The Productivity Paradox: Has AI Failed to Deliver Productivity Gains?


The promise of AI is to increase productivity and efficiency for every business. According to an article by Georgios Petropoulos, a research fellow with Breugel, efficiency gains in the production process realized by AI have not contributed to productivity and GDP growth. How is this possible? Petropoulos calls this phenomenon the “Productivity Paradox,” and he has four separate explanations for it. 

1. AI has created false hopes in the workplace

AI, which is more accurately called machine learning, has been “oversold” by many tech companies as the automagical solution to very real labor problems. It is oversold because in many cases, it cannot do what the sales reps say it can. A software solution may automatically take in leads from a website, but a real person usually needs to parse through the leads to score them on quality and throw out spam. Yes, the software does do some of this, but anyone who works in the trenches with it will tell you that it does not do it well enough, and “junk” frequently slips through. 

There are limited use cases where AI is used to parse “big data” sources, usually in law firms where such solutions are needed to search for precedent-setting cases or sort through other laws and regulations. Those may increase productivity by saving lawyers and their staff research time, but this is a very niche use case of AI technology improving productivity in one sector. 

2. Bad measurement of AI productivity gains

The success of AI is usually measured in cost savings and not how it has improved productivity in a company. In his article, Petropoulos argues that even if there is mismeasurement happening and productivity gains actually exist because of  AI, it still doesn’t explain the global slowdown in GDP and productivity. It’s worth pointing out that his article was written pre-pandemic, so that doesn’t explain it either. 

3. Only Enterprise-level businesses stand to gain from AI

If you are at a bank, a Fortune 500 company, or any other large corporation, chances are high thatAI has touched the way you work in some way. However, for the 99% of American businesses that are classified as “small business,” there is little to no interaction with AI solutions. This is because these solutions tend to come with a high price tag that a smaller business cannot afford. Over time, this will inevitably change. 

4. Implementation lags of general-purpose technology

A general-purpose technology is one that has a far-reaching scope and a long implementation time. The use of computers, for example, is a general-purpose technology. Just like the AI solutions of today, only enterprise-level businesses and organizations could afford computers at the beginning of the “computer age” from the 1960’s to mid 1980’s. 

To boil this all down, AI is at its nascent stage at the moment, and may not be mature enough to offer significant productivity gains at a grand scale though there are still massive implications for what we can learn from it. All of the industries that would have had an IBM mainframe in the 1970’s now have some form of AI technology, while the “little guy” is likely left out. However, the “little guy” can probably perform just fine without it. A small business rarely has to parse through the amount of data that a bank does, for instance, and the AI solutions in use at large corporations still need a human to clean up data and provide inputs. That’s an important piece to remember - all of the data in the world is meaningless without humans behind it. Productivity data, as an example, is not valuable without oversight from people.  

Don’t count on AI to solve short-term labor problems

All the buzz about AI and automation has some managers - and shareholders - relying on it as the future of work. Where these attitudes exist, they have produced a “chilling effect” in the workplace when it comes to people power. Unless you’re talking about something like a cashier, there is no technology that can effectively replace most jobs - at least not yet. Skilled knowledge workers don’t really have anything to worry about. After all, would you replace your accountant with a calculator? 

If you think your employees are going to be replaced at some point soon, it stands to reason that you would stop investing in their long-term futures with the company. This negative attitude on the part of management hampers productivity and may even cause staff to leave their jobs well ahead of when they would be replaced with automation. If you hired smart people, you can’t be surprised if they start looking for other jobs when they sense significant changes that may impact their roles.  

It’s important for managers not to make the mistake of thinking key team members are replaceable, and especially that they are replaceable by machines. That mindset sets them up to make bad decisions about important things such as compensation and training. Any human resources person will tell you that keeping employees engaged and having a positive work environment will increase productivity, bolstering profits in the process. An employee who thinks they have no future at your company is not going to turn in the best work, and eventually no work at all when they quit. 

When AI becomes more advanced, it’s coming for managers too

Let’s look at all of the basic tasks a manager performs every day. Fills out reports, checks progress of tasks, and ensures that team members are on point and have the resources they need to do their jobs. 

Doesn’t that…sound like something a machine could do? 

Obviously, a software solution couldn’t fill in all of the blanks between those tasks, such as taking on more challenging projects that junior team members can’t handle. But it could automate a manager’s day-to-day tasks very easily. 

In an ideal world, such a solution would simply eliminate those items from a manager’s plate and provide them more time to focus on their team, but in the minds of senior management looking to cut costs, such a solution may mean the death of middle management. If you really want to make yourself sweat, consider all the good arguments this article makes for automating the role of the CEO. But given the fact that most AI is still an immature technology only accessible by corporations with deep pockets, chances are good you are safe. For now. 

Knowing how and when when to automate and use AI 

Corporate social responsibility (CSR) is no longer just a buzz term. As a cornerstone of Environment, Social, & Governance (ES&G) investing, it is here to stay for a very long time. Part of having a good CSR reputation is being good to your employees. The concepts of automating jobs and being a good corporate citizen seem to be polar opposites. But according to an article in MIT Management Sloan School, they don’t have to be. If your business legitimately can replace jobs with automation, then your employees need to be involved in the process and retrained for different positions within the company. 

If you are in the position of evaluating AI solutions to automate your workforce, cut through all the “way of the future” sales speak. When you factor in the loss of the human element, that technology may not be saving you as much money as you think. Budget may be better spent on investing in your people to increase their education level or training in a very specific area. In the end, technological solutions and data cannot replace humans when it comes to higher functions, such as critical thinking. Data should not be relied upon at face value, and always requires a human to interpret it or question its results. 

Prodoscore can highlight your team’s productivity 

Prodoscore is a productivity intelligence solution that provides insight into employee productivity. Before replacing people, consider collecting this data - conveniently assisted by machine learning - to show how good they are at their jobs. It’s hard to have a “productivity paradox” when you have the data available to tell you exactly what your employee’s working habits are. If you, like any good manager, are data-driven and rely on numbers to make decisions, Prodoscore will help you visualize the  value that your employees bring to your business every day. Contact us to find out more