparation & Coordination fall within the category of professional writing where these productivity gains were observed, suggesting 52% quality parity with human baseline on structured writing components.
Stanford University Human-Centered AI Institute — Stanford HAI AI Index Report 2025 (2025) · Confirmed
Roadmap Prioritisation
The Stanford HAI AI Index Report 2025 documents AI systems achieving expert-level performance on graduate-level science questions and professional coding tasks. For tasks like Roadmap Prioritisation, current AI capability is estimated at 40% of human expert performance, with strongest results on well-defined problems and weaker performance on open-ended reasoning.
Evidence: 2025Roles: product manager
The deck ends here
See your own Task Map
Ten signals only scratched the surface. Your role has thirty tasks the graph is tracking. Take the free assessment and Runway will surface the specific ones moving against you — with the same evidence grade you saw on these cards.
Tier definitions: Confirmed = peer-reviewed or multi-source corroboration · Plausible = single credible source · Early Signal = preliminary, unverified. See methodology.