Divya van Mahajan

Workers Are Excited About Ai Agents So Why Are Companies Screwing It

ai-daily-brief-podcast

Overview

This episode of the AI Daily Brief podcast analyzes a new EY study examining how U.S. workers feel about agentic AI in the workplace. The host (affiliated with Super Intelligent, an AI advisory firm) uses the study to explore a central paradox: workers are broadly enthusiastic about AI agents, yet companies are failing to harness that enthusiasm through poor communication, inadequate training, and underprepared management structures. The talk situates the EY findings within a broader 2025 context of rapid agent deployment and a well-documented executive-employee perception gap.

Source video URL: (not provided)

Prerequisites

  • Basic familiarity with AI terminology: large language models, AI assistants/copilots, and AI agents
  • General awareness of enterprise AI adoption trends in 2025
  • Understanding of the distinction between AI as a productivity tool (copilot) versus AI as an autonomous actor (agent)
  • Familiarity with organizational change management concepts (communication strategy, upskilling, workforce management)

Main Points

Rapid Growth of Agent Deployment in 2025

  • Per a KPMG study, agent deployment in enterprises nearly quadrupled from 11% in Q1 2025 to 42% in Q3 2025
  • Agents are rapidly transitioning from theoretical to operational within large organizations
  • This acceleration makes understanding worker sentiment toward agents increasingly urgent

The Executive–Employee Perception Gap (Context-Setting)

  • A December 2024 Writer.ai study of 800 employees and 800 executives revealed significant disconnects:
    • 73% of executives said their AI approach was “well-controlled and strategic”; only 47% of employees agreed
    • 64% of executives believed their company had high AI literacy; only 33% of employees agreed
    • 89% of executives said their company had an AI strategy; only 57% of employees agreed
  • This gap forms the backdrop against which the EY findings must be interpreted

EY Study Methodology

  • Over 1,100 U.S. employees surveyed across six industries (banking, wealth management, consumer products, manufacturing, oil and gas, technology)
  • All respondents worked at companies with annual revenue of $1 billion or more
  • Near-even split between managers/supervisors (586) and non-managers (562)
  • Broad generational representation: Gen Z, Millennials, Gen X, Baby Boomers
  • Field dates: August 8–September 3, 2025 — post-GPT-5 release, making it relatively current

Finding 1: Workers Are Broadly Enthusiastic About Agents

  • 84% of workers said they were eager to embrace agentic AI in their role
  • Expected positive impacts cited: productivity (86%), work experience (83%), work-life balance (82%)
  • Workers appear to view agents as tools that reduce drudgery and return time, not just as productivity metrics for employers
  • 90% of those already using agentic AI reported confidence in their ability to use it

Finding 2: Job Security Remains a Significant Concern

  • 56% of workers expressed concern about job security — overlapping substantially with the 84% who are enthusiastic
  • The host frames this as rational: workers can simultaneously see personal benefit and structural risk
  • Concern is higher among non-managers (65%) than managers (48%), reflecting differential exposure to displacement risk

Finding 3: Information Overload Is Widespread

  • 61% of all respondents feel overwhelmed by the constant influx of new agentic AI information
  • Even among current agentic AI users, 64% feel overwhelmed by the pace of new tool introductions
  • 54% overall feel they are falling behind their peers on agentic AI adoption (48% managers, 61% non-managers)
  • General overwhelm is translating into specific anxiety about relative competitiveness

Finding 4: Management of Hybrid Human-Agent Workforces Is a Novel and Unresolved Challenge

  • Workers broadly recognize that managing hybrid human-agent teams requires fundamentally new skills
  • 53% of current managers are concerned they are not equipped to integrate and manage such teams
  • 63% of non-managers said they are less inclined to pursue management roles due to these concerns
  • 82% of managers said adding AI agent supervision would make their role more challenging
  • Key open questions include: task allocation between humans and agents, handoff workflows, and role boundary (“role bleed”) management
  • The host notes that no one is truly expert in this area yet — the skill gap between the average manager and the best in the world is narrower than in almost any other domain

Generational Differences in Management Confidence

  • Baby Boomers: Potentially well-positioned for leadership but not placing high emphasis on agentic AI importance
  • Gen Z: Familiar with and optimistic about agents but focused on personal impact rather than management; most likely to cite lack of training as a barrier
  • Millennials: Experiencing a “confidence crisis” — 62% of millennial managers (vs. 53% overall average) are concerned about managing AI-augmented teams; caught between rising responsibility and displacement anxiety
  • Gen X: The “Goldilocks” cohort — high enthusiasm, high usage, and 94% confident in their ability to manage AI-augmented teams

Imperative 1: Clear Communication

  • Organizations that clearly communicate their AI strategy show dramatically better outcomes:
    • Actual agentic AI usage: 66% (clear communicators) vs. 39% (unclear communicators)
    • Willingness to embrace AI: 87% vs. 69%
    • Familiarity with agentic AI: 74% vs. 46%
  • Positive sentiment is also substantially higher at clearly communicating organizations:
    • Productivity improvement expectation: 89% vs. 76%
    • Decision-making improvement: 87% vs. 67%
    • Work-life balance improvement: 85% vs. 69%
  • Among workers who have already used agents, 92% at clear-communication firms said agents positively impacted their productivity, versus only 62% at non-communicating firms — a 30-percentage-point gap
  • The host notes that “clear” is the operative word: overcommunication or unclear messaging can also create overwhelm

Imperative 2: Effective Training

  • 89% of respondents believe upskilling and reskilling are crucial for staying relevant
  • Yet 59% cite lack of adequate training as a key organizational barrier
  • Only 52% of organizational leaders say their company has a fully developed agentic training program
  • 85% of workers are learning about agents outside of work; 83% describe their agentic AI knowledge as self-taught
  • The host distinguishes agentic AI management from prompt engineering — it requires workflow thinking, comprehensive planning, and a different cognitive approach
  • Market gap: few robust off-the-shelf training programs exist; most organizational learning is homegrown (sandboxes, internal champions, peer learning)
  • Security and quality risks arise when employees self-train using unvetted sources or input confidential data into personal tool subscriptions

Imperative 3: Rethinking Management

  • The management model must evolve as every employee — including junior staff — increasingly manages teams of agents
  • Traditional management training is insufficient; agent-specific management training is needed at all levels
  • EY recommends organizations build new frameworks for supervising hybrid workforces, including role definition, workflow design, and accountability structures

Key Concepts

  • Agentic AI / AI Agents: AI systems that autonomously take actions and complete multi-step tasks, as distinct from passive AI assistants or copilots
  • Hybrid Human-Agentic Workforce: A workforce model in which both humans and AI agents perform work tasks, requiring new management approaches
  • Executive–Employee Perception Gap: The documented divergence between how organizational leaders and frontline workers perceive the maturity, clarity, and effectiveness of company AI initiatives
  • Role Bleed: The phenomenon where expanded AI-enabled capacity causes employees to take on tasks outside their traditional role boundaries, raising questions about specialization and accountability
  • Context Engineering: Described as an emerging discipline beyond prompt engineering, focused on structuring inputs, memory, and information flow for effective agent use
  • Agentic Training Program: A structured organizational effort to develop employee skills specifically for working with, directing, and managing AI agents — distinct from general AI literacy programs
  • Communication Overhang: A state in which the gap between what leaders believe has been communicated about AI strategy and what employees have actually internalized creates organizational friction and underperformance

Summary

The EY study, covering over 1,100 U.S. employees surveyed in late summer 2025, reveals that the primary barrier to successful agentic AI adoption in large enterprises is not worker resistance but organizational failure. Workers are largely enthusiastic — 84% eager to embrace agents — and understand both the personal productivity gains and the structural risks these tools represent. However, companies are squandering this goodwill through three compounding failures: unclear or inconsistent communication about AI strategy (which the data shows has a direct, measurable negative effect on adoption rates and worker sentiment), insufficient training (with over 80% of workers self-teaching outside of work), and an absence of frameworks for managing hybrid human-agent teams (leaving managers and aspiring managers alike feeling unprepared and deterred). The host’s overarching message is that agent adoption is no longer a future question — it is a present operational reality — and organizations that invest now in clear communication, structured training, and new management models will see compounding returns, while those that do not risk both underperformance of their AI investments and erosion of the employee enthusiasm that currently exists as a genuine organizational asset.