For 20 years, I’ve been telling the women in my training rooms to stop softening their language at work.
That means deleting every unnecessary ‘just wondering if’, every ‘sorry to bother you’ and every ‘hope that’s okay’ from their emails. And it means no more cushioning every opinion and
recommendation with statements like ‘I might be wrong, but…’.
And my message has been getting through. Slowly but surely.
Days, months and even years later, women get back in touch to tell me what happened once they stopped writing that way. They talk about feeling more confident and being taken more seriously. Many also tell me they hadn’t realised just how often they were undermining themselves until they stopped doing it and how liberating that realisation has been.
But now, something new is threatening to undo all that progress – the tool millions of us are using every day to write: AI. The scenario I’m seeing goes like this.
A woman drafts her email. It’s clear and confident. Just as it should be. But then, she runs it through an AI tool – ChatGPT, Copilot, Grammarly – just to check that her tone is ‘professional’ enough. Then sends AI’s new version, trusting its judgement over her own.
And what she may not notice is what got added back in. A sneaky ‘just’ in her opening line. A gratuitous apology midway down. And unnecessary caveats to close it out. Individually, those changes seem harmless. Together, they dilute the confidence she started with and alter how she’s perceived.
So why does AI do this?
The robots didn’t invent what ‘professional’ looks like. They’ve learnt this style from the billions of real emails and documents they’ve been trained on. And historically, ‘professional’ has meant more hedging and more deference. More ‘I hope this is all okay with you’ and less ‘here’s what I recommend’.
To be clear, I’m not arguing that AI is biased against women (that’s a different conversation entirely). However, there’s no question that women are more likely to accept an apologetic tone as the ‘right’ way to write. After all, we’ve been conditioned to believe that sounding too sure of ourselves comes at a cost. It’s no secret that while a confident man is likely to be seen as ‘leadership material’, an equally confident woman is often seen as ‘bossy’. Or if we’re really unlucky, ‘bitchy’.
Is it any wonder then that so many women accept AI’s softer rewrite without question? For many, that style of writing has become an instinctive survival strategy.
At the same time, it baffles me how quickly we’ve trusted AI’s judgement over our own.
Think about it. We question our managers. We question our colleagues. We question ourselves all the time. So why do so many of us accept AI’s rewrite without stopping to ask whether it has genuinely improved our message?
Of course, this isn’t just a women’s issue. Plenty of men accept AI’s suggestions without questioning them too. The difference is that women have more at stake.
For decades, we’ve been working to redefine what professional credibility for women looks and sounds like. We’ve challenged the idea that authority needs to be wrapped in endless politeness – and that leadership is somehow incompatible with warmth.
That’s why I worry about AI becoming the new authority on what ‘good’ workplace writing sounds like. It doesn’t know which communication habits have served women well and which have held us back. It simply reflects what it has seen before.
What’s the answer then? Stop using AI to help you write at work?
Absolutely not (that ship’s sailed, baby!) But we do need to stop assuming every rewrite is serving us. And start asking ourselves before accepting the softer version:
Is this really how a confident woman should sound?