الذكاء الاصطناعي في الإدارة: دليل عملي لرفع إنتاجية الفريق

By MTMS Global · Last updated: 27/10/2026 · Reading time: 8 minutes · Category: AI in Management

Introduction

Managers today ask two questions about AI: where does it genuinely raise my team’s productivity, and where does it expose me to error or a data leak? An evidence-based answer is better than an impression, so this article starts with the results of two published field studies, then moves to managerial use and safeguards.

What the studies say

Brynjolfsson, Li and Raymond (2025) followed 5,179 customer support agents who used a generative-AI-based assistant. Productivity rose by an average of 14%, and by 34% among novice and low-skilled workers. In other words, the tool narrowed the gap between newcomers and experienced staff in that setting.

In an experiment by Dell’Acqua and colleagues (2023) with 758 consultants, users completed 12.2% more tasks, 25.1% faster, and at more than 40% higher quality on tasks within the tool’s capabilities. But on a task deliberately chosen to fall outside the tool’s “technological frontier”, users were 19 percentage points less likely to produce a correct solution than those who did not use it.

The practical lesson is that the effect depends on the type of task: productivity rises on tasks within the tool’s capabilities, and the risk of confident error grows outside them. The results apply to the work studied (support and consulting) and should not be generalized automatically to every job.

Areas of managerial use

Davenport and Ronanki (2018) classify organizational applications of AI into process automation, data understanding, and engagement of customers and employees. This is reflected in a manager’s work as follows:

Usage safeguards

• Privacy: do not enter personal data, contracts or confidential figures into a tool your organization has not approved. Egypt’s Personal Data Protection Law No. 151 of 2020 protects personal data, so consult a legal adviser before entering individuals’ data into any tool.
• Verification: review every output that contains figures, references or judgments before using it, because a result can look coherent and still be wrong.
• Accountability: the decision and the signature belong to the manager, not the tool, and what the tool was used for should be documented.
• A risk framework: the AI Risk Management Framework issued by the U.S. National Institute of Standards and Technology (NIST AI RMF 1.0) provides a structure for governing, measuring and managing risk (NIST, 2023), and it can be aligned with the risk register we explained in Enterprise risk management: the framework, the process and the risk register (Arabic).

Steps for a safe pilot

1. Choose one recurring, low-risk task, such as summarizing meeting minutes.
2. Measure the current situation before the pilot: time taken and number of errors.
3. Define the data that may and may not be entered, and write it on one page.
4. Pilot with a small group for four weeks, keeping a human review on every output.
5. Compare the results with the previous situation, then decide whether to expand, stop or adjust.

An illustrative example

The following example is hypothetical, to illustrate the application; its figures are not real data. It assumes a sales manager in a distribution company who wants to reduce the time spent writing weekly meeting minutes.

Conclusion and recommendations

Field studies point to real gains on specific tasks and to a genuine risk outside the tool’s frontier. A manager is therefore advised to:

1. Start with one low-risk task and measure before and after the pilot.
2. Write a clear rule for what may and may not be entered.
3. Keep a human review on every output that contains figures or judgments.
4. Not rely on the tool for decisions about individuals without governance and legal review.
5. Train the team to tell tasks inside the tool’s capabilities from those outside them.

Frequently asked questions

Does AI really raise productivity?

Field studies indicate that it does on specific tasks; in a study of 5,179 support agents, productivity rose by an average of 14% (Brynjolfsson et al., 2025). But the effect depends on the type of task and should not be generalized to every job.

What is the biggest risk in managerial use?

Two: entering confidential or personal data into an unapproved tool, and trusting outputs that look coherent but are wrong, especially on tasks outside the tool’s capabilities (Dell’Acqua et al., 2023).

Who benefits most from these tools?

In the support study, novice and low-skilled workers benefited most, with a gain of 34% (Brynjolfsson et al., 2025); this applies to the setting of that study.

How do I start without major risk?

Choose one recurring, low-risk task, measure the current situation, define the data that is allowed and prohibited, pilot with a small group under human review, and compare results before expanding.

Putting it into practice

MTMS Global offers training programs for managers and teams on using AI tools with clear safeguards, and consulting services for drafting usage policies. For the fundamentals of managerial work, read Business management: managerial functions, manager roles and skills (Arabic). You can also contact us to discuss your organization’s needs.