Monday, February 27, 2017

"...technological unemployment will be worse than you think".

From Daniel Susskind:

A MODEL OF TECHNOLOGICAL UNEMPLOYMENT, OXFORD UNIVERSITY Discussion paper, NO. 819, February 2017 

DOWNLOAD: VERSION 22 FEBRUARY 2017

In the past 15 years a new area of the economics literature has emerged to explore the consequences of technological change on the labour market. This is the ‘task-based’ literature. The account that emerges is optimistic about the prospects for labour in the 21st century. The central argument of this paper is that this particular optimism is unjustified.

In this paper I try to make two contributions. The first is to show that the literature’s current conception of how the latest machines operate and the capabilities that this implies -- what is known as the ‘ALM hypothesis’ -- is incorrect. Put simply, the ALM hypothesis implies that while machines can perform ‘routine’ tasks, they cannot perform ‘non-routine’ tasks. Tasks are ‘routine’ when human beings find it straightforward to explain how they perform them (rather than because they are boring or dull). The ALM hypothesis argues that because human beings cannot easily articulate the rules they follow when performing ‘non-routine’ tasks, it is therefore hard to write a set of rules for a machine to follow to perform these tasks. As a result, it is claimed that these ‘non-routine’ tasks cannot readily be automated.
The problem is that the ALM hypothesis is dated. It assumes that the only way to automate a task is to understand, articulate, and replicate the way a human being performs that task. But recent technological advances in processing power, data retrieval and storage capabilities, and algorithm design, means that this constraint no longer holds. It is no longer necessary to replicate human thinking and reasoning processes in order to outperform human beings....MORE

Headline from and HT to Marginal Revolution