AI writing at work: Why your team needs skills, not just tools

Many organisations have already given employees access to AI tools. The harder question is whether people know how to use them well enough to improve the work.

This is especially clear in workplace writing. Employees are using AI to draft emails, summarise information, prepare reports, improve proposals and shape internal updates. But the quality is uneven. Some people use it confidently. Some are experimenting quietly. Some have barely started. And even where AI is being used, it does not always translate into clearer writing, faster delivery or better business outcomes.

For HR, People and Culture, and L&D teams, this is the real challenge: AI adoption does not automatically create AI capability. Without training, organisations may see more content being produced, but not necessarily better communication.

The issue is not access. It is inconsistent capability.

The adoption picture is more complicated than the headlines suggest. EY’s 2025 Work Reimagined Survey found that AI usage at work is now widespread, with 88% of employees using AI in some way. Yet only 28% of organisations are positioned to turn AI deployment into high-value business outcomes.

In Australia, the gap is also clear. EY’s Australian AI Workforce Blueprint found that only 32% of workers rate their AI proficiency as high, and only 35% have received formal AI training from their employer.

That matches what many facilitators see in writing workshops. Roughly half the room has tried AI, often for personal or low-risk tasks. A smaller group uses it reasonably well at work. Another group has barely touched it.

The problem is not that people are unwilling. It is that most organisations do not yet have a shared way to build the skills, judgement and confidence people need to use AI well.

Writing is where the gap shows up quickly

Writing is one of the easiest ways to start using AI. It is also one of the easiest ways to use it badly.

A polished AI draft can look finished. The sentences may flow. The structure may appear neat. The tone may sound professional. But that does not mean the message is clear, accurate, useful or appropriate for the audience.

Poor AI writing often creates work for someone else. A manager has to rewrite the message. A colleague has to clarify the point. A client receives a response that sounds pleasant but does not answer the question. A report includes confident statements that have not been checked.

This is why AI writing is not just a technology issue. It is a business communication issue.

The productivity gains are not automatic

AI can improve productivity when people know how to use it well. A major MIT study found that professionals using ChatGPT for business writing completed tasks faster and produced higher-quality output.

That is the opportunity. But the benefit does not come from the tool alone. It comes from how people use it: to shape thinking, organise information, improve structure, reduce blank-page time and refine the final message.

Without those skills, AI can create a productivity illusion. The first draft appears quickly, but the review, correction and rewriting take longer than expected. Harvard Business Review has described this problem as “workslop”: AI-generated work that looks polished but lacks the substance needed to move the task forward.

For organisations, this is where the gap becomes costly. People may feel more productive individually, but the business does not always see the same lift in quality, speed or decision-making.

AI writing also creates risk

The risks are not only about vague writing or generic tone. They are also about judgement.

Employees may paste sensitive or confidential information into tools they do not fully understand. They may rely on AI output without checking it. They may accept inaccurate claims because the wording sounds confident. They may generate content that does not match the organisation’s voice, standards or compliance requirements.

Most of these mistakes are not malicious. They happen because employees are trying to be efficient without enough guidance.

That is why responsible AI writing needs to be taught as a practical workplace skill. People need to understand what they can use AI for, what they should avoid, how to check the output and where human accountability remains essential.

What good AI writing capability looks like

Effective AI writing is not about letting AI do the work. It is about using AI as a writing partner while keeping human judgement in control.

Employees need to learn how to:

• give AI clear context, purpose and audience
• use prompts that support better thinking, not just faster drafting
• improve clarity, structure, tone and brevity
• check facts, assumptions and recommendations
• protect confidential and sensitive information
• edit AI output so it sounds credible, useful and human
• challenge AI on its always confident but often incorrect output.

When these skills are in place, AI can become genuinely helpful. It can support clearer emails, sharper summaries, stronger reports, better first drafts and more confident editing. It can also help less confident writers improve, because they are no longer starting from a blank page.

Why HR and L&D teams need a scalable approach

If employees learn AI writing informally, capability will remain uneven. One person may use AI responsibly and save time. Another may produce generic content that creates rework. Another may avoid AI altogether because they are unsure what is allowed.

A shared training approach gives the organisation a common standard. It helps employees understand how AI can support their writing, where the boundaries are and what good output looks like.

It also gives managers and leaders more confidence. Instead of hoping employees are using AI appropriately, organisations can build the skills and expectations that make responsible use more consistent.

How ICML helps teams write better with AI

ICML’s AI for Business Writing, AI for Government Writing and AI for Local Government Writing workshops helps employees move beyond experimentation and into confident, responsible use.

The workshops focus on real workplace writing tasks. Participants learn how to use AI to plan, draft, refine and improve written communication while still applying human judgement. They also learn how to handle sensitive and confidential information responsibly, so productivity gains do not come at the expense of trust, accuracy or compliance.

For organisations, the opportunity is not simply to get more people using AI. It is to help people use AI well.

If your team has access to AI but is not yet seeing the quality, consistency or productivity gains you expected, the next step is capability.

Ask for a quote or request a proposal, and we will tailor a Writing with Practical AI workshops to the writing tasks, communication standards and risk profile your organisation needs to improve.