Keeping Communication Human When AI Drafts Your Team's Messages

A growing share of workplace communication now passes through AI before anyone reads it. Emails are drafted by assistants, meetings are summarised automatically, chat messages are polished or translated in a click. Combined with remote and hybrid working, this means many colleagues now see each other less, write to each other more, and increasingly read words that neither of them fully wrote.

For multicultural teams, this creates a specific challenge. Communication styles differ across cultures – in directness, in how much context is spelled out, in how politeness is expressed. AI tools tend to flatten those differences into one generic tone. That can help, but it can also strip out the signals people rely on to build trust. This article looks at those everyday effects and how teams can manage them. For the organisational picture, see our article on why company culture decides whether AI adoption succeeds.

How AI changes the way teams communicate

• Less face-to-face, more text. Async tools and AI summaries reduce the need for meetings, but also the informal moments where relationships form.
• More direct by default. Short, efficient messages – often suggested by AI – can read as curt to colleagues from cultures where warmth and context come first.
• Translated, not interpreted. Machine translation conveys words well but can miss tone, humour, hierarchy markers and implied meaning.
• Summarised, not heard. Meeting summaries capture decisions but often lose hesitations, objections voiced softly, and who was left out of the conversation.

Where culture shapes the outcome

Direct and indirect styles

In some cultures, clear and blunt feedback is a sign of respect. In others, the same message is softened, framed with context, or delivered privately. When an AI tool rewrites a message to be "clearer", it may remove the softening that made it acceptable to the recipient – or add politeness that makes an urgent request sound optional.

High- and low-context communication

Some colleagues expect messages to say everything explicitly. Others rely on shared context, relationships and what is left unsaid. AI summaries tend to favour the explicit, so the meaning carried by context can disappear.

Hierarchy and voice

In cultures with greater respect for hierarchy, junior colleagues may express disagreement very gently. In an AI summary, a gentle "perhaps we could also consider…" may not register as an objection at all.

Illustrative scenarios

The "efficient" reply. A project lead uses an assistant to reply quickly to a partner abroad. The reply is accurate but omits the greeting and personal note the partner always includes. The partner reads it as a sign the relationship has cooled. A one-line personal opening would have prevented the misunderstanding.

The missing objection. An automated meeting summary records that a plan was "agreed". A colleague who raised a cautious concern in her second language does not see it reflected and concludes that speaking up is pointless.

The translated joke. A light-hearted comment in a team channel is auto-translated literally and lands as an insult. Nobody notices until relationships are strained.

How to keep AI-assisted communication human

Review before you send

Treat AI drafts as a first version. Before sending, ask: does this sound like me, and does it suit this person and this relationship?

Keep some communication live

Protect regular video or in-person time for relationship-building, difficult conversations and feedback. These are the moments where AI adds least and risks most.

Check summaries with participants

Ask attendees to add anything an automated summary missed, especially concerns. Make it normal to say "That is not quite what I meant."

Be transparent about AI use

Agree as a team when it is fine to use AI for drafting or translation, and when people would prefer a personal message.

A quick self-check before you send

Before sending an AI-assisted message to a colleague or partner from a different background, take ten seconds to ask:

• Relationship: would this person normally expect a greeting, a personal line or some context first?
• Directness: is the request clear enough to be understood as a request, and soft enough not to sound like an order?
• Meaning: if it will be translated, have I avoided idioms, sarcasm and jokes that may not survive?
• Channel: should this be a call instead of a message?
• Ownership: would I be comfortable if the recipient knew exactly how this was written?

This small habit costs little and prevents many of the misunderstandings that erode trust in distributed teams.

A team communication charter for the AI era

Many teams find it useful to agree a short charter. A starting template:

1. We use AI to draft and summarise, but we read and own everything we send.
2. Feedback, bad news and sensitive topics are discussed live, not by AI-written message.
3. Meeting summaries are shared for corrections before they become the record.
4. We say when a message has been machine-translated, and we ask rather than assume if something reads oddly.
5. We keep a personal touch with partners and colleagues for whom it matters.
6. We review this charter every few months and adjust it together.

Teams communicate best when their members understand their own communication tendencies. The free Culture XI CQ Brindo Sphere™ assessment helps people see how they adapt to cultural differences. Explore more in our cultural intelligence articles, and read how culture also protects teams from AI-driven work intensification and burnout.

Frequently asked questions

Is it rude to use AI to write messages to colleagues?

Not in itself. What matters is whether the message suits the relationship and reflects what you mean. Review drafts and keep sensitive conversations personal.

Can machine translation replace learning about other cultures?

No. Translation handles words; cultural understanding handles meaning, tone and expectations. Both are useful.

How do we make sure quieter colleagues are heard in AI summaries?

Share summaries for correction, invite comments in writing, and leaders can explicitly ask for concerns that may not have been captured.

Should every team have an AI communication charter?

It is especially useful for distributed or multicultural teams. Keep it short and review it regularly.