How to adapt to AI at work

Have you ever imagined what your workday would look like if Jill from Accounting were actually a robot? It's not a distant sci-fi scenario anymore: AI is quickly becoming an office mate. And as AI continues to integrate more deeply into our workflows, the way we spend our time at work will change significantly. The big question is: how can you prepare, and what skills should you build now to thrive in this AI-enhanced future we're living in?

Make the mental shift from doing to deciding and describing

One of the biggest changes many of us will need to make is a shift from being the one who does the work to being the one who decides and describes what work needs to be done, leaving the AI to execute it. This transition has been described as moving from creation to allocation. 

In other words, the actual hard work of production is about to get a whole lot easier thanks to AI—but the mental labor beforehand is going to be more demanding than ever. Instead of just cranking away at tasks, you'll need to think at a higher conceptual level about what you're aiming to achieve and provide clear instructions to the AI. For instance, instead of manually creating reports, you might describe the report's desired format and contents to an AI, which will then generate a draft. Then you'll have to review its contents to make sure it's accurate and edit it to meet your expectations. 

This shift makes the nature of your work more complex. You'll need to be precise in your instructions and have a thorough understanding of the big picture to ensure the AI produces the correct output. The plus side? The overall time it'll take to do something will be lower, and the quality will be better, which hopefully means some serious productivity gains.

Develop your plotter skills

In author circles, writers often identify as "pantsers" or "plotters." Pantsers are the fly-by-the-seat-of-their-pants types who write from the heart and then develop the plan as they go. Plotters are the outliners and meticulous architects: they develop the full plan first and then execute. 

I'm mostly a pantser, and having to switch to plotter-mode and write a comprehensive, big-picture prompt before diving into work with an AI can be positively painful at times. But for a lot of tasks, it's going to become a critical skill to develop proficiency with prompting and AI more broadly. So if you're a pantser, you might have to build up those skills like I do.

Find the best AI workflow for you

There are new workflow styles specifically for working with AI that have been proposed: centaurs and cyborgs. 

• Centaurs divvy up the task and get the AI to handle discrete parts.

• Cyborgs work rapidly back and forth with the AI systems, asking the AI to make changes and sculpting the output through the tool in more of a symbiotic dance. 

I've found that I switch between the two styles depending on what I'm working on, but I tend to prefer the centaur style most. It's worth it to try both and see what you like. Best practices for prompt engineering and AI workflow management are still emerging, and AI tools and their capabilities are constantly evolving. Continuous learning will be the norm as these technologies evolve, so be willing to experiment with new techniques.

You also need to decide which workflow will bring you more satisfaction in your work. As more and more work becomes primarily done by AI, with humans hired more as polishers than original creators, it will be more important than ever to hang on to that fundamental creative drive that makes work meaningful.

Spend time reviewing AI outputs and don't shut your brain off

I learned how to do long division in school, and I remember my dad, a math and science teacher, telling me that he had also learned how to do square roots by hand. I was flabbergasted. He tried to teach me, but I just reached for a calculator instead. 

In the future, I think we're going to start just reaching for an AI to do most things, and skills that were once essential will become largely obsolete, much like doing long division and calculating square roots by hand.

There's a danger that many high-skilled human tasks and areas of expertise may become less important as AI capabilities increase. For those who have honed their deep subject matter expertise over many years, that's a tough pill to swallow. But I do think that there will always be a place for human expertise, and that AI-augmented human expertise will be the most valuable of all. Even as hallucinations become less of a problem, you'll need to continue to trust your own brain. 

Because reviewing is so fundamentally different from creating, there's a risk that critical errors might slip through if we aren't vigilant enough. When you're creating something from scratch, you engage deeply with the material, allowing for a thorough understanding and connection with the output: you created it, after all. In contrast, reviewing can lead to a more surface-level engagement, which might make it easier to miss subtle but important mistakes. This means we will need to work hard to maintain focus and kick our critical thinking into high gear.

Early research has also shown that when we work with AI tools, their efficiency has the potential to be a double-edged sword. Sure, you get it done quicker, but because you spend less time making it, you also remember it less. This matters because the act of creating and engaging with the material helps to reinforce memory and understanding. When AI handles the creation, you miss out on this cognitive reinforcement, potentially leading to a weaker grasp of the content and a reduced ability to recall and apply the knowledge in future tasks. I think we all need to take this seriously for plenty of reasons. Imagine the nightmare of your boss asking you to explain something you wrote in a memo and not being able to remember what was in there. 

There's also the question of originality. If we're all using the same AI tools in the same way, will AI tools stifle creativity, leading to a homogenization of ideas and outputs? Early evidence shows there's a risk of this, so I think it's crucial to maintain a unique voice and perspective in your work.

Pull AI into your existing workflows

AI might disrupt our work—hopefully for the better—but there's no reason it has to work against our existing workflows. By adding AI into your current processes, you'll get the best of both worlds.

When you use Zapier, you can connect AI tools to the other apps you already use, so you can do things like pull the power of ChatGPT straight into your CRM, marketing tool, or any other app you use. Learn more about how to pair AI with automation, or take a look at these examples for inspiration.

Stay flexible

There are so many potential drawbacks and losses that we can't even foresee or predict yet as AI is introduced into more work processes. Maybe we don't care if people can't do square roots by hand anymore. But maybe it'll be something more profound that we'll only realize once it's gone.

I think that if we approach the integration of AI thoughtfully—and we're flexible about human-AI collaboration—there are plenty of opportunities to enhance the workplace and retain the human touch in our work. The promise has always been that AI can take over repetitive and mundane tasks, freeing up more time for humans to focus on creative, strategic, and higher-level problem-solving activities.

But even if you spend most of your day palling around with AI colleagues, there's really no substitute for human-to-human collaboration and interaction. So why not see what Jill from Accounting is up to? The robots might be coming to the workplace, but they haven't taken over the break room.

Related reading:

• Using generative AI? Forget everything you know about computers

• AI will change how you search—here's how

• How to make AI less scary for your team

• What is enterprise AI?