Sadie Crabtree has worked at Prostate Cancer UK since 2012, joining as digital engagement manager and rising through the communications team to assistant director of communications. She led development of the online risk checker tool that transformed the organisation’s approach to public health campaigns, designed award-winning research into the barriers that stop men acting on prostate cancer risk, and she currently leads its AI Interest Group. Before moving into the UK charity sector, she spent a decade in the United States as a communications lead for campaigning organisations and trade unions.

Don’t leave AI to your IT team – get in the room with them. It’s not a tool like other tools. It’s conversational, it’s natural language, nobody needs training to use it. And most of what makes a chatbot work is opaque, so a trouble-ticket approach to problem solving doesn’t get you very far – a chatbot doesn’t throw error messages when it’s making mistakes. What you need alongside the IT mindset (focused on governance, security and stability) is a digital transformation mindset, actively looking for opportunities and how quickly you can find the value. That’s why our interest group and our ongoing conversations about AI and culture are deliberately cross-organisational, not a technology project.

AI skill development is closer to people management than tool training. You can’t train someone in an afternoon to get good output from AI, because what you get back depends on the quality of what you put in, and on learning its weaknesses when it can’t tell you them itself, and figuring out how to get the best out of it. That’s closer to learning how to manage a person than learning how to operate a machine, and it comes from experience, ideally shared: people showing each other what worked and what didn’t.

Human conversations are more valuable than ever – and transcripts are a tangible asset. One of the best uses I’ve found: record the conversation you’re already having, and let AI turn it into the structured brief someone would otherwise spend an hour writing afterwards. Delivery teams want a brief to be clear and complete upfront, but that clarity only ever comes from a proper back-and-forth. People are used to treating conversations as ephemeral. I want people to treat them as an asset with real value as rich context for AI tools to accelerate your work.

Technology should bring people together, not stand between them. The bad use of AI is the one that lets you avoid a conversation you should be having with a colleague or a manager. The good use removes the friction from coming together, and makes sure meetings always tick more tasks off your list than they create. That distinction is a big part of whether AI helps an organisation or just adds to the noise.

When being right feels good, treat that as a cue to check yourself. I was a lot more sure of myself in my twenties, but I like myself more now. Because I’m less committed to my own ideas, I trust myself more to get to the right answer. When believing something feels really good, that’s a feeling to be suspicious of, not to hide behind. Often it’s when I’ve cast myself as the hero and someone else as the villain. That shuts down your ability to reflect.

Technical accuracy in what you say has little to do with the accuracy of what the audience hears. Sometimes what the audience takes away is the opposite of what you thought you said. We tested a message that told men to talk to their GP “even if you don’t have symptoms,” since prostate cancer often has none. What the language in the test actually did was reinforce the idea that symptoms are main reason to go, and made men less likely talk to visit their GP. Designing public messages to satisfy internal stakeholders rather than meet the audience need can do real harm.

Give people a simple door in, then bring them the nuance. The honest answer on prostate cancer risk is complicated: there’s no national screening programme, and testing has both risks and benefits. You can’t put that on a poster. What you can put on a poster is “check your risk.” Build a simple entry point, then have the nuanced conversation once someone’s walked through it.

Slow down, and understand the problem deeply before you try to solve it. If I could tell my younger self one thing, it wouldn’t be a career tip. It would be to listen properly to people who don’t agree with me, because they know things that are worth understanding. I spent a long time going as fast as I could with my own ideas. I’ve learned to pay more attention. I spot patterns and make a lot of connections. I have a lot of ideas while people are talking to me and it’s easy to grab hold of one and get attached to it, and jump in with an solution. Sometimes I’m wrong because I haven’t listened enough to fully understand the problem. Sometimes I’m right but it would have been better to let the other person work it out for themselves. Either way, it’s usually more valuable to listen than to talk. What was important for me to realise is that my thoughts are abundant and therefore cheap. I can ignore them and the good ones will come back to me. It’s other people’s thoughts that are important – if I don’t listen I might miss the chance to understand something that’s critical to solving the problem.

—

Cause & Effect is a series from Hope, a charity branding and marketing agency, in which leading figures who have been involved in building and promoting good causes tell us what they’ve learned from their experiences. Interview by Michael Isaacs.