Artificial intelligence has swiftly developed from an exceptional technology to a part of our daily work life. ChatGPT assists employees in writing their emails, Microsoft Copilot helps during meetings by summarising conversations, and Google Gemini helps during document analysis, while other AI technologies assist our working processes by writing code, creating presentations, and automating repetitive tasks. Research shows that the adoption of AI has significantly developed, with more and more workers using generative AI every week.

At the same time, it is interesting to mention an apparent contradiction that exists in the use of AI technology in various countries.

People use AI technology widely, but there is no significant increase in productivity. Organisations invest huge amounts of money into AI technologies, yet most leaders are unable to measure their effect on their companies’ performance, and many companies report either remarkable results or no results at all.

So, what is the reason for this paradox? It is not in AI itself but in how companies use and integrate AI into their work processes.

The Advancement of AI Is Faster Than Ever Before

Generative AI has quickly become one of the technological revolutions in just a short period of time. Just a couple of years ago, AI was in the early stages of development.

Now employees are using AI for:

  • Report writing

  • Presentation design

  • Software development

  • Customer service

  • Data analysis

  • Marketing understanding

  • Idea generation

  • Summarization

  • Translating

  • Taking notes on meetings

Leaders of the technology world have embedded AI into productivity software so that AI is present in the e-mail programs, office programs, CRM systems, graphic design programs and computer programs.

In general, this means only one thing:

Today everyone can use AI.

The productivity puzzle.

For many years, economists have anticipated that technological breakthroughs will lead to considerable advancements in productivity levels.

In the past:

  • Electricity revolutionised production processes.

  • Computers changed office tasks.

  • The Internet transformed the process of communication.

  • Mobile devices made it easier to cooperate on a global scale.

  • Experts believed that artificial intelligence would be the next phenomenon to contribute to productivity levels.

However, national statistics on productivity do not seem to have changed significantly. This situation is similar to what economists referred to as the productivity paradox early on in the computer era.

In the 1980s, the economist Robert Solow pointed out that it is possible to see the impact of computers everywhere except for productivity statistics.

Many analysts assume that, similar to the concept of the productivity paradox, today AI is likely undergoing the same process.

The Reason behind AI’s Failings in Efficiency

Illustration depicting various AI tools in a workplace setting.

AI Aids with Time Saving, not Entire Processes

Most users employ AI for specific activities.

For instance:

  • Email writing

  • Spreadsheet correction

  • Meeting summarisation

  • Proposal preparation

Although such final actions will result in time saved, unless certain processes get enhanced, productivity is almost the same.

Consider a customer care representative who can use AI to provide answers to clients’ queries. The final output will still have the same time constraints in the way of approvals, reviews, and old software.

The only change in the process is that the problem is just being moved elsewhere.

Employees Use the Time to Double-Check the AI Performance

AI output is rarely perfect. Moreover, users spend time checking the correctness of information.

As a rule, people spend time on:

  • Verifications

  • Editing

  • Improvements

  • Checking calculations

  • Hallucinations correcting

According to many professionals working with knowledge, AI is often perceived as “a speed-first draft/advice” rather than a definite solution.

The Problem of Increased AI-Related Complexity

Numerous companies are presently utilising more than one AI tool all at once.

For instance, a marketing division may deploy the following:

  • ChatGPT

  • Claude

  • Midjourney

  • Canva AI

  • Adobe Firefly

  • Grammarly

  • Notion AI

Tech specialists can make use of GitHub Copilot, Cursor and multiple code correction packages, as well.

Far from facilitating and making their work easier, employees must instead make decisions on the following:

  • Which tool to consider using

  • What prompt to use to get the right assistance

  • The way to combine separate results

  • How to keep subscriptions to so many tools

The Stealth Price of Context Transformations

An overlooked productivity destroyer involves frequent context transformations.

At present, employees go back and forth between the following:

  • E-mails

  • Slack messaging

  • Meetings

  • Artificial intelligence discussions

  • Notes

  • Dashboards

With each interruption, the brain has to change the context.

Studies in cognitive psychology have proven that often switching tasks impairs efficiency and increases mental tiredness. Paradoxically, AI produces even more content to interpret or approve.

More jobs are created than eliminated

Graph showing the relationship between AI adoption and productivity levels.

The “More Content” Effect

When the process of content creation is easier, organisations tend to generate more of it.

Some typical examples include:

  • More reports

  • More presentations

  • More documentation

  • More meeting notes

  • More campaigns

  • More emails

Employees have to spend more time reading, sorting, and replying to the messages. As a result, the amount of work has increased, although it takes less time to create a single document. This situation is comparable to what happened after the email came into use

Higher Performance Demands From Managers

Though AI makes workers more productive, expectations also increase.

Rather than reducing the amount of work employees do, companies frequently ask them to:

  • Present more reports

  • Complete more projects

  • Organise more meetings

  • Give quicker answers

  • Provide more services

As a consequence, people have to work even harder while not necessarily improving their productivity.

AI Performance Increases when Workflows are Updated

Companies that are making productivity progress are not just implementing AI systems. They change workflows to accommodate AI systems.

Real-world example: Customer service

Previously, many customer support teams used AI just to provide suggestions for responses.

The productivity gains were few and insignificant.

Then, companies altered their entire workflow in the support department:

  • AI sorted tickets on its own.

  • Urgent cases went first.

  • Routine matters were managed by AI agents.

  • Human operators dealt only with difficult cases.

Thus, instead of saving a few minutes on every ticket, the whole process in the department became much faster.

Real-world example: Software development

Many developers utilise AI coding assistants.

However, the more meaningful benefits arise when AI is integrated into the complete development cycle:

  • Code creation

  • Documentation

  • Unit testing

  • Bug detection

  • Code examination

  • Security checking

As a result, instead of speeding up the coding process only, AI influences the entire software delivery process.

Businesses that change the process in software development usually experience shorter release time and faster feature delivery.

Expert Opinions on the Productivity Controversy

A team meeting with AI tools assisting in discussions and presentations.

Erik Brynjolfsson: AI Requires Changes in Entities

Economist Erik Brynjolfsson, affiliated with MIT, is among the top analysts of digital productivity in the world. He points out that advanced technologies seldom create an immediate effect in the area of productivity.

Brynjolfsson claims that organisations have to rethink the way they run their businesses, retrain employees, and reconsider their management principles before they can start receiving the major advantages provided by AI.

He states that the introduction of AI to an organisation is similar to the emergence of electricity in factories. It was only after factories were optimised for the use of electricity that productivity went up, while just the replacement of steam engines for electric motors didn’t bring any changes.

Ethan Mollick: AI Is a Strong Partner

Wharton professor Ethan Mollick believes that AI can perform better if it acts as a partner rather than a substitute. His studies show that the use of artificial intelligence improves the efficiency of workers, especially those with little experience.

However, Mollick notices that companies need to educate and experiment to benefit from AI.

The Problem of Measuring Productivity

Traditional measures of productivity are concerned with:

  • Time spent working

  • Output

  • Revenue

On the other hand, knowledge work is on a different level.

How can organisations measure the following?

  • Improved decision-making?

  • Learning speed?

  • Idea quality?

  • Teamwork?

  • Creativity levels?

If, for example, an executive makes a strategic business decision with the help of AI that counts in millions, traditional indicators are not focused on that value.

The Importance of Skills over Software

Many organisations agree that there is a notable difference in AI benefits among employees.

A proficient AI user understands how to:

  • Do effective prompting.

  • Validate outputs quickly.

  • Make good use of different tools.

  • Automate mundane processes.

  • Create scalable AI processes.

An inexperienced user, simply:

  • Accepts erroneous responses.

  • Rewrites poor outputs.

  • Uses AI erratically.

  • Relies on general prompts.

This creates an environment where two people with access to the same software might generate vastly different results. Thus, the need for training and education about AI is now as important as having the software available to one.

The upcoming AI era: From assistants to independent agents.

Currently, AI is performing tasks merely as an ally.

Coming next is an AI agent system that can manage the whole process without any human help.

In the years to come, AI agents would be able to:

  • Organize meetings

  • Collect data

  • Analyze info

  • Create reports

  • Get confirmations

  • Deliver results

If proven efficient, these systems can have an effect on productivity, being capable of automating all business processes instead of only single activities.

A lot of companies are working with agentic AI in different fields like finance, procurement, customer care, etc., but the widespread application is still in its early stages.

Problems That Organisations Should Tackle

Though AI is widely used nowadays, it is important to note that companies may still have serious challenges regarding its use:

Quality of Output

AI being only as good as the quality of its input may render an organisation crippled with useless machines. Poor quality data leads to improper decisions and ineffective machines.

Compliance

Every organisation dealing with sensitive private or financial data of their customers must ensure that they comply with locally set privacy regulations and internal procedures. Proper governance is vital for establishing confidence.

Employee Issues

Some employees may fear that AI will put their jobs in jeopardy or will change them tremendously. Therefore, without the availability of proper communication, training and upskilling opportunities, the process of adopting AI will slow down.

Conclusion

The paradox of high AI use and low productivity is less puzzling when taken into the bigger context. AI can do tasks faster than before, but it does not mean that organisations will achieve transformation just through speeding up processes.

Organisations that do things the right way make changes to their operations and invest in training their employees, improve their data management and rethink their way of working.

Such examples from history show that new technologies become a source of their largest benefits only when companies adopt processes that allow them to utilise them to the maximum.

In the coming years, there will be a shift from ‘should we use AI?’ to ‘how can we adapt our work to AI?’. As a result of advancements in autonomous AI and organisations becoming skilled in using them for business processes, the paradox of productivity will be converted into substantial economic growth from the technologies of the time.

Frequently asked questions

Why is productivity falling despite increased AI usage?

While AI is widely adopted, many companies struggle to integrate it effectively into their workflows, leading to no significant productivity gains.

What do experts say about the integration of AI in businesses?

Experts like Erik Brynjolfsson emphasize that organizations must rethink operations and retrain employees to fully benefit from AI technologies.