4 min read

Daron Acemoglu, a professor from MIT (Massachusettes Institute of Technology) and Pascual Restrepo, a professor from Boston University published a paper earlier this month, titled, “The Wrong Kind of AI? Artificial Intelligence and the future of Labor Demand”.  

In the paper, professors talk about how recent technological advancements have been biased towards automation, thereby, changing the focus from creating new tasks to productively employ labor. They argue that the consequences of this choice have been ceasing the labor demand, decreasing labor share in national income, increasing equality, and low productivity growth.

Automation technologies do not increase labor’s productivity

Professors state that there’s a common preconceived notion that advancements in tech lead to an increase in productivity, which in turn, leads to an increase in demand for labor, thereby, impacting employment and wages. However, this is not entirely true as automation tech does not boost labor’s productivity. Instead, this tech replaces labor’s productivity by finding a cheaper capital substitute in terms of tasks performed by humans. In a nutshell, automation tech always reduces the ‘labor’s share in value added’.

“In an age of rapid automation, labor’s relative standing will deteriorate and workers will be particularly badly affected if new technologies are not raising productivity sufficiently”, states professors. But, the paper also poses a question that if automation tends to reduce the labor share then why did the labor share remain constant over the last two centuries? Also, why does productivity growth go hand-in-hand with commensurate wage growth?


Professors state that in order to understand this relationship and find an answer, people need to recognize different types of technological advances that contribute to productivity growth. They state that Labor demand has not increased over the last two centuries due to technologies that have made labor more productive rather due to new technologies that have eliminated labor from tasks in which it previously specialized.

The ‘Wrong kind of AI’

Professors state that economists put a great deal of trust in the Market’s ability to distribute the resources efficiently. However, there are many who disagree. “Is there any reason to worry that AI applications with the promise of reinstating human labor will not be exploited and resources will continue to pour instead into the wrong kind of AI?state the professors.

Professors listed several reasons for market failures in innovation with some specific reasons that are important in terms of AI. Few of these reasons are as follows:

  1. Innovation creates externalities and markets do not perform well under such externalities.
  2. Markets struggle in case there are alternative and competing technological paradigms.  This is because in case a wrong paradigm moves ahead of the other, it can get very difficult to reverse the trend and benefit from the possibilities offered by an alternative paradigm.
  3. The research paper states that there are additional factors that can distort choices over what types of AI applications to develop. The first one being that in case of employment creation having a social value beyond what is in the GDP statistics, this social value gets ignored by the market.
  4. Recently, the US government has been frugal in its support for research and its determination to change the direction of technological change. A part of this change is because of:
    1. reduction in resources devoted to the government for support of innovation.
    2. the increasingly dominant role of the private sector in setting agenda in high-tech areas. This shift discourages the research related to future promise and other social objectives.

To sum it up, professors state that although there is no ‘definitive evidence’ that research and corporate resources are getting directed towards the wrong kind of AI, the market for innovation does not provide a good enough reason to expect an efficient balance between different types of AI. Instead of contributing to productivity growth, employment, and shared prosperity, automation advancement would instead lead to anemic growth and inequality.

“Though many today worry about the security risks and other..consequences of AI, we have argued that there are prima facie reasons for worrying about the wrong kind of AI from an economic point of view becoming all the rage and the basis of future technological development”, reads the paper.

Read Next

Stanford University launches Institute of Human Centered Artificial Intelligence; receives public backlash for non-representative faculty makeup

Researchers at Columbia University use deep learning to translate brain activity into words for epileptic cases

Researchers discover Spectre like new speculative flaw, “SPOILER” in Intel CPU’s


Subscribe to the weekly Packt Hub newsletter. We'll send you the results of our AI Now Survey, featuring data and insights from across the tech landscape.