In my hometown of Palo Alto, a hotbed of technological innovation, the motto of many successful companies is “move fast and break things.” However, this motto could lead to severe consequences down the line. In particular, artificial intelligence (AI), which imbues computers and robotic systems with reasoning and learning skills, and automation, a consequence of increasingly powerful AI and better engineering, could lead to widespread job loss, which in turn could lead to a mass economic downturn.
To many, the risks of advanced AI and widespread automation seem they belong more in a far-future sci-fi story than today’s world. Yet, this opinion could not be further from the truth. Research spanning computer science, economics, and political science has revealed that, not only do automation and AI pose a risk to nearly half of all jobs in the short-term, but that low-paying jobs will be impacted the most.
Imagine, training for years and spending most of your working life in a profession that you’re proud of, then being fired because a computer program or robotic system can do it slightly cheaper. From truck driving to mining to even computer programming, it seems that no job is truly safe, provided we don’t make radical change soon.
This radical change has to come from a societal awakening, that many jobs are at risk, and that automation is a real and tangible threat. For this reason, in this project, I sought to create a musical piece that, while hopefully enjoyable, conveys a narrative subtly, without using words or lyrics. I hope that this method makes the story accessible and increases the wider political will to change.
What are artificial intelligence and automation… really?
Much of modern artificial intelligence research is in “machine learning,” or creating systems that start off inept and untrained, and over time, become smarter and more accurate. These systems are incredible versatile, and, most importantly, need no human input. For instance, a machine can learn to classify species of birds from images alone, to learn to play video games, and even to drive in the real world.
A Waymo self-driving car. While we don’t know exactly what machine learning components are inside the car, self-driving cars are only possible with modern computer science and engineering.
Modern machine learning essentially falls into three categories: supervised learning, unsupervised learning, and reinforcement learning.
In supervised learning, algorithms are given a set of known true data and are asked to extrapolate from this data to new data. One example is speech-to-text: once the machine sees enough examples of speech and the corresponding text, it can extrapolate to new speech.
In unsupervised learning, algorithms are asked to find structure in data. For instance, an algorithm might be able to detect outliers or suspicious behavior when surveilling an area.
Finally, in reinforcement learning, algorithms learn by interacting with a simulated environment. An algorithm may learn to play a game, like chess or Go, or even learn to drive. Importantly, a human is never involved in the machine’s learning process – this means that reinforcement learning could, in theory, scale up indefinitely to any well-defined interactive task. Through reinforcement learning, a machine could figure out how to best allocate resources in a company or find the best strategy to trade stocks.
Similar technologies can be applied to areas you might not think of. For instance, in mining, ore needs to be transported from place to place. Typically, you would need to pay for a human worker, and by hiring a human, the employer faces wage and liability issues.
Enter the self-driving ore carrier. This futuristic technology is already is use – a fleet of 80 are already used in mines in Western Australia. In fact, according to an article in the MIT Technology review, AI and automation will eliminate between 2 to 3 million jobs in vehicle-related technologies alone.
These and many other technologies promise to displace jobs in a wide variety of sectors, and as our technological capabilities continue to grow, so will their ability to displace jobs.
How many jobs are at risk?
Estimates of job risks vary in scope and timeline, but most predict the long-term risk to be extremely disruptive. A report by Frey and Osborne, one of the most cited works in this space, estimated that, because of automation, up to 47% of American jobs could disappear in the next two decades.
Another report by the Boston Consulting Group found that there would be 4 million installations of robots replacing human jobs in the next 10 years.
These figures would not be an issue if more jobs were created by the AIs, however that scenario is unlikely. A report from the Obama administration found that the jobs created by AI and automation would not come close to outweighing the job displacement brought about by these technologies.
As perhaps the strongest indicator of Ai’s potential being realized by the wealthy elite, venture capital investment in AI has also skyrocketed.
Who is impacted and how can this issue shape our communities?
One of the most concerning aspects regarding the rise of AI and automation, and its job-displacing effects is that it disproportionately hits the lower class.
An Obama white house report found that 83% of jobs paying less than $20/hour would be at risk of being automated away, in comparison to 34% of jobs paying more than $20/hour. This presents grave challenges to income inequality and social stability, since lower classes are disproportionately impacted.
Not only will daily life be obviously changed by robots, but the widespread hollowing out of the working class due to automation will be felt worldwide. Families will be impacted, and inequality will increase.
These risks directly connect to the UN Sustainable Development Goals, specifically: no poverty, decent work & economic growth, and industry innovation, & infrastructure. Jobs are, in some ways, a core pillar of the human experience – this is reflected in the impacts listed by the UN.
While the issue affects everyone — if we assume that every job could theoretically, at some point in the far future, be automated, then everyone has to contend with that possibility — the issue is especially relevant to my local community. First, because the tech businesses are attempting to capitalize on AI, local businesses will be affected by the development of the technology, which is shaped in part by policy and social impacts.
But much more importantly, Palo Alto has massive wealth inequality. In a relatively small town, there are millionaires living a stones’ throw away from families living out of rented garages. This inequality will only deepen if drastic action is not taken, since automation both has the potential to destabilize the economy, putting people living paycheck-to-payheck at risk, and directly worsen inequality, by automating jobs. Tragically, some hardship brought on by the issue will have been created by tech entrepreneurs living in this same city.
For these reasons, action must be taken now.
What can we do?
The most important thing we can do right now is raise awareness of the issue and start taking is seriously. Right now, many don’t fully understand the danger that they and their jobs are in, and if we can bring tangible examples to the table, we can convince more people that urgent change is needed.
Beyond raising awareness, we can begin to explore bold new policy proposals. For instance, universal basic income, known as UBI, is a proposal to give each person a fixed amount of money per year, regardless of income. This initiative promises to offer more flexibility than existing welfare programs, since people receive the money regardless of income. This avoids situations when people are disincentivized to earn more, for fear of losing welfare. Regardless of how you feel about these programs, you can help by supporting candidates like Andrew Yang in their efforts to get these issues on the national debate stage.
Other policy proposals include limiting access to AI, regulating AI, and requiring companies to disclose their AI and automation secrets to the government. One more idea is a “robot tax” whose proceeds go to welfare or UBI, therefore reducing the economic incentive to stop hiring people. By writing letters to our representatives, we can take this issue to the top and let politicians know that we care about this issue.
However, to make any of these changes on the national level, we need the political will. These changes must come from the middle and working class, since many business owners may gain in the short-term by replacing humans. This is where the music comes in.
The objective of the music I composed, named A Journey Through Time, is to, as the title suggests, expose the listener to a brief history of music and civilization, and to give the listener a glimpse into the future.
There are no lyrics – this is intentional. Most resources that educate people on the dangers of AI and automation use verbal communication. I hope that my music can reach a different subset of the population that has an auditory learning style.
Without any lyrics, as a classmate pointed out, the song takes on an air relevant to the issue. Lyrics evoke a sense of humanity, as human ears are tuned to listen to voice. The absence of lyrics may create an emptiness, a void of humanity, which is related to the issue of widespread job loss. This absence hopefully strengthens the emotional connection.
The end of the song sees the nature sounds in the background fade out, replaced by metallic clanking. I hope that evokes an emotional response in the listener that can help compel them to action.
Baggaley, Kate. “Robots Are Replacing Humans in the World’s Mines. Here’s Why.” NBCNews.com, NBCUniversal News Group, 21 Dec. 2017, www.nbcnews.com/mach/science/robots-are-replacing-humans-world-s-mines-here-s-why-ncna831631.
Frey, Carl Benedikt, and Michael A. Osborne. The Future of Employment: How Susceptible Are Jobs to Computerisation? University of Oxford, 2013, The Future of Employment: How Susceptible Are Jobs to Computerisation?, www.oxfordmartin.ox.ac.uk/downloads/academic/The_Future_of_Employment.pdf.
Furman , Jason, et al. Artificial Intelligence, Automation, and the Economy. 2016, Artificial Intelligence, Automation, and the Economy, obamawhitehouse.archives.gov/sites/whitehouse.gov/files/documents/Artificial-Intelligence-Automation-Economy.PDF.
Gray, Alex. “These Charts Will Change How You See the Rise of Artificial Intelligence.” World Economic Forum, 18 Dec. 2017, www.weforum.org/agenda/2017/12/charts-artificial-intelligence-ai-index/.
Rotman, David. “Technology and Inequality.” MIT Technology Review, MIT, 21 Oct. 2014, www.technologyreview.com/s/531726/technology-and-inequality/.
Rotman, David. “The Relentless Pace of Automation.” MIT Technology Review, MIT, 13 Feb. 2017, www.technologyreview.com/s/603465/the-relentless-pace-of-automation/.
Salian, Isha. “NVIDIA Blog: Supervised Vs. Unsupervised Learning.” The Official NVIDIA Blog, 20 Sept. 2018, blogs.nvidia.com/blog/2018/08/02/supervised-unsupervised-learning/.
Sirkin, Hal, et al. “How Robots Will Redefine Competitiveness.” Https://Www.bcg.com, 23 Sept. 2015, www.bcg.com/publications/2015/lean-manufacturing-innovation-robots-redefine-competitiveness.aspx.
Waters, Richard, and John Burn-Murdoch. “Waymo Builds Big Lead in Self-Driving Car Testing .” Financial Times, Financial Times, 13 Feb. 2019, www.ft.com/content/7c8e1d02-2ff2-11e9-8744-e7016697f225.