Society · Trend
AI Agents Reshaping Knowledge Work
Enterprise push toward agentic systems that execute end-to-end business tasks across workflows, data, and tools.
Forecast horizon
Open-weight models now match 2024 frontier performance at a fraction of the cost, democratising capable workplace agents.
Connections
Connections · 6
How this node ties into the rest of the map, and the evidence behind each link.
from · drives 4/10
Generative AI Erosion of Human Temporal Learning As AI agents automate knowledge work execution, the economic pressure against human temporal learning intensifies.
+4 growth from · drives 3/10
Risks of Mass AI Agent Interactions Mass-market arrival of agents carrying out tasks without human oversight creates systemic interaction risks as AI agents reshape knowledge work.
+3 growth from · informs 3/10
LLM Agent Performance in Dynamic Environments EvoArena benchmark reveals current agents struggle in dynamic environments, a key challenge for AI agents in real-world knowledge work.
+3 growth from · enables 3/10
LLM-Automated Scientific Reproducibility Assessment LLM-automated reproducibility assessment demonstrates AI agents taking over knowledge-intensive scientific audit tasks.
+3 growth from · drives 3/10
South Korea AI Adoption and Social Integration South Korea's deep societal AI integration is driving rapid adoption of AI agents across professional and consumer contexts.
+3 growth from · constrains 3/10
AI Agent Data Leakage in Non-Adversarial Enterprise Scenarios Data leakage risks in non-adversarial enterprise AI agent deployments constrain safe adoption of AI agents in knowledge work.
+3 growthSignal sources
Signal sources
Dated facts from primary sources in this direction.
AI in government Sep 2025
An OECD review of ~200 government AI use cases found 57% support automating or tailoring public services and 45% enhance decision-making — most still stuck in pilots.
OECD — Governing with AI → Labor signal Nov 2025
A Stanford payroll study found a 13% relative decline in employment for workers aged 22–25 in AI-exposed occupations since late 2022, while older peers held steady.
Stanford Digital Economy Lab → Information integrity Oct 2025
Microsoft reported AI-enhanced phishing reached a 54% click-through rate (4.5× traditional) and AI-generated fake IDs grew 195% globally in 2025.
Microsoft — Digital Defense Report 2025 →