Dave Birss on using artificial intelligence at work

Like most people, I have a difficult love-hate relationship with AI. I can’t live with it, but I can’t live without it either. One day, I praise it. The next, I criticize it and test what it can do, only to find myself caught in the same cycle all over again. I don’t believe AI can replace me at its current stage of development, but conversations about artificial intelligence increasingly come down to one question: will it take our jobs? Dave Birss takes this question a step further. What if we willingly hand our jobs over to AI, and do so with a smile? We might do it because our company tells us to, because everything feels urgent, or simply because we are lazy. Birss has had a career that is hard to fit into a traditional biography. He has worked in music, comedy, advertising, journalism, creative thinking, and the practical use of AI tools. “I’m a magpie. Show me something shiny and I’m gone,” he tells me. Today, he is the author of Iconic Advantage, How to Get to Great Ideas, A User Guide to the Creative Mind and Friction. Through LinkedIn Learning, he has taught more than 1.5 million professionals how to use generative AI in responsible, creative and thoughtful ways.

Before the first UEPS Masterclass, which he will lead with Boris Šurija on 20 October at Belgrade’s MTS Hall, he speaks to Vogue Adria about the future of AI, the danger of accepting AI solutions that are only “reasonably good,” prompts that can stop AI from simply agreeing with us, and why curiosity is our best defence against losing the ability to think for ourselves.

Think with AI, not like AI

Birss’s message, “Think with AI. Not like AI,” begins with a comparison to older technologies. Calculators reduced our need to do arithmetic in our heads, GPS weakened our sense of direction, and search engines made us rely less on memory. “Those were specific technologies with specific casualties. AI is a general-purpose technology, so it can take much more from us. And it will do so while telling us our average ideas are brilliant.”

Research is beginning to support his concerns. A study by Microsoft Research and Carnegie Mellon University, presented at the CHI 2025 conference, surveyed 319 knowledge workers who used generative AI at least once a week. People who felt confident in their own abilities were more likely to question and carefully check AI-generated answers. Those who felt less confident about a task were more likely to assume the answer was good enough. The researchers summed up their findings this way: greater confidence in AI was linked to less critical thinking, while greater confidence in one’s own abilities was linked to more. There is an important limitation to this study: it relied on people reporting their own behaviour. Participants described how much they felt they had thought critically, rather than reporting how many mistakes they had actually found.

Birss has reached a similar conclusion through his own experience. “What I keep seeing is the opposite of the fear that AI will take our jobs: people are happily handing their jobs over to AI, whether because of instructions from their company, imposter syndrome, deadline pressure or simply sheer laziness. And the more they do it, the less valuable they become. Every time you let AI do your thinking for you, a little of that ability drifts away and doesn’t come back.” His main concern, then, is not that AI is getting smarter, but that people are not.

An ocean of mediocrity

AI can now produce something that is, for the most part, reasonably good in just a few seconds. Yet most people stop there and do not think about how to improve it. Looking at the media and content creation industries, I get the impression that many people who call themselves “creators” see nothing wrong with copying AI-generated text word for word. They fail to notice sentences that sound unnatural or phrases that are clearly typical of AI. Is using AI making us less ambitious about the quality of our ideas? “Absolutely. But I’d place most of the blame on organisational hierarchies. People aren’t settling for good enough because they’ve lost their ambition. They’re settling because some executive somewhere has fallen in love with the idea of AI productivity and decided they want twice the output in half the time. I’ve watched this happen throughout my career, long before AI came along. If you demand that your employees work at a speed that only allows for shallow thinking, that’s exactly what you’ll get. That’s how the world becomes an ocean of mediocrity.”

Birss has written about how deciding whether something is good enough is becoming one of the most important skills in the age of AI. But how can we develop this ability to judge quality? He explains that it once came naturally as people learned a skill. “As you learned to make something, you also learned to judge whether it was good. AI has separated the two. Now anyone can produce something that looks brilliant in seconds without ever developing the discernment to recognise that it isn’t. That’s where slop comes from. It’s the ability to create content without the ability to evaluate it.” The good news is that we can develop better judgement quite easily. Birss recommends studying good work and asking useful questions. That is all it takes. “If you analysed one excellent piece of work every day and wrote down three things you could learn from it, within a few weeks you’d be unbearably good. Try it. Your colleagues might not thank you, but your work will.”

If you expect your employees to work so quickly that they can only think about things on a surface level, that is exactly what you will get. That is how the world becomes an ocean of mediocrity.

Besides good judgement, Birss believes another human skill will become essential in the future: the ability to question what everyone else is doing and choose a different path. He reminds me that the quality of a company’s work depends on the quality of its people’s judgement, and no tool can fix that. “Research shows that AI tools tend to converge on the same outputs because their training rewards the strongest patterns in the data, and the strongest patterns are, by definition, typical and obvious. Creativity is the opposite. It’s about forging your own path and finding your own voice. If you rely too heavily on AI, you’ll blend in with everyone else who relies too heavily on AI, and that’s a crowded place to be. The people who can push back against that and stand out will have an advantage in any market.”

Humans first

I recently saved a phrase I came across on social media: Created by a human is the new luxury. It is a worrying thought in a world that puts AI first. “I wince every time I hear a company describe itself as AI-first,” Birss tells me. “It’s a lovely marketing trick until you realise it’s telling employees that humans come second. Or third, once you’ve accounted for profit. Your people are still the most valuable part of your business. Neglecting them is madness.” He adds that, for individuals, putting AI first means skipping the first stage of thinking. Yet that is exactly the stage that helps us become good at our work.

Which part of the creative process would he never let AI take over? And are some tasks better suited to automation than others? “Every creative process is different, and different parts matter to different people, so I’m not going to give you a neat list of what AI should do and what humans should do. It would be wrong for half your readers.” Instead, he offers practical advice: work out which parts of your process bring you the most joy. These might be the moments when you quietly celebrate a success at your desk or want to high-five a colleague who would rather avoid it. Write those moments down. Then promise yourself that you will never hand those parts of your work over to AI, because once you do, it can be very hard to get them back. If you must use AI for these tasks, use it to improve your own skills: let it show you things you might have missed and help you go further than you thought possible. Do not let it do the work for you.

A good prompt asks for criticism

We also discuss the tendency of AI to agree with and flatter its users. How can we turn AI into a useful, reliable critic that does not agree with everything we say? Is that even possible? “It’s absolutely possible. You just have to make it less pleasant to be around. Most AI tools allow you to add custom instructions, a few paragraphs telling the tool how you want it to behave. Mine turn it into a fairly difficult colleague. I tell it not to praise me. I ask it to conduct an academic critique of its own answer, checking for reliability and accuracy. And I ask it to end every response by telling me what I might have missed and how it could have been done better. It makes every answer longer and a little more painful. It also regularly makes me rethink what I’m doing from the ground up. The criticism is often the most valuable part of the entire process. I’d rather have an honest editor than a cheerleader.”

He often explains that a good prompt is not about finding magic words. What matters most is giving AI clear instructions and enough information. “I used to give people elaborate frameworks covering everything a prompt should contain. These days, I find it simpler to say: imagine AI is an incredibly smart intern who started working for you two minutes ago. Brilliant, eager, and knowing absolutely nothing about your organisation, your task, how you work or what you consider good. Your job is to tell it. If you don’t, it will make assumptions, and the result will only be as useful as those assumptions deserve.”

For any important task, he adds a request like this to his prompt: do not make assumptions. If I have not given you enough information to produce an excellent answer, stop and ask me questions one at a time until you have everything you need. This immediately improves the results, as long as we actually answer the questions. An intern can only work with the information we give them.

Curiosity will save the world (and us)

Towards the end of our conversation, I ask Birss how we can encourage curiosity today. He tells me that curiosity was in short supply long before AI appeared. Most people simply do what their company has always done without asking why. “Curiosity is asking why, over and over again, like an incredibly persistent five-year-old. It’s irritating. And it should be. It’s the only reliable way to break out of habitual thinking.” He also believes in experimenting just for the sake of trying something new. “A surprisingly large number of my experiments turn into actual products. I’ve built well over a hundred digital tools. Many are available for free on my website, and several are about to launch as proper apps, which my accountant finds reassuring. Curiosity moves the world forward and gives you knowledge nobody else has. Without it, the world would stagnate.”

He sums up the message of his Belgrade Masterclass in one sentence: putting people first will take us further than putting AI first. “I’d print that on a mug, but what I really want is for people to put it into practice.”