AI Adoption May Reduce Productivity Before Improving It, New DEC Report Finds

AI Adoption May Reduce Productivity Before Improving It, New DEC Report Finds

• AI deployment is outpacing the skills needed to use it effectively
• Sales, administrative, and computer-related roles face the highest share of automatable tasks
• As AI absorbs entry-level tasks, pathways for young professionals to build expertise may be narrowing

SINGAPORE, 13 May 2026 — The Digital Education Council (DEC) has released AI Skills Opportunity Map, a report analysing how AI is transforming professional work across 11 job families and what institutions and employers must now do differently. Google.org engaged DEC for this new global AI readiness research initiative.

AI is reshaping work unevenly across job families

Sales, administrative, and computer-related roles face the highest share of automatable tasks, where work is structured and rule-based. Banking and healthcare retain a larger share of human-led work as accountability and risk-sensitive decision-making create a firm ceiling on AI integration. In education, business and management, AI could support the majority of work and become embedded into everyday workflows.

Productivity gains are not immediate

The report identifies a productivity paradox: AI adoption may initially reduce productivity before delivering sustained gains. 

The core problem is that organisations are scaling AI output faster than they can evaluate it. This output-review imbalance is compounded when professionals misapply tools they do not yet fully understand, and worsens further as AI takes on more complex work that users lack the expertise to assess. 

The result is friction that accumulates quietly, and often goes unrecognised until productivity has already declined.

The dual imperative for skills development

To help institutions and employers respond to the productivity paradox, the report highlights a dual imperative for skills development: professionals need both applied AI skills and enduring human capabilities.

The report identifies five applied AI skills across the AI interaction cycle:

• AI-assisted workflow design
• AI interaction and instruction
• AI output evaluation and adaptation
• Elevation and decision-making
• AI ethics-aware practice

These are complemented by six enduring human capabilities:

• Industry expertise
• Systems thinking
• Problem framing
• Organisational and social intelligence
• Adaptive agency
• Metacognition and cognitive discipline

While applied AI skills will continue to evolve as tools change, the report shows that enduring human capabilities will be essential for sustaining effective and responsible AI use over time.

AI increases the value of expertise while making it harder to build

AI does not benefit all workers equally, the report finds. 

Professionals with deeper domain expertise are better positioned to judge where AI should be applied and when human involvement is needed.  But AI is increasingly taking over the entry-level and routine work through which that experience is typically built for young professionals.

Preparing students and early-career professionals for an AI-enabled workforce requires intentionally designing learning, training, and work experiences that develop the expertise needed to use AI well, not only the technical familiarity to use it at all.

"Organisations are discovering that deploying AI and benefiting from AI are two very different things. The gap between them is a skills gap, and closing it requires a fundamentally different approach to how we educate and train people," said Alessandro Di Lullo, CEO of the Digital Education Council.

"Every institution and employer is asking the same question right now: how do we prepare people for a workforce that AI is changing faster than our curricula can keep up with. This report gives them something concrete to work with," said Danny Bielik, President of the Digital Education Council.

The AI Skills Opportunity Map is available for download here.

‍