原创翻译:龙腾网 http://www.ltaaa.com 翻译:G19 转载请注明出处

4.5 – What economic, legal and regulatory constraints might restrict automation in practice?


So far the analysis has focused on the technical feasibility of automation based on the characteristics of the jobs (e.g. the tasks required to be done) and their typical workers (e.g. education levels). But, in practice, we recognise that actual future levels of job automation may fall below these levels – or at least take longer to reach them than we might expect on purely technological grounds.


Economic constraints


The first important constraint here is economic – just because it is technically feasible to replace a human worker with a robot, for example, does not mean that it would be economically attractive to do so. This will depend on the relative costs of robots (including energy inputs, maintenance and repairs) relative to human workers, as well as their relative productivity.


In recent years, we have seen rapid total employment growth in the UK, driven in part by relatively subdued (or negative) real wage growth.


Furthermore, a relatively flexible UK labour market that has been open to migration from the EU in particular has made it a comparatively attractive option for companies in many sectors to expand by hiring more people, rather than incurring potentially large up-front costs by investing in new technologies such as AI and mobile robots, which will also seem relatively risky as they may not have been widely tested in practice.


Why take the risk of such investments when there is a low risk, low cost human alternative? Such considerations would apply in sectors like transport, retail and wholesale, hotels and restaurants, and food processing.


Over time, however, we would expect these economic factors to become less significant as the cost of the new digital technologies falls (quite possibly very rapidly if a robotic version of Moore’s Law turns out to apply) and they become more widely adopted, leading them to seem less risky and untested by companies that were not early adopters. But it remains highly uncertain in many sectors with low current adoption of robots when the ‘tipping point’ to much higher adoption will be reached. Organisational inertia and legacy systems may push back the timing of any such shifts towards automation even if they become technically and economically feasible.


Legal and regulatory constraints


Even if economic barriers to adopting automation can eventually be overcome, however, there could also be significant legal and regulatory hurdles to negotiate.


In the case of driverless vehicles[10], for example, the issue of who bears the liability for accidents is a difficult one to resolve – is it the car manufacturer, the manufacturer of the sensors on the car, the provider of the computer software that makes driving decisions, or some combination of these and other suppliers? What about the liability of the human passenger if he or she is expected to take manual control of the vehicle when signalled to do so by the vehicle’s computer but failed to do so? And should driverless cars be expected to satisfy higher safety standards then human drivers if that is one of their key selling points?


In the long run, we would expect these kind of legal and regulatory barriers to be overcome in those industries where automation makes economic sense and is technically feasible. But there may often be powerful vested interests arguing against too rapid an advance in automation, so it may well be that realisation of the full potential automation may occur significantly later than the early 2030s timescale we assume in this report (in line with the original FO study).


4.6 – What offsetting job and income gains might automation generate?


Another key caveat noted earlier in this article is that we have focused so far on estimating the potential job losses from automation. In practice, however, there should also be significant gains from these technologies in terms of:


• completely new types of jobs being created related to these new digital technologies; and


• more significantly in quantitative terms, the wealth from these innovations being recycled into additional spending, so generating demand for extra jobs in less automatable sectors where humans retain a comparative advantage over smart machines.


These offsetting gains are not easy to quantify, but in an earlier PwC study[11] with Carl Frey, we estimated that around 6% of all UK jobs in 2013 were of a kind that did not exist at all in 1990. Moreover, in London, this proportion rose to around 10% of all jobs. These were mostly related to new digital technologies such as computing and communications. Similarly, by the 2030s, 5% or more of UK jobs may be in areas related to new robotics/AI of a kind that do not even exist now. It is very difficult to know what these new types of jobs will be in advance, but past experience suggests that there will be some job gains from this source, albeit probably significantly less than the around 30% potential job losses from automation discussed above.

这些补偿性的工作机会不容易量化,但是在早前普华永道与卡尔弗雷的研究中[注11] ,我们估计2013年英国所有职位中,有约6%是在1990年时根本不存在的工作种类。更有甚者,在伦敦,这个比例高达10%。这些主要是新数字技术比如计算机和通讯方面的工作。与此类似,到2030年,英国可能将会有5%或者更多的现在没有的机器人/人工智能相关的领域的工作。现在难以提前预测到这都会是些什么样的工作,但是过去的经验告诉我们确实会有,虽然其数量可能会大大少于之前所讨论过的自动化带来的30%的潜在岗位流失。

The more significant offsetting factor is that these new automated technologies will boost productivity considerably over time[12] (if not, they will not be adopted on economic grounds). This will generate extra incomes, initially for the owners of the intellectual and financial capital behind the new technologies, but feeding into the wider economy as this income is spent or invested in other areas. This additional demand will generate increased jobs and incomes in sectors that are less automatable, including healthcare and other personal services where robots may not be able to totally replace, as opposed to complement and enhance, workers with the human touch for the foreseeable future[13] .

更加显著的补偿效应是,这些新自动化技术将会随着时间推而移提高生产力[注12] (如果不能做到这一点,它们也就不会在经济上被采用)。这将带来额外的收入,最初是给这些新技术背后的知识产权人以及金融资本,然后随着这些收入被花费或者投资在其他领域而扩散到更广泛的经济领域。额外的需求将给不容易自动化的行业带来新增的工作机会和收入,包括在可预见的未来机器人难以完全替代的医疗和其他个人服务,在这里人类不是作为补充和增益,而是做有人情味的工作者[注13]。

The historical evidence suggests that this will eventually lead to:


• broadly similar overall rates of employment for human workers, although with different distributions across industry sectors and types of jobs than now;


• higher average real income levels across the country as a whole due to higher overall productivity;


• but quite possibly also a more skewed income distribution with a greater proportion going to those with the skills to thrive in an ever more digital economy – this would put a premium not just on education levels when entering the workforce, but also the ability to adapt over time and reskill throughout a working life.

但是很有可能收入分配也会更加扭曲,更大的份额会被那些拥有技能可以在前所未有的数字化经济里仍能发展的人所占有—— 这将不仅仅取决于人进入职场时的教育水平,还需要有能力去适应形势并且在整个职业生涯里不断的学习新技能。

4.7 – What implications might these trends have for public policy?


The latter point raises important public policy issues. The less controversial one is that the government, working with employers and education providers, should invest more in the types of education and training that will be most useful to people in this increasingly automated world. Exactly how to identify the skills that will be required and develop the training is much more complex of course – for many people, this will involve an increased focus on vocational training[14] that is constantly updated to stay one step ahead of the robots. It also requires better matching of workers to the new opportunities that will arise in an increasingly digital economy. But the principle of investing more in education, skills and retraining seems widely accepted.


Central and local government bodies also needs to support digital sectors that can generate new jobs, for example through place-based strategies centred around university research centres, science parks and other enablers of business growth. This place-based approach is one of the key themes in the government’s new industrial strategy and its wider devolution agenda. It also involves extending the latest digital infrastructure beyond the major urban centres to facilitate small digital start-ups in other parts of the country.


More controversial is whether governments should intervene more directly to redistribute income[15]. In particular, the idea of a universal basic income (UBI) has gained traction in Silicon Valley and elsewhere as a potential way to maintain the incomes of those who lose out from automation and (to be hard headed about it) whose consumption is important to keep the economy going. The problem with UBI schemes, however, is that they involve paying a lot of public money to many people who do not need it, as well as those that do. As such the danger is that such schemes are either unaffordable or destroy incentives to work and generate wealth, or they need to be set too low to provide an effective safety net.

更具争议的是,政府是否应该直接干预收入再分配[注15] 。尤其现在全民基本收入(UBI)(G19注:社会福利)的理念已经在硅谷和其他地方获得关注,作为一种可以维持那些在自动化进程中被淘汰者的收入的潜在方式,或者说白了,让经济保持运行需要这些人的消费。然而问题是,全民基本收入计划把很多属于公众的钱给了很多并不需要人,虽然有的人是真的需要。危险在于,这样的计划要么难以承受(G19注:在经济上太昂贵)或者摧毁了人们通过工作来创造财富的热情,要么就只能标准设置的很低以致于并不能构建成一个有效的安全网。

Nonetheless, we are now seeing practical trials of UBI schemes in a number of countries around the world including Finland, the Netherlands, some US and Canadian states, India and Brazil. The details of these schemes vary considerably, and it is beyond the scope of this article to review them in depth, but it seems likely that more pilot schemes of this kind will emerge around the world and that they will come on to the policy agenda in the UK as well. For the moment, the need to reduce the UK budget deficit may be a significant barrier to any such scheme on a national level, as well as concerns about the social acceptability of giving people ‘money for nothing’. But the wider question of how to deal with possible widening income gaps arising from increased automation seems unlikely to go away.


4.8 – Summary and conclusions


Our analysis suggests that around 30% of UK jobs could potentially be at high risk of automation by the early 2030s, lower than the US (38%) or Germany (35%), but higher than Japan (21%).


The risks appear highest in sectors such as transportation and storage (56%), manufacturing (46%) and wholesale and retail (44%), but lower in sectors like health and social work (17%).

风险在运输与仓储(56%),制造业(46%) 和批发与零售(44%)等行业最高,但是在医疗与社会工作(17%)等行业较低。

For individual workers, the key differentiating factor is education. For those with just GCSE-level education or lower, the estimated potential risk of automation is as high as 46% in the UK, but this falls to only around 12% for those with undergraduate degrees or higher.


However, in practice, not all of these jobs may actually be automated for a variety of economic, legal and regulatory reasons.


Furthermore new technologies in areas like AI and robotics will both create some totally new jobs in the digital technology area and, through productivity gains, generate additional wealth and spending that will support additional jobs of existing kinds, primarily in services sectors that are less easy to automate.


The net impact of automation on total employment is therefore unclear. Average pre-tax incomes should rise due to the productivity gains, but these benefits will probably not be evenly spread across income groups. The pay premium for higher education and non-automatable skills will also probably rise ever higher.


There is therefore a case for some form of government intervention to ensure that the potential gains from automation are shared more widely across society through policies in areas like education, vocational training and job matching. Some form of universal basic income scheme might also be considered though this does face problems relating to affordability and potential adverse incentive effects that would need to be addressed.



[10] For a more detailed discussion of these issues, see PwC Strategy&’s 2016 Connected Car report here: http://www.strategyand.pwc.com/reports/connected-car-2016-study


[11] C. Frey and J. Hawksworth (PwC, 2015): http://www.pwc.co.uk/assets/pdf/ukeo-regional-march-2015.pdf
[12] See, for example, this 2015 PwC report on the potential productivity benefits of service robots:
http://www.pwc.com/us/en/technol ... ivity-platform.html


[13] Of course, eventually, we may reach the science fiction scenario where robots become indistinguishable in all ways from humans. At present, that seems likely to be much further off than the early 2030s time horizon we are focusing on in this study, though this could always change given recent rapid advances in AI and robotics.


[14] An area where the UK lags well behind countries like Germany as highlighted in our 2016 Young Workers Index report here:
http://www.pwc.co.uk/services/ec ... -workers-index.html


[15] Another idea here is the recent suggestion of Bill Gates to tax robots where these displace human labour. However, it is not clear that such a specific tax on investment in robots would be economically efficient. Other labour-saving technologies do not face such specific taxes, so why single robots out for such treatment and potentially lose productivity gains from such innovation and investment?