Daily Archives: May 4, 2017

If you can measure a jobs productivity, you can probably replace that job with a machine, so when it comes to humans in the workplace we should be measuring different things. Photograph: John Macdougall/AFP/Getty Images

A new report by the US-based National Academies of Science Engineering and Medicine suggests that not only has the automation of work barely begun but that the ways in which we measure the effects of technology on employment are inadequate to the task.

The authors argue that to understand how automation is transforming our workplaces, we need better ways of tracking technological change. Put simply, they are saying that if we are what we measure that is, if policy is driven by the information we collect then we are collecting the wrong information.

Data on many of these trends are elusive, reflecting [the] changing nature of society and the economy, and gaps in [the] statistical infrastructure, the report says.

It points out, for instance, that we dont have a regular source of information about workers in part-time and other sorts of casual employment. Nor do we have good information about investment in computer technology at either the level of the company or of any given occupation.

Also lacking is long-term information about the way in which skills within particular jobs are changing, as well as data on how effective educational practices are in preparing people for work. Such information gaps undermine our ability to respond appropriately to technological change and its effects on employment.

This is a huge wake-up call for governments and businesses around the world who are proving slow to engage with the changing nature of work and who have tended to hide behind the mantra of jobs and growth, as if that will take care of everything. It is a reminder to all of us that we are long way from understanding what the future of work really looks like.

The authors call for three new indices to be developed, tools that can be used to plug holes in conventional measures such as GDP, productivity and the unemployment rate a technology progress index, an artificial intelligence progress index and an organisational change and technology diffusion index.

They set out the parameters of each in some detail and, in so doing, open up a much-needed discussion about the data used to help form public policy.

We tend to think of measures like GDP and productivity as eternal truths of economics and, indeed, they have proved their worth over time. Nonetheless, some of them are not only reasonably recent inventions, dating from around the second world war, but are designed to measure activity in an economy of mass manufacturing, a sector increasingly being displaced by the information economy as the primary source of global wealth. This means the measures themselves are also increasingly irrelevant.

As the economics professor Richard Holden wrote: The IMF model suggests Australian unemployment falling to 5.2% in 2017 and to 5.1% in 2018. But that is a pre-2008 model of how the labour market and macroeconomy interrelate. Maybe thats still the right model but I wouldnt bet on it.

As the entrepreneur and founder of Wired Magazine Kevin Kelly has said on the subject of productivity: Productivity is for robots. Humans excel at wasting time, experimenting, playing, creating and exploring. None of these fare well under the scrutiny of productivity. That is why science and art are so hard to fund. But they are also the foundation of long-term growth.

To help understand the point Kelly is making, consider that a quarter of Britains top actors have been kept in work over the last decade by Harry Potter films. So although JK Rowling may be a billion-dollar industry, her value as a contributor to national wealth does not improve by subjecting her to a stopwatch and increased output to improve her productivity.

What Kelly is saying is that, if you can measure a jobs productivity, you can probably replace that job with a machine, so that when it comes to humans in the workplace we should be measuring different things. [Our] notions of jobs, of work, of the economy dont include a lot of space for experimenting, playing, creating and exploring, Kelly says, but those are the very skills that are likely to become more valuable in the workplaces of the future.

So the value that humans will increasingly bring to the workplace is to be not a robot, which will mean measuring our contribution by something other than inputs and outputs.

The National Academies report is not arguing for a wholesale replacement of traditional measures of economic activity but it is saying we need vast new supplementary data to better understand the ways in which new technologies are affecting the work that we do. Until we develop and implement these measures, it will mean that, on everything from education to welfare to employment policy, governments are flying blind.

The concerns of the reports authors are being driven by their belief that the technological disruption of employment has barely begun. They write: Opportunities for digitising and automating tasks are far from exhausted. In particular, the workforce will be increasingly affected as more and more cognitive tasks become fully or partly automatable ... and as advances in robotics yield enhanced physical dexterity, mobility and sensory perception in machines. These trends will almost surely change the demand for the workers performing these tasks and the nature of the organisations in which they work.

And so the sooner we start accurately measuring what is happening, the better.

Read this article:

We need to track more than GDP to understand how automation is ... - The Guardian

Editor's Note: The following is a guest post from Richard Bush, chief technology officer of NYIAX (New York Interactive Advertising Exchange).

Its hard not to think of airline pilots whenever someone insists that automation in the advertising industry will make people obsolete. Pilots, after all, have witnessed so much automation in the cockpit over the past few decades that many believe their role in flying the plane is superfluous.

However, at the same time, aviation experts are concerned that pilots who rely too much on machines risk losing key skills a warning that implies machines dont truly replace humans. Perhaps this is why a Pew survey found that a majority of Americans believe that within 50 years, robots and computers will do much of the work currently done by humans, but that very few of the respondents expect their own professions to experience substantial impacts. Is this disconnect a matter of denial, or do workers know something about their jobs that the conventional wisdom on automation fails to capture?

The answer hinges on the definitions we use. Automation isnt a monolith. Broadly speaking, there are two types of automation shaping the future of ad tech.

Within advertising, machine-to-machine automation describes the vast infrastructure of the LUMAscape. Inside that infrastructure, automation is increasingly deployed to ensure the real-time delivery of targeted ads. Without this automated infrastructure, programmatic buying and the real-time-bidding-driven ecosystem would, at most, be a shell of what it currently is.

While not without its flaws, this infrastructure is incredibly powerful. Automation not only achieves what humans cannot (the delivery of ads, across an immense media ecosystem, in real-time); automation also enables a virtuous circle of optimization as algorithms identify opportunities for improving performance. Still, like all infrastructure, no matter how sophisticated it becomes, the technology can never be totally self-propelled because people are essential for determining strategy and application.

People-to-process automation is machines learning from people. Here is where the fears expressed in the Pew study materialize. One common example of people-to-process automation is the commercial truck driver a profession machines will supposedly make obsolete in the next few years. But while automation will certainly transform trucking, why are we so certain humans wont have a role to play?

Drawing from our experience with commercial aviation, a human driver will be seen as an essential safeguard. But its not just about having a backup. What will an automated truck do if detained by local law enforcement, or a client refuses to pay upon delivery? Successful resolution requires judgment. The truck driver might join the Pony Express rider in the history books, but an automated truck driver could just as easily open the door for a new human role one thats more merchant and customer service representative than driver.

A similar story is playing out inside the media business. For all the high-tech advances around advertising infrastructure, media buying and selling remains a relatively low-tech affair of personal connections and Excel spreadsheets. People-to-process automation will free buyers and sellers from the minutia of tasks better left to automation. More importantly, once free from the grunt work of finding counterparties, tracking contractual details, and dozens of other tasks, buyers and sellers will be empowered in the same way that similar technologies have empowered Wall Street traders to standardize and scale trading operations. At that point, buyers and sellers will be far from obsolete, but the skill sets and job descriptions we associate with those positions will take on a more important and strategic outlook.

Trading has always been deemphasized in the media buyers job media buyer is the preferred title on the demand side, while sales covers the supply side. To the extent that machines are able to learn from media buyers and sellers, those people will be free to act as true media traders. Without the friction of executing each deal, traders will focus on a more holistic and forward-looking view that emphasizes planning and strategy, as opposed to managing and optimizing automated infrastructure.

The transformative effects of people-to-process automation wont be limited to the role of media trader or the function of a specific organization. The inefficiencies media traders experience determine the nature and scope of the market in which they participate. When traders are free to actually trade media contracts, the skill of media trading will begin to revolve around what they plan to buy, sell (or resell) tomorrow. Thats a significant departure from todays market one that will be felt most when it comes to planning. At the moment, planning remains a hollow promise without the insights that can be gleaned from a forward-looking market. As people-to-process automation frees up humans to create a forward-looking market, it also empowers them to use their full expertise in the marketplace.

Machine learning is a people-first endeavor. Looking at the technologies a firm is able to build around, people-to-process automation will inevitably reflect the unique skills of the people working in that enterprise. This wont happen overnight, but even if it happens faster than we think, humans will still find themselves pioneering new skills to address whatever new challenges emerge. The alternative isnt just the stuff of bleak science fiction; its a recipe for stagnation because automation without humans will optimize but will never innovate.

In the coming years, the enterprises that succeed wont be the ones that automate to workforce zero; they will be the firms that build their own tech, from their own data, to solve the challenges they find most pressing. This transformation will be widespread, but automation will be as diverse as the human experience, because just like every tool since the first stone hammer, the utility of the technology will only make sense in a context defined by humans. What that context will be, however, is a topic for the future, because automation is only now unleashing the next wave human innovation.

Read the rest here:

People power: How the next automation wave will turn media buyers into traders - Marketing Dive

If one subscribes to the idea of technological unemployment, which is the loss of employment caused by the increasing use of, and development of, technology in the workplace, then the advancing tide of technology is a frightful sight. From the relief of labour workers through machine muscle to the reduced need for human input in analysing large data, more automation is reducing the demand for human workers.

Whether one believes that technology only creates short-term unemployment with no long-standing effects (optimist) or whether one believes technologies can create a lasting decline in human employment (pessimist), it is inescapable that more and more companies are seeking to use technology to augment their business processes.

Currently, the oncoming tide of unemployment through automation is yet to make a significant splash. Mckinsey & Company, in their 2015 quarterly report Four fundamentals of workplace automation noted that in most cases of automation and computerisation, the worker was not entirely displaced, but rather experienced certain elements of their work being performed through automation. One can see such a highlight with the legal Industry, with more firms and practitioners relying on techniques and systems, such as ROSS Intelligence, to facilitate greater legal work, rather than being entirely replaced.

Furthermore, the current US unemployment rate fell to 4.5% in March 2017, from 4.7% in the previous month. This was the lowest jobless rate since May 2007, as the number of unemployed persons declined by 326,000, to 7.2 million.Furthermore, The International Bar Associations (IBA) 120 report April 2017, titled Artificial Intelligence and Robotics and Their Impact on the Workplace, found that of 1,000 manufacturing professionals, two-thirds stated they had not witnessed job losses as a result of automation, with a further 37% commenting that automation had increased job creation.

However, if there were to be a shift in the power of automation, how would employment be impacted? How would the law cope?

PricewaterhouseCoopers found, in their March 24th, 2017report that up to 30% of existing UK jobs could fall victim to automation in the next 15 years, as compared to the US (38%), Germany (35%) and Japan (21%). These impact figures ring truer in more manual and physical labour sectors such as transport and manufacturing and lower in education, health and social work. Upon this, John Hawksworth, chief economist at PwC, commented that manual and routine tasks are more susceptible to automation, while social skills are relatively less automatable. That said, no industry is entirely immune from future advances in robotics and AI.

Additionally, The IBA report pointed to the example of a German car worker costing more than 40 (34) an hour, whereas a robot was priced at between only 5 and 8 per hour.

A production robot is thus cheaper than a worker in China, the report notes. Nor does a robot become ill, have children or go on strike and [it] is not entitled to annual leave. The poignant example is the erosion of the competitive advantage of poorer, emerging economies which utilise cheaper workforces, as automated manufacturing and computer systems undercut the cost of human labour.

If these statements are held to be true, one could see a rise in technological unemployment in the coming years. Employment is more than simply a source of income, however it is a status symbol, a source of social interaction, a place to develop technical skills and abilities. Employment can provide fulfilment through to structure and security in ones life.

On the other hand, unemployment can cause mental health issues, with on average, those who are involuntarily out of work have higher levels of psychological distress than those who have work, with such illnesses becoming a barrier to re-employment.

Firstly, business restructuring, brought on by the introduction of automated work processes, will ultimately reduce the demand for a particular kind of work and could thus lead to redundancies and unemployment. This would also create new opportunities in the workforce, from the relocation of employees to retraining.

For example, an AI research programme would free up legal time, allowing junior lawyers to conduct higher-level work, and so on, thus promoting the value of work along the progression chain.

Currently, it is legally acceptable for an employer to dismiss an employee (by reason of redundancy) if the human can be replaced by a robot.Simply put, the employer need only show they have a reduced demand for employees to carry out work of a particular kind. Yet, robots lack legal personality (and thus the ability to carry liability) and thus cannot be classified as employees. Therefore, it could arise that there is both a situation of redundancy and the employer still requires that employees work to be carried out.

In this regard, it can be argued that the law should step in to either offer further protections, through the narrowing of the redundancy definition or regulating the legal status of robots. This is also relevant to issues of health and safety, particularly occupational safety hazards through working with automated machinery. This includes wearable technology such as exoskeletons, designed to enhance worker performance as these technologies could impact general health, from altering posture to straining the body.

In addition, the principle of vicarious liability and its interaction with robots creates a further legal question. This is the principle that an employer can be held liable for the wrongdoings or negligence of an employee, yet because robots cannot create primary liability nor do they hold a duty of care under the law, triggering vicarious liability would be difficult for any claimant.Thus, it is suggested that the law requires a significant review in this area, especially considering automated cars and logistics, such as unmanned drones for delivery.

This would also necessitate new skills being tutored to employees, to integrate them into the new processes. An element which the law could protect would be requiring the company to first offer retraining to displaced employees rather than simply firing them and hiring a new worker. Automation will instate the mandatory upskilling the workforce in order to efficiently and effectively utilise the systems, and thus the law could facilitate this. However, one could also argue that this is the law complicating what is essentially a business decision.

One issue that is brought to the forefront is workplace privacy. A key issue is that of behavioural analysis which can be used in the recruitment process. Employment laws in Western countries commonly protect classes of applicants, aiming to eliminate biased hiring practices. Using behavioural analysis in the recruitment process could have the unintended consequence of automating prejudice, unintentionally.

Data on behaviour and other indications of a candidates skill collected by the robotic system would be compared to similar data of successful workers already at the company or in the industry. Algorithms in these HR systems sift through resumes to find the top candidates and should the system produced biased results as against age, sex or race, then it would contravene the law.

Yet, this could easily occur if the system prioritises a certain experience or role, inadvertently discriminating against candidates if a class was more likely or less likely to exhibit those experiences. Whats more, if AI and automated systems are used to analyse candidate interviews, it could construe biological responses incorrectly.

A further consideration is the effect on working time regulations. Take for example the EU Working Time Directive, which requires EU member states to guarantee that all employees have a minimum set of rights, from the limitation of weekly working hours to an average of 48, including overtime to paid annual leave of at least four weeks per year.

The immediate consideration is in the reduction of hours that would be created by the introduction of automation, rather than being limited by the hours of work, an employee utilising automation would be limited by the amount of work available. As before, they could only work 48 hours (unless a legally-compliant waiver is signed), whereas the question is now whether there is enough work to fill those 48 hours. Consequently, this could lead to greater reliance on flexible working hours, with employers moving to contract employees in such a way as to respond to short-term fluctuations in demand for work, disrupting the traditional working week. Certainly, this can have its advantages, from enabling more time to be given to travelling workers to facilitating more familial time for those with children.

Automation also impacts collective rights and bargaining power. As aforementioned, there is currently very little protection against redundancy as a result of automation. Consequently, this also impacts collective rights as employers can, cheaply and effectively, replace swathes of workers with little ramifications, undermining social protections and job security. Whether greater security is achieved through extending individual protections or strengthening group rights, the current challenges submitted against Uber can be instructive in this regard.

Automation has raised several important questions, of both economic and political significance. Does automation lead to a rapid relocation and concentration of wealth? Is a universal minimum wage required? Should the government impose human quotas?

However, whilst these questions are certainly important, the underlying tool that can be used to facilitate the safe implementation of automation is the law. What is for certain is that laws, as they currently stand, are not adequate to address the incoming tide of automation, AI and Big Data usage.

Continue reading here:

The Law's Role in Pushing Automation Forward - The Market Mogul