Jrcb4 The Impact of Artificial Intelligence on Learning final

Figure 4: O*NET task and skill structure for Middle School teacher occupation 
Source: Based on O*NET (www.onetonline.org
) 
The path-breaking study by Frey and Osborne asked experts in robotics and AI what are 
those technical bottlenecks that limit the automation of work tasks.
44
Using these 
automation bottlenecks as a starting point, they then asked the experts to classify a set 
of O*NET occupations based on whether automation of their tasks seemed possible. 
Those jobs that didn’t contain hard-to-automate tasks were classified as having a high 
risk of being automated. One important outcome of the Frey and Osborne study is that it 
predicted that about half of U.S. occupations is at high risk of being automated in the 
near future using current technologies. Whether this estimate is accurate or not, it still 
highlights the point that educational systems will be under considerable pressure to 
address this wide-spread change. Traditional educational planning has tried to predict the 
future demand for different types of education based on estimated labour market 
developments. Frey and Osborne show that AI will have radical impact on the labour 
market, and create discontinuities in many trends that currently underpin educational 
planning and policies. We, therefore, need to reconsider both the content and the 
functions of education in this new environment. 
43
O*NET data can be accessed online at http://www.onetonline.org/. 
44
Frey and Osborne (2013). 

---

20 
2.4.2
 
Skill-biased and task-biased models of technology impact 
Many earlier studies on the impact of computers and automation were based on skill-
biased models of technological change. In skill-biased models, jobs that do not require 
educated, experienced, and skilled workers are susceptible to automation. In such 
models, computers are expected to be used mainly for tasks that require limited skill. It 
becomes then natural to assume that to avoid unemployment people need more and 
higher-level education. In contrast, recent studies on computerization have adopted a 
task-biased approach. It assumes that those tasks that can be exactly described can be 
programmed with a computer. In these studies, occupations that consist of routine tasks 
are susceptible to automation. This has typically led researchers to assume that 
occupations that require human-like intelligence are not susceptible to automation. The 
implication for educational policy could be that education should focus on non-routine 
cognitive tasks, often labelled as 21
st
century skills. Frey and Osborne used a task-biased 
model, but they argued for a different approach. In their view, the impact on AI and 
robotics should be studied based on current technological bottlenecks. AI is rapidly 
becoming able to perform tasks that have traditionally been understood to require 
human cognition. According to Frey and Osborne, it is therefore important to ask experts 
what computers cannot do. All those tasks where technical bottlenecks do not exist may 
be automated, and if an occupation consists of such tasks, it is susceptible to 
automation.
Beyond such an occupation-level analysis, it is interesting to drill down to specific 
occupations and consider how AI could change them. In Table 1 we do this for the O*NET 
Middle School Teachers. The table lists some of the teacher’s tasks, as they are listed in 
O*NET, in their order of importance. The potential impact of AI on tasks is based on 
author’s estimate, and should be taken as indicative.