Language: English
Request a service

How Much Time Does AI Really Save at Work? What the Research Says

Claims about AI productivity are everywhere. Here is what controlled studies actually found about time savings in customer support, writing, consulting and coding, and how to apply it in your own business.

Hasan Gültekin5 min read

“AI will save you hours every day.” You have probably heard this many times. But how much time does AI really save at work, and for which tasks?

Instead of vendor claims, let’s look at what controlled research found. Then we translate it into practical guidance for your business.

The short answer

For the right tasks, well-designed studies show substantial gains: tasks completed roughly a quarter to over half faster, with similar or better quality. For the wrong tasks, AI can make results worse. The difference between the two is the most important thing to understand.

What the research found

Customer support: 14% more issues resolved per hour

In “Generative AI at Work” (Brynjolfsson, Li and Raymond, NBER Working Paper 31161, 2023), researchers studied thousands of customer support agents who got access to an AI assistant that suggested responses.

  • Productivity, measured as issues resolved per hour, rose 14% on average
  • Novice and lower-skilled workers improved by 34%
  • The most experienced agents saw minimal impact

Takeaway: AI helps spread the knowledge of your best people to everyone else. New employees get up to speed much faster.

Writing tasks: much less time, better quality

In an experiment by Noy and Zhang (MIT, published in Science in 2023), professionals such as marketers, grant writers and HR staff completed realistic writing tasks, with or without ChatGPT.

  • Time spent dropped substantially. The working paper reports about 37% less time on the measured task
  • Output quality, as judged by evaluators, went up
  • The gap between weaker and stronger writers narrowed

Takeaway: first drafts of emails, reports, job descriptions and press releases are a strong use case.

Consulting work: faster and better, until it is not

In “Navigating the Jagged Technological Frontier” (Dell’Acqua et al., Harvard Business School Working Paper 24-013, 2023), consultants at Boston Consulting Group did realistic tasks with and without GPT-4.

For tasks inside the AI’s capabilities:

  • 12.2% more tasks completed
  • 25.1% faster
  • Significantly higher quality

For a task deliberately chosen to be outside the AI’s capabilities:

  • Consultants using AI were 19 percentage points less likely to reach the correct answer

Takeaway: AI’s abilities are “jagged”. It is excellent at some tasks and confidently wrong at others that look similar. People need to know where the boundary is.

Software development: 55.8% faster on a defined task

In “The Impact of AI on Developer Productivity: Evidence from GitHub Copilot” (Peng et al., 2023), developers were asked to write an HTTP server in JavaScript.

  • The group with the AI pair programmer completed the task 55.8% faster

Takeaway: well-defined coding tasks benefit strongly. Real projects include reviews, design and debugging, so overall gains are usually lower than in a lab task.

Summary table

Study Work type Main result
Brynjolfsson, Li, Raymond (2023) Customer support +14% issues resolved per hour; +34% for novices
Noy & Zhang (2023) Professional writing Much less time, higher quality
Dell’Acqua et al. (2023) Consulting +12.2% tasks, 25.1% faster, higher quality inside the frontier; worse outside
Peng et al. (2023) Software development 55.8% faster on a defined task

Why your results may differ

Lab and field studies measure specific tasks. In real businesses, results depend on:

  • Task fit: drafting, summarising, classifying and searching benefit most
  • Data access: AI that can see your documents, orders and CRM saves much more time than a generic chatbot
  • Training: people who learn good prompting and verification get much more value
  • Review overhead: time saved on drafting can be lost if every output needs heavy correction
  • Workflow integration: copying text between tools wastes time; integrated AI does not

Where businesses typically save time

Task Why AI helps
Writing first drafts Removes the blank-page problem
Summarising meetings, documents, tickets Reading long content is slow
Answering repetitive questions Same answers, many times
Extracting data from documents Manual data entry is slow and error-prone
Searching internal knowledge Finding the right document takes time
Translating Fast first versions for review
Coding and scripting Boilerplate and repetitive code

How to measure time savings in your own company

  1. Pick one process (for example answering support emails)
  2. Measure the baseline: time per task, volume, quality
  3. Introduce AI for a pilot group
  4. Measure again after a few weeks of use, not on day one
  5. Include review time in the calculation
  6. Check quality: errors, customer satisfaction, rework

My guide on calculating the ROI of AI automation shows how to turn this into a business case.

Frequently asked questions

Does AI save the same amount of time for everyone?

No. Research repeatedly shows bigger gains for less experienced workers on tasks where AI performs well. Experts gain less, but still benefit from faster drafting and research.

Can AI make people slower?

Yes, when it is used for tasks it does not handle well, or when its output is trusted without checking. Training on where AI fails is as important as training on where it helps.

Is the time saved used productively?

That depends on management. Saved time can go to more customers, better quality or new work. Without a plan, it often disappears into other low-value tasks.

Want to find where AI saves your team the most time?

I help businesses identify, measure and implement AI use cases that deliver real time savings. See my AI integration service or request a consultation.

Sources