AI in your workplace: what changes and what doesn’t
Most desk workers already use AI, often with tools they bring themselves. Here’s what it really saves, where it trips up, and which parts of the job stay yours.
Video transcript
Chances are AI is already part of your working day, whether you chose it or not. So what really changes for you, and what stays the same?
In a 2024 survey by Microsoft and LinkedIn, three in four people who work at a desk said they use AI at work. And most of them bring their own tools.
In a UK government trial with twenty thousand civil servants, people said AI saved them 26 minutes a day on average, especially on searching for information and routine tasks.
Careful experiments found tasks took between a quarter and more than half less time. And in a study of customer support agents, the least experienced gained the most.
But there’s a catch, called the jagged frontier. AI can be excellent at one task and poor at a similar one. In one study, consultants using it on the wrong kind of task got more answers wrong.
So what changes? First drafts, summaries and searching all get quicker. What doesn’t? Your judgement, your relationships, and the fact that your name is on the work.
Use AI where it’s strong, check it where it matters, and stay in charge. Read the full guide below for a prompt to try today.
In 30 seconds
- Most desk workers already use AI, and studies show big time savings on the right tasks.
- It’s uneven: excellent at some tasks, poor at others that look similar, so check its work.
- AI speeds up drafts, summaries and searching. Judgement, relationships and accountability stay with you.
Maybe a colleague’s emails have become suspiciously polished, or a meeting summary now lands in your inbox before the call has even ended. AI has arrived at work, and much of it came in through the back door.
In a 2024 survey of 31,000 people in 31 countries, Microsoft and LinkedIn found that 75% of knowledge workers, people who mostly work at a desk, use Jargon busterGenerative AI: AI that creates new text, images, sound or video, rather than only sorting or labelling things. at work. And 78% of those users bring their own AI tools to work.
How much time does it save?
a day saved, on average, by UK civil servants in a 20,000-person government trial of an Jargon busterAI assistant: A chat app such as ChatGPT, Claude or Gemini that answers questions and does tasks you describe in plain language. built into their office software. The figure is self-reported.
Source: UK Government Digital Service, Microsoft 365 Copilot Experiment: Cross-Government Findings Report, June 2025Over 70% of those in the trial said it cut the time they spent searching for information and on routine tasks. They saw its limits on complex, nuanced or data-heavy work.
Controlled studies go further. They give people the same task, some with AI and some without, and measure the difference.
In three controlled studies, AI cut the time a task took by 25% to 56%
Reduction in time taken, controlled studies, 2023Each bar is a separate experiment, with its own task and people, so compare them with care. Sources: Peng and others, arXiv, February 2023; Noy and Zhang, Science, July 2023; Dell’Acqua and others, Harvard Business School, September 2023
In one study, the biggest gains went to less experienced people. Customer support agents given an AI assistant resolved 14% more issues an hour on average, and newer agents 34% more.
The catch: a jagged frontier
AI’s skills are uneven. Researchers call this the Jargon busterJagged frontier: Researchers’ name for AI’s uneven abilities: excellent at some tasks, surprisingly poor at others that look similar.: it can be excellent at one task and surprisingly poor at a similar-looking one, and it won’t warn you which is which.
In a 2023 Harvard Business School study of consultants at Boston Consulting Group, those using AI on tasks it handled well were faster and did better work. But on a task outside its strengths, they were 19 percentage points less likely to get the right answer than colleagues working without it.
So the risk isn’t that AI fails. It’s that it fails fluently, in the same confident tone it uses when it’s right.
What changes, and what doesn’t
| What changes | What doesn’t |
|---|---|
| First drafts of emails, reports and posts arrive in seconds. | Judgement: deciding what’s true, what matters and what to do. |
| Summaries of long documents and meetings take moments. | Relationships with customers and colleagues, built person to person. |
| Searching for information, policies and past work gets quicker. | Accountability: if your name is on it, you answer for it. |
In short, AI gives you a faster start, not a finished job. You still decide what’s right, what gets sent and who to trust.
Where to start
Check your organisation’s rules first, and keep personal and confidential information out of any tool it hasn’t approved. Using AI at work safely explains how.
Then pick one dull, regular task, such as catching up on a long email thread, and hand it over.
Here’s an email thread: [paste the thread]. Summarise it in five bullet points: what was decided, what’s still open, and what I need to do, with any deadlines. Quote the exact words for anything I need to act on, so I can check them.
Wondering what all this means for your job in the long run? Read our honest answer to whether AI will take it.
Check yourself
3 quick questions nothing is savedTools in this guide
- MMicrosoft CopilotMicrosoft’s assistant across Windows, Edge and Microsoft 365 apps.
- CChatGPTOpenAI’s general-purpose assistant for writing, questions, analysis and images.
- CClaudeAnthropic’s assistant, strong at long documents, careful writing and code.
- GGeminiGoogle’s assistant, built into Gmail, Docs and Android.
Sources (6)
- 2024 Work Trend IndexMicrosoft and LinkedIn, May 2024
- Microsoft 365 Copilot Experiment: Cross-Government Findings ReportUK Government Digital Service, June 2025
- The Impact of AI on Developer Productivity: Evidence from GitHub CopilotPeng and others, arXiv, February 2023
- Experimental evidence on the productivity effects of generative artificial intelligenceNoy and Zhang, Science, July 2023
- Navigating the Jagged Technological FrontierDell’Acqua and others, Harvard Business School, September 2023
- Generative AI at WorkBrynjolfsson, Li and Raymond, NBER, April 2023
Spotted a mistake? Tell us and an editor will check it.