Jrcb4 The Impact of Artificial Intelligence on Learning final

2.4.4
 
Trends and transitions 
Econometric studies on the effects of automation, computerization, and AI are therefore 
interesting and important but they do not capture the future well. In general terms, there 
is no obvious reason why historical trends would remain valid in socio-economic 
transitions. Econometric models may be important for understanding the present in the 
light of the past, but they can predict the future only if nothing important changes. This 
is simply because these models are based on data, and we don't have empirical data 
about the future.
49
They are, however, important because they suggest that we can 
predict the future in a very specific way: If nothing important changes, wide use of 
already existing AI technologies will imply a future that will be very different from what it 
used to be. This somewhat paradoxical result shows that, if for nothing else, this is 
because paid labour used to be such a central factor in shaping the industrial age, its 
institutions, and our everyday life. 
49
More detailed discussion on this problem can be found in Tuomi (2012).Productivity is also often difficult to 
measure when quality change and innovation are important. This will be the case for AI, in particular, as it 
does not only replace existing functions but transforms existing ones and creates novel productive tasks. 
For example, the impact of computers has been measured using "quality adjusted prices" that take into 
account developments in technical characteristics of computer equipment, such as processor clock speed, 
memory bandwidth, and number of transistors on chips. Because of the almost exponential improvements 
in many of these technical features, computers have become important factors in productivity growth. It is, 
however, not clear how such productivity measures correlate with common-sense ideas of productivity. For 
example, it is difficult to say how much more productive a person is writing texts with a computer that has 
a thousand times faster processor than two decades ago. 

---

23