There are calls from many forums starting from World Economic Forum to Think Tanks and research centers for uplifting education and skill for responding to the robots. We hear loud calls for skill revolution. A proposal has been around for offering higher skills and education to make 1 billion people eligible for jobs by 2030. On the other hand, these same organizations are also raising the flag of the march of robots to take over millions of jobs. For example, MGI predicts that the world will likely lose 800 million jobs by 2030. Hence, there are several questions to ponder to reason and respond to. To begin with, what does it take a robot to take our jobs? What kinds of education and skills are needed for the jobs which will be leftover by robots? Do we need more or less knowledge and skills in the Robot era?
Occupation, tasks, and execution complexity determine knowledge and skills needed
An occupation comprises the eligibility of performing a set of tasks. For example, a registered nurse needs to perform 28 tasks, as per the O*net database. Similarly, a taxi driver or chauffeur requires to perform 21 tasks. To perform these tasks, they need knowledge and skills, and most importantly a set of innate abilities. For instance, the registered nurse applies 24 innate abilities in performing related tasks. Human beings acquire the eligibility of performing tasks through education and training, experience, and innate abilities. By born, human beings earn a set of innate abilities. Fifty-two of these innate abilities have been broadly divided into four categories in the O*net database.
Through education and training, human beings earn Codified Knowledge and skill. The experience offers the opportunity to acquire additional knowledge and skill, however, in tacit form. Therefore, human beings make them eligible for performing tasks through their i. codified, and ii. tacit and iii. innate abilities.
Degree of complexity in making machines eligible for executing tasks
Machines are made of inanimate materials. As opposed to human beings, by born, they are devoid of task execution capacity. Innovators or machine designers build task execution capacity in machines through designs. Of course, with the support of technology. Hence, they have been expanding the technology portfolio. In order to build task execution capacity in machines, designers need a codified representation of underlying knowledge, skill, or human-like innate abilities. Hence, the codified capability is the first target of automation. For example, how to solve a differential equation or calculate the standard deviation of a set of numbers is a codified capability. Over the last several decades, machine designers made significant progress in building machine capability for performing tasks that mostly require codified capacity. In fact, due to this progress, we have been observing net job loss in the organization’s middle segment.
To transfer Tacit capability to machines, we need to go through the codification exercise first. It has been found to be highly complex. Hence there has been far less progress in building machines to take over those tasks requiring high-level tacit capacity. However, progress has been made in reducing the need for tacit ability, primarily through job division.
Complexity in automating humans’ innate ability is exceptionally high
On the other hand, the codification of innate abilities has been found to be the most complex exercise. Even after codification, machines designers find it extremely difficult, perhaps not impossible, to build such abilities in machines. Hence, there has been very little progress in automating tasks that require high-level innate abilities. Of course, progress is being made to simplify tasks and improve structuredness in the work environment to reduce the demand for human-like innate abilities. Despite such progress, it often needs high R & D investment to automate tasks demanding deep engagement of innate abilities.
For example, glare sensitivity is an innate ability. It’s about the ability to see objects in the presence of glare or bright lighting. To perform a driving job, its importance is more than average. Of course, human drivers successfully apply it to avoid confusion in detecting bright objects on sunny days. Unfortunately, Tesla’s autopilot system failed to detect shiny white trailers on sunny days due to a lack of adequate glare sensitivity. Consequentially, the car faced a fatal accident, causing the death of the occupant.
Decreasing knowledge and skills in the Robot era is needed
Since 1961, robots are taking over dull, dirty, dangerous, and repetitive tasks. In executing those tasks, robots keep applying codified manipulation abilities. In the recent past, there has been progress in automating tasks requiring codified knowledge. On the other hand, progress in automating tasks requiring innate ability is extremely slow. Apart from it, job division has decomposed high-level tasks into lower-level subtasks with a concentration in codified, tacit, and innate abilities. As explained, progress in automating tasks requiring codified capability is high. The progress in automating tasks requiring tacit ability is moderate. Hence, increasing penetration of automation and robots will leave those tasks for humans that will require mostly innate abilities. Consequentially, the need for humans’ codified capability, even experience, will be diminishing. Hence, human workers will require less knowledge and skills in the Robot era.
The above hypothesis appears to be already in action. For example, apparel and shoemaking are mostly demanding innate abilities. For this reason, workers in developing countries, having very little or no formal education and training, qualify for jobs in contributing to the global manufacturing value chains. Moreover, it has been also observed that there has been a weakening salary differentiation based on experience. Hence, while automation and robotics are qualifying increasing numbers of tasks, why will we require more education and training for humans to be eligible for jobs in the future? Of course, we will need more qualified machine designers for automating tacit capability and imitating human-like innate abilities in machines. But these machines (robots) are increasingly going to leave behind those tasks for humans, which will require less and less codified and tacit capability.
Are we misinterpreting the need for knowledge and skills in the robot era?
It seems that we are misinterpreting the effect of automation and robotics on education and training requirements. For example, some smartphones have supercomputers like computational power. But the knowledge and skill requirement in using them is far less than what we needed to use a supercomputer in the 1980s. On the other hand, automobiles are getting day by day technologically more sophisticated. They have increasing automation features, starting from adaptive cruise control to automatic gear changing. Do we need increasingly higher education and skill to drive those progressively advanced machines? Fortunately, the answer is no. Rather technology advancement is demanding a decreasing amount of knowledge and skill from the drivers.
There appears to be a disconnect in our thinking about how automation progresses and its implication on education and training. Prediction of 800 million job losses by 2030 appears to have very little substance. Lets’ look into our historical progress. Over the last 70 years, only about 2 million robots became technologically and economically feasible to take over roughly 10 million jobs. The prediction that automation will take over 800 million jobs appears to be highly difficult to justify. Many of those 800 million jobs will require high-level human-like innate abilities. For example, it was widely predicted that autonomous vehicles would take over millions of jobs.
After investing more than $80 billion in R&D, the progress has been stalled due to the insurmountable barrier of automating human drivers’ innate abilities. Another widely speculated area was the uprising of humanoids to take over service jobs. But the failure of Honda’s R & D program over 32 years, with an estimated cost of $500 million, to make Humanoid ASIMO qualify for elderly care offers us hard lessons.
Are we misinterpreting or misleading?
Often we think that labor-intensive tasks requiring low skill are highly amenable to automation. It appears that there has been a mistake in interpreting what it takes to make automation or robotics qualify to execute tasks. Due to such misinterpretation, have we come up with a job loss prediction figure which is highly inaccurate? Furthermore, our understating about education and training requirements for making human workers eligible in the robot era appears to be grossly inaccurate. As opposed to wide speculation of increasing knowledge and skill, human workers will mostly require innate abilities to perform the tasks that robots or automation are going to fail to take over.
