Generative AI can make business writing faster and easier, but it can also introduce errors and flatten your individual voice. The answer is not necessarily whether to use it, but how. Read more...
Writing occupies a significant part of the working day. Emails, reports, proposals, meeting notes and presentations all compete for our time and attention.
It is easy to understand the appeal of generative AI. A tool that can produce a draft in seconds appears to offer an immediate solution to information overload.
However, speed is not the only measure of effective communication. Accuracy, originality, confidentiality and trust also matter.
The most productive approach is to use AI as an assistant while retaining human control over the purpose, judgement and final message.
Where AI Can Help
Generative AI can be useful during several stages of the writing process.
It can help organise a collection of rough ideas, suggest a structure, summarise lengthy information or identify gaps in an argument. It can offer alternative headings, simplify complicated language and provide feedback on tone.
For someone facing a blank page, asking AI to generate questions or propose an outline may be more valuable than asking it to write the entire document.
These applications reduce routine work while keeping the writer actively involved.
Protect Your Own Voice
AI-generated writing often sounds polished, but it can also sound generic. If people rely on it to create complete drafts, their communication may gradually lose the experiences, language and personality that make it recognisably theirs.
Your voice is not simply a writing style. It reflects your judgement, expertise, relationships and understanding of the audience.
Use AI-generated text as material to evaluate rather than finished copy to accept automatically. Rewrite it, challenge it and add the details that only you can provide.
Check Every Claim
Generative AI can produce information that sounds convincing but is incomplete, misleading or entirely incorrect.
Every factual statement, statistic, quotation and reference must be checked against a reliable source. This is particularly important in legal, medical, financial and professional communications where an error could cause serious harm.
The person sending or publishing the document remains responsible for its contents, regardless of how the first draft was produced.
Consider Privacy and Confidentiality
Before entering information into an AI system, consider whether it contains private, personal or commercially sensitive material.
Client information, employee details, unpublished strategies and confidential documents should not be submitted without understanding the organisation’s policies and the tool’s data protections.
Check which AI tools your organisation permits and what information can safely be used.
Be Transparent Where Necessary
There is no universal rule requiring disclosure every time AI helps improve a sentence. However, transparency becomes more important when AI has made a substantial contribution or when authorship and originality are central to the work.
Organisations should establish clear expectations so employees understand when AI assistance is appropriate and when it should be acknowledged.
Use AI Without Giving Up the Skill
Writing is part of thinking. The process forces us to clarify what we mean, identify weaknesses in an argument and make decisions about what matters.
If AI performs the entire task, we may save time but lose some of that valuable thinking.
Use the technology to support the difficult parts of writing, not to avoid engaging with the subject altogether.
The Takeaway
The strongest approach is to treat AI as a capable but imperfect assistant. Give it defined tasks, examine its output critically and ensure that the final document still reflects your knowledge, values and voice.
This article is an independently written summary of ideas presented in “Should You Write with Gen AI?” by Alexandra Samuel, published by Harvard Business Review on 9 December 2024. Read the original HBR article for the author’s complete guidance on using generative AI responsibly.
