Lorem ipsum dolor sit amet, consectetur adicing elit ut ullamcorper. leo, eget euismod orci. Cum sociis natoque penati bus et magnis dis.Proin gravida nibh vel velit auctor aliquet. Leo, eget euismod orci. Cum sociis natoque penati bus et magnis dis.Proin gravida nibh vel velit auctor aliquet.

  /  Project   /  Blog: It Takes Two to Tango: Machines And Humans in the Age of Automation

Blog: It Takes Two to Tango: Machines And Humans in the Age of Automation

Go to the profile of HireHand
Exhibit 1: The story of technological advancement — a partial take?

1) Introduction

Investors ploughed $50 billion into automation start-ups last year¹. While the industries and specific angles of these businesses differ, all share the belief that machines represent the future with human-led activity a vestige of the past. Indeed, a cursory read of these businesses’ funding round releases highlights two points: first, just how fervently these “Apostles of Automation” believe in the redemptive abilities of artificial intelligence; second, how uninterested they are in the practical ways in which humans will continue to engage with technology as it develops.

When the story of technological change is told, the rapid advancements in hardware (and software) frequently dominate the narrative. The above tri-glyph (see Exhibit 1) outlining the progression of the washing machine over the past 200 years from scrub board to the rotary drum to the automated electronic version is typical. An equally important and frequently overlooked part of the story of technological change, however, is the shifting ways humans engage these machines to reach their shared objective. In the case of the washing machine, the rapid human advancement is from eight hours of pure elbow grease (on the scrubbing board), to four hours of turning the crank (on the rotary drum), to two minutes of preparing and starting the load (automated electronic version). Told from this angle, the story is one of time-saving and reduced manual labour but continued human involvement. And if humans are still required for as basic a task as washing this late into the innovation cycle, we will undoubtedly continue to play a meaningful role in the more complex areas of working life that today find themselves in the early stages of automation.

To the extent that the Apostles of Automation consider workers in their master narrative, they envision a post-work utopia where humans occupy themselves with creative writing classes and the collection of their Basic Income payments. Needless to say, this is an incorrect and costly miss that belies historical precedent and common sense. To employ another metaphor, even if one partner is a machine (and the most advanced one in the world at that), it still takes two to tango. Given this fact, our thesis is that the choreographers of this constantly evolving dance will deliver more economic (and social) value than all the Apostles combined.

This article will look at how advanced automation is affecting shift-work in blue collar industries and how it is likely to influence the white collar world.

2) How automation ACTUALLY changes shift work in blue collar sectors

Our review of how the pattern of human work changes with increases in automation will focus on blue-collar industries given the relatively advanced stage of automation in these industries. Let’s see how this dance is working in practice. (See Exhibit 2).

Exhibit 2: How increased automation (and cost pressures) actually influence the shift patterns of workers

2A) When a business is considering automation, the unit of activity at which they consider the costs and benefits is at a task level as opposed to job level. As one automation expert notes, “Automation does not replace the human in the process. It replaces the robot in the human.”² And once machines are able to take over tasks for which humans are presently used, the decision to automate is not a fait accompli. Rather it is a complicated cost-benefit analysis including, but not limited to, cost, convenience and business capability to implement. And only rarely, to the quote above, is a human completely a robot, or does the entirety of a worker’s job consists of tasks more suitable for machines, thus leaving a number of human-led tasks in place.

2B) Cost considerations are primary driver of whether a shift worker will maintain the same activity level by having those automated tasks replaced by additional manual ones or whether her activity will be reduced. An unhelpful platitude often uttered by “change management” experts to the front-line is that automation will free them up to pursue more “high-value work”. The unsaid assumption here is that the overall objective of the exercise is not purely cost reduction — which for many low-margin blue collar industries is precisely the case. In such cases, the offer of additional work is rare — the total value to the business is in cost reduction. A caveat to this cost reduction story is the fact we are seeing more favourable (for the worker) dynamics emerge in select regions. In the United States, for instance, blue collar workers are more scarce than white collar workers for the first time in decades and are experiencing proportionally higher wage growth³ . In such environments, many businesses will maintain workers at full employment in order to retain them.

2C) If a shift worker’s activity has been reduced the remaining tasks are often dispersed unevenly, leading to the fragmentation or splitting of shifts. This is particularly the case in industries such as food retail where business activity is not controlled but rather responsive to customer demand. In such instances, businesses will focus their automation efforts on reducing the need for labour in low activity periods (in order to run a more lean operation) and continue to rely on workers in times of high activity. Hence why Pret a Manger in London has up to five times as many people on shift at prime breakfast and lunch hours relative to the quieter periods in between meals.

2D) Shift workers are still required in rapidly automating industries but their hours are likely to be reduced and their shifts split. Exact statistics are hard to come by, but it is estimated that the usage of split shifts has increased 3-fold in the London food retail sector in the last 5 years. A product of this phenomenon can be seen in macroeconomic trends in the US and UK where employment levels are at all-time highs but so are the number of individuals categorised as “working poor” (i.e., in employment but living below the poverty line — an estimated 16M people in the US and UK)⁴ .

Picture 1: The dance of the “touch screen” and the greeter

An illustration of these four factors can be seen with McDonalds and its adoption of touch-screens for ordering (See Picture 1). Contrary to public narrative, automation has not displaced the job of an entire class of workers (the front of house) but rather a selection of tasks (taking orders and payment) of the many overall (others including greeting the customer and collecting food) the worker has done (See 2A). As this innovation was part of a cost reduction exercise, the front of house worker has not been given extra work but rather has experienced a reduction in activity (See 2B). McDonald’s continues to employ front of house team members but increases the number utilised around its lunch and dinner rushes (See 2C). To organise shifts to this need, McDonald’s increasingly uses 2-hour flexible shifts around these peaks (See 2D).

The emotion that likely comes to the mind of the typical white-collar reader is one of sympathy. You may, however, want to tap into previously untouched stores of empathy as this phenomenon is on your doorstep.

3) From Blue Collar to White Collar — More similar than you would think

Automation is already requiring white collar workers to deal with the strikingly similar effects on blue collar workers addressed above, despite differences in the nature of work (less physical) and driving technological innovation (more machine learning than machine). Indeed, we are already seeing automation in areas unimaginable to most people five years ago: from robot-assisted surgeries, automated anesthesiologists and machine learning diagnoses; AI-driven insurance claims assessments and online therapy and chatbot driven customer service. Areas of service that were previously assumed to be beyond the reach of technology are now being revolutionised, and this trend will only continue.

3A) It’s all about the tasks in White Collar as well

One reason that this trend is catching many by surprise is that — as with shift work (See 2A) — automation is affecting these workers on a task by task basis rather than rendering their entire jobs obsolete. As a joint report from MIT and Carnegie Mellon on machine learning notes, “Few — if any — jobs can be fully automated using machine learning…machine learning technology can transform many jobs in the economy, but full automation will be less significant than the re-engineering of processes and the reorganization of tasks.”⁵ It is important not to sleep on this trend, however, as far more tasks are already capable of automation than the isolated ones in relatively isolated industries addressed to date. Indeed, McKinsey estimates that “45 percent of work activities could be automated using already demonstrated technology” with the main areas ripe for disruption being “analyzing reports and data to inform operational decisions, preparing staff assignments, and reviewing status report(s).”⁶ Given that white collar automation isn’t theoretical or dependent on hypothetical technology — but is already present in many quarters — the question becomes not if, but when, its disruptions will be felt on a similar scale as in the blue collar world?

3B) The prize for automation in White Collar work is larger because it is not all about cost

While cost reduction (the machines can do it cheaper!) is the primary reason for businesses to automate blue collar tasks (See 2B), there are additional considerations at play in the white collar context. The space for additional consideration comes from the growth posture of these industries — the services sector in particular — which account for 70% of US GDP and are responsible for the majority of economic growth in western economies. As a result, the competitive advantage of automation is more about capturing additional share of a growing pie. In this context, the platitudes of “change management” experts we mocked in the blue collar context can apply. Businesses genuinely do want to free up their valued workforce from monotonous activities to design, build, and innovate. And the heights that a star performer can reach in the knowledge economy is disproportionately high.⁷ Put simply one outstanding saleswoman or software developer is the equal of 20 average ones. The aims of automation, in this context, are to replicate, surpass, and scale that star performer’s achievements. Hence, why the thrust of Amazon’s automation efforts are in white-collar jobs like demand forecasting and pricing conversations due to “algorithms that take in thousands of inputs and are always running smarter than any human.” A caveat to this more optimistic take on white-collar automation is that it relies on industry growth — the pie getting bigger. The recent stagnancy of the services sector in the UK, however, suggests a cost-oriented mindset may not be far away.

3C) What does this ACTUALLY mean for the White Collar worker

Like the blue collar worker, the majority of the tasks a white collar worker completes can be automated, but, unlike the blue collar worker, there is a higher likelihood of those automated tasks being replaced by other high-value work. Given this, what does automation in the white collar industries mean for the work patterns of its workforce? Will those patterns stay the same given a never-ending stream of high-value activities to pick up? Will automation and cost pressures eventually dry up that stream, requiring a more fragmented work schedule built around required interactions with technology (See 2C)? The answer likely lies in between — employment in a job that looks more like shift-work than the salaried position today. Blue collar workers — under significantly more pressure today — would take this outcome in a heartbeat. White collar workers, however, have typically enjoyed more flexibility and autonomy in their work (See Netflix). Thus even a taste of the fragmentation, fewer hours worked, less predictable work patterns and lower salaries that many blue-collar workers have swallowed whole (See 2D) will be poorly received at best and hugely disruptive at worst.

This begs the question of how quickly we are likely to see this disruption? Given that the technology for disruptive automation in the white collar sector already exists, the profit and market share incentives mean that automation of any white collar industries facing competition could happen much, much more quickly than we are seeing in blue collar work. Think cascade, not steady stream. What will drive the speed of automation then is less a series of cost and benefit analyses and more a concern of “keeping up with the Joneses”. All that is required is an industry leader making the first move.

4) Conclusion

Automation occurs at the task — not job — level; cost considerations — of lack thereof — determine if those tasks are replaced; even in cases where they are not and shifts become fragmented, workers are still required (see McDonald’s); these maxims apply to both blue collar and white collar industries regardless of differences in the nature of automation.

If you accept these assertions, then it must follow that humans will be an essential dance partner to even the most advanced machines for the foreseeable future. This partnership can lead to greater human fulfilment — more flexible work patterns, higher hourly pay, and a way of working in which advanced technologies augment human endeavour to chart new frontiers. Or it could result in worker exploitation, widespread economic insecurity, and the snuffing out of human agency — picture your office as the Amazon fulfilment center from hell. Steering our future course away from the latter to the former will require our best minds (and technology) dedicated to designing the ideal dance between humans and machines instead of denying the need for such choreography in the first place.


[1] https://www.pwc.com/us/en/moneytree-report/moneytree-report-q4-2018.pdf

[2] https://blog.kinaxis.com/2018/06/is-automation-killing-white-collar-jobs/

[3] https://www.bloomberg.com/news/articles/2018-12-13/blue-collar-worker-shortage-turns-u-s-labor-market-on-its-head

[4] https://www.jrf.org.uk/work/in-work-poverty

[5] https://www.forbes.com/sites/joemckendrick/2018/08/14/artificial-intelligence-will-replace-tasks-not-jobs/#1652dcf1a7fa

[6] https://www.mckinsey.com/business-functions/digital-mckinsey/our-insights/four-fundamentals-of-workplace-automation

[7] For more information on the economic literature behind this phenomenon, see Sherwin Rosen, The Economics of Superstars The American Economic Review, Vol. 71, №5. (Dec., 1981), pp. 845–858.

Source: Artificial Intelligence on Medium

(Visited 3 times, 1 visits today)
Post a Comment