随笔0619/速记——读THE FUTURE OF EMPLOY

今天恰巧看到有人推荐《技术陷阱》，作者是Carl Benedikt Frey —— 所以先找论文看看核心观点。

看摘要：

We examine how susceptible jobs are to computerisation. To as-sess this, we begin by implementing a novel methodology to estimatethe probability of computerisation for 702 detailed occupations, using aGaussian process classifier. Based on these estimates, we examine ex-pected impacts of future computerisation on US labour market outcomes,with the primary objective of analysing the number of jobs at risk andthe relationship between an occupation’s probability of computerisation,wages and educational attainment. According to our estimates, about 47 percent of total US employment is at risk. We further provide evidencethat wages and educational attainment exhibit a strong negative relation-ship with an occupation’s probability of computerisation.

所以，具体做法是针对每个职业，考虑其可以被计算替代的可能性 —— 也是就是能被AI替代的可能性。

具体来说，如果追溯到task model：

The task model of Autor, et al. (2003) has delivered intuitive and accuratepredictions in that: (a) computers are more substitutable for human labour inroutine relative to non-routine tasks; and (b) a greater intensity of routine in-puts increases the marginal productivity of non-routine inputs. Accordingly,computers have served as a substitute for labour for many routine tasks, whileexhibiting strong complementarities with labour performing cognitive non-routine tasks.

其相应的Cobb-Douglas生产函数就可以写成：

其中：

where LS and LNS are susceptible and non-susceptible labor inputs and C iscomputer capital.

作者在其基础上进行了进一步延伸与分解：

其中：

where LPM, LC and LSI are labour inputs into perception and manipulationtasks, creative intelligence tasks, and and social intelligence tasks.

随后作者对三种工作进行了进一步的阐释 —— 这里就不展开了，总之：

Hence, in short, while sophisticated algorithms and developments in MR, building upon with big data, now allow many non-routine tasks to be automated, occupations that involve complex perception and manipulation tasks, creative intelligence tasks, and social intelligence tasks are unlikely to be substituted by computer capital over the next decade or two.

举例来说：

这里作者没有展开，我想，即便不容易被替代，也不代表完全不被影响、不需要重塑工作方式，积极拥抱AI。

作者的实证研究是基于：

The 2010 version of O∗NET contains information on 903 detailed occupations, most of which correspondclosely to the Labor Department’s Standard Occupational Classification (SOC).

方法上：

Instead, our implementation strategy builds on the literature examining the offshoring of information-based tasks to foreign worksites, consisting of different methodologies to rank and categorise occupations according to their susceptibility to offshoring (Blinder, 2009; Jensen and Kletzer, 2005, 2010).

Hence, in short, by hand-labelling occupations, we work around the issuethat O∗NET data was not gathered to specifically measure the automatability of jobs in a similar manner to Blinder (2009). In addition, we mitigate some of the subjective biases held by the researchers by using objective O∗NET variables to correct potential hand-labelling errors. The fact that we label only 70 of the full 702 occupations, selecting those occupations whose computerisation label we are highly confident about, further reduces the risk of subjective bias affecting our analysis. To develop an algorithm appropriate for this task, we turn to probabilistic classification.

嗯，只手工标注了70个职业，并将三个特性和数据库中的表述link了起来，然后进行模型训练；再然后，应用于所有职业：

Having validated our approach, we proceed to use classification to predict the probability of computerisation for all 702 occupations. For this purpose,we introduce a new label variable, z, denoting whether an occupation is truly computerisable or not: note that this can be judged only once an occupationis computerised, at some indeterminate point in the future. We take, again, a logistic likelihood...

结论是：

Turning first to the expected employment impact, reported in Figure III, we distinguish between high, medium and low risk occupations, depending on their probability of computerisation (thresholding at probabilities of 0.7 and 0.3). According to our estimate, 47 percent of total US employment is in the high risk category, meaning that associated occupations are potentially automatable over some unspecified number of years, perhaps a decade or two.

图表是这样：

要注意的是，为什么Medium Risk可以是Medium：

More specifically, the computerisation of occupations in the medium risk category will mainly depend on perception and manipulation challenges.This is evident from Table III, showing that the “manual dexterity”, “fingerdexterity” and “cramped work space” variables exhibit relatively high values in the medium risk category.

所以，也就难怪，附录中的“最不容易被替代”的职业排名，是这样子了：

结论：好好学习心理学、学习共情、学会体谅别人 —— 好好做人，不要做工具人。

后记：

- 某所长有视频解读这一文章的结论，写完才发现

- 另外，作者写的这本书，是在这篇论文的基础上，更加关注后续的挑战，包括谁受益、谁受损，受损的人会如何行动，以及整个社会应该如何行动