automatic News for 06-01-2019

Site Reliability Engineering

While we believe that software-based automation is superior to manual operation in most circumstances, better than either option is a higher-level system design requiring neither of them-an autonomous system. There’s an additional benefit for systems where automation is used to resolve common faults in a system. 29 For truly large services, the factors of consistency, quickness, and reliability dominate most conversations about the trade-offs of performing automation. The context for our automation is often automation to manage the lifecycle of systems, not their data: for example, deployments of a service in a new cluster. Although all of these automation steps are valuable, and indeed an automation platform is valuable in and of itself, in an ideal world, we wouldn’t need externalized automation. The people most affected by automation bugs were no longer domain experts, so the automation became less relevant and less competent. 

A team not running automation has no incentive to build systems that are easy to automate. A product manager whose schedule is not affected by low-quality automation will always prioritize new features over simplicity and automation. Another way to understand the development of our attitude toward automation, and when and where that automation is best deployed, is to consider the history of the development of our cluster management systems. 31 Like MySQL on Borg, which demonstrated the success of converting manual operations to automatic ones, and the cluster turnup process, which demonstrated the downside of not thinking carefully enough about where and how automation was implemented, developing cluster management also ended up demonstrating another lesson about how automation should be done. Analogous discussions about the impact of automation in the noncomputer domain-for example, in airplane flight33 or industrial applications-often point out the downside of highly effective automation:34 human operators are progressively more relieved of useful direct contact with the system as the automation covers more and more daily activities over time. We, too, have experienced situations where automation has been actively harmful on a number of occasions-see Automation: Enabling Failure at Scale-but in Google’s experience, there are more systems for which automation or autonomous behavior are no longer optional extras. 

In public, many executives wring their hands over the negative consequences that artificial intelligence and automation could have for workers. In private settings, including meetings with the leaders of the many consulting and technology firms whose pop-up storefronts line the Davos Promenade, these executives tell a different story: They are racing to automate their own work forces to stay ahead of the competition, with little regard for the impact on workers. They crave the fat profit margins automation can deliver, and they see A.I. as a golden ticket to savings, perhaps by letting them whittle departments with thousands of workers down to just a few dozen. Few American executives will admit wanting to get rid of human workers, a taboo in today’s age of inequality. 

They point out that some automation helps workers by improving productivity and freeing them to focus on creative tasks over routine ones. None were willing to say publicly that replacing human workers is their ultimate goal. Terry Gou, the chairman of the Taiwanese electronics manufacturer Foxconn, has said the company plans to replace 80 percent of its workers with robots in the next five to 10 years. In a letter to shareholders last year, Jeff Bezos, Amazon’s chief executive, said that more than 16,000 Amazon warehouse workers had received training in high-demand fields like nursing and aircraft mechanics, with the company covering 95 percent of their expenses. There are plenty of stories of successful reskilling – optimists often cite a program in Kentucky that trained a small group of former coal miners to become computer programmers – but there is little evidence that it works at scale. 

Automating work is a choice, of course, one made harder by the demands of shareholders, but it is still a choice. Even if some degree of unemployment caused by automation is inevitable, these executives can choose how the gains from automation and A.I. are distributed, and whether to give the excess profits they reap as a result to workers, or hoard it for themselves and their shareholders. The choices made by the Davos elite – and the pressure applied on them to act in workers’ interests rather than their own – will determine whether A.I. is used as a tool for increasing productivity or for inflicting pain. 

Now we’re seeing robots that promise a cheap solution to the last link in that chain-getting a single pizza to a home at the right time to the right person. If we want to see robots create economy-scale effects in the short term, we should skip the last mile and look to what I call the last motion. These are the short task sequences between previously automated streams of work in highly structured environments, where humans currently handle various uncertainties involving physical inputs, control actions, and outputs. There are major technical barriers to getting systems like these to economically handle the last motion between two automated processes, but those barriers are much smaller in these structured environments. You need to inexpensively fuse diverse, high-quality, high-frequency sensor data to help robots maintain situational awareness. 

You need advanced software to help robots learn new tasks quickly, adjust to minor surprises, behave predictably for nearby humans, and improve task performance over time. While many of us would gladly accept the increased productivity that came from automating last-motion work, our society does not seem ready for this kind of change. Less educated workers are increasingly finding themselves in these last-motion jobs, picking items out of bins, assembling electronics, moving car parts and materials from line to line, often pressing the equivalent of go and stop buttons on production equipment. Many mature workers who used to handle more complex operator jobs are finding themselves in this kind of work as the processes around them get automated. As much as we may see these jobs as dehumanizing, low-paying, or de-skilling, they are increasingly plentiful, they pay better than retail and other service jobs, and many of us depend on them to provide for our families and make ends meet. 

In a very real way, we have automated a lot of people into these last-motion jobs, and many of them are at the end of their proverbial rope when it comes to work, skill, and making a livable wage. If we want a society where working hard today offers a better tomorrow, we need to do better than that.