As communications professionals, we are wordsmiths. We obsess over sentences, headlines, and whether something sounds like us. So when an AI tool produces a paragraph of corporate enthusiasm, our instinctive response is that we could have written it better ourselves. 

Often we could. That is partly why I resisted AI for longer than I’d like to admit. But as I’ve started using AI tools more regularly in my day-to-day work within a small in-house marketing and communications team, I’ve stopped asking whether AI writes better than we do. The more useful question is whether it gives us the time and headspace to communicate better. 

The biggest gains have come from the least glamorous tasks. Using an approved AI platform, I rely on it to summarise long email chains, tidy spreadsheets, organise rough notes and keep social media schedules on track. It’s just as handy for structuring agendas for routine catch-ups and turning hectic weeks into manageable to-do lists. 

The brand risks of AI-generated content

AI has earned its place in research and media monitoring too. It can sort coverage into themes and flag issues worth exploring, though the important articles still need proper reading. It’s also a useful second pair of eyes on awards entries and plans. I often get more from asking “What have I missed?” than “Can you write this?” 

None of these tasks is the job itself. Our value lies in judgement and creativity, both of which need time. If AI hands some of that time back, it’s doing something worthwhile. 

But it isn’t all positive. One area where its limitations become particularly clear is helping a spokesperson find the right words. AI can condense notes into a useful starting point, but what someone says publicly needs to reflect their own views, experience and way of speaking. Ask AI to put words in their mouth, and you tend to get something polished and plausible that could have come from any organisation and ultimately says very little. 

That’s where the expertise of press officers and communications professionals comes in: knowing an organisation, its people and its audiences well enough to recognise when something rings true, or when it might land badly with a journalist. 

A supposedly quicker draft loses its advantage if you have to spend longer stripping out generic phrasing. The view has to come from the person. Our job is to make sure it’s heard clearly. 

Then there’s AI slop: generic posts and images that feel generated rather than considered. An analysis by AI-detection firm Pangram flagged around 41% of long-form LinkedIn posts in its sample as fully AI-generated. This is not a representative measure of the platform, but most of us recognise the pattern in our feeds. 

Why communications professionals need to teach judgement, not just AI skills

For communications teams, the newer risk sits outside the team. In any large organisation, plenty of content is published without passing through communications. Colleagues rightly want to share their achievements, but AI now lets anyone generate posts and images in seconds.

It’s easy to imagine an image showing a workplace that doesn’t exist, a distorted logo or people who were never there, or copy that exaggerates what happened. 

This kind of content can undermine a carefully built brand. But stopping colleagues from communicating would be the wrong response. A better approach is to make doing it well the easiest option, with authentic photographs, straightforward templates and guidance on when to ask for a second pair of eyes. 

Confidentiality matters too. Using an approved AI tool doesn't give you blanket permission to upload sensitive or unpublished information. The rules about what can be shared still apply, however sophisticated the technology becomes. 

Judgement is one of the most important skills I try to develop in junior colleagues. Whether they like it or not, AI will be a big part of their careers, so I encourage them to use and challenge its output. Would a journalist believe this? Is it what the audience needs to hear? Being able to judge an answer will matter more than generating one. 

Writing for people and machines

I was fascinated by a Digiday piece about The Economist, which is experimenting with marketing pages structured for AI agents. My first thought was whether communications teams such as mine should be doing the same. 

Then I saw what it chose not to do. It paused a trial to turn its 300-page style guide into an automated copy checker. Its vice-president of generative AI was blunt: “Nobody wants to read an AI-written Economist.” 

That tension feels right for communications. We may soon be writing for two readers: the journalist or colleague, and the AI tool that summarises our words for them. We need to be discoverable by machines, but keep the words human. 

So what would I tell a head of communications starting tomorrow? Begin with low-risk tasks, use approved tools, follow confidentiality policies and check everything intended for public consumption. Test the results honestly, including the time spent fixing them, and don’t reward output for its own sake. 

AI is certainly making us faster. Whether it makes us better depends on what we choose to do with the time it gives back.

Jonny Bisby is a senior marketing and communications manager at a global facilities management company, and has previously worked in roles at the BMA and the Royal Pharmaceutical Society (now the Royal College of Pharmacy).

Further reading

Why AI-generated reputations are never standing still

Why AI is changing the rules of reputation management

Is AI phasing out the industry expert?