Thinking of Implementing AI in Your Company? Prepare for an Initial Drop in Productivity

Prepare for Chaos

For years, artificial intelligence has been promoted as a silver bullet for all business problems. AI in a company increases efficiency, lowers costs, and provides a competitive advantage. This perspective is tempting for many entrepreneurs, but at the beginning, the opposite is often true. This is shown by research conducted by experts from the University of Toronto, Stanford, and Oklahoma on tens of thousands of American manufacturing companies.

Large-scale analyses have shown that artificial intelligence often causes a drop in productivity. Efficiency may fall for a while before it rises.

“Artificial intelligence doesn’t work like a plug-in. You have to change the entire system, and that always means problems at the beginning,” explains Professor Kristina McElheran from the University of Toronto, co-author of the study The Rise of Industrial AI in America, in an interview with MIT Sloan.

You might like to read: Gen Z and the Workplace: Why the Traditional 9-to-5 Is Dead

Why Does AI in a Company Work with a Delay?

The phenomenon is linked to the “J-curve”. This is a chart showing that at the beginning of a new technology implementation, productivity decreases, but after some time, it grows rapidly, creating a shape that resembles the letter “J.” This reflects initial difficulties that eventually subside and lead to significant benefits.

The study showed that after implementing artificial intelligence, companies on average record a 1.33 percent drop in productivity. And if you include the so-called selection effect (meaning that more ambitious companies are more likely to decide on AI), the actual drop can be as much as 60 percent. This happens because using AI requires much more work than just installing a system. It involves a complete restructuring of processes. What exactly do companies need to do when implementing artificial intelligence? Among other things, they must:

• collect and analyze data,
• conduct training for employees,
• reorganize day-to-day work.

The study shows that older companies have the biggest problems. Their structures are rigid, teams are attached to old methods, and the infrastructure is often unprepared for change.

“Older companies showed a lower use of organized management procedures after AI implementation. This alone accounted for one-third of the productivity losses,” says McElheran.

Younger companies cope better with the revolution. They implement new solutions faster because they don’t have outdated habits, which makes them more flexible.

How to Survive the Initial Decline?

The researchers emphasize that it is possible to mitigate the “J-curve decline”. However, several key things need to be taken care of before artificial intelligence is even launched. These are the essential elements:

• high-quality data,
• a trained and committed team,
• processes adapted to work with AI,
• clear goals and an implementation plan.

AI in a Company is a Process, Not an Event

Artificial intelligence has huge potential, but it is not a magic spell. AI in a company means change, uncertainty, and frustration. And this can last for a long time before the first results appear. However, those who endure this initial stage can gain a lot.

Read the original article: Planujesz wdrożyć AI w firmie? Najpierw odnotujesz spadki