<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Morgan R. Frank on Medium]]></title>
        <description><![CDATA[Stories by Morgan R. Frank on Medium]]></description>
        <link>https://medium.com/@morganfrank_21645?source=rss-4a0776c7d25------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*oPXqSaZicITn0ba7m7B7JA.jpeg</url>
            <title>Stories by Morgan R. Frank on Medium</title>
            <link>https://medium.com/@morganfrank_21645?source=rss-4a0776c7d25------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 00:19:31 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@morganfrank_21645/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The evolution of AI research and the study of its social implications]]></title>
            <link>https://medium.com/mit-media-lab/the-evolution-of-ai-research-and-the-study-of-its-social-implications-4a9598b3d7db?source=rss-4a0776c7d25------2</link>
            <guid isPermaLink="false">https://medium.com/p/4a9598b3d7db</guid>
            <category><![CDATA[research]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Morgan R. Frank]]></dc:creator>
            <pubDate>Fri, 08 Mar 2019 15:18:53 GMT</pubDate>
            <atom:updated>2019-03-12T23:23:27.677Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>This blog post summarizes the key findings in our new article <a href="https://www.nature.com/articles/s42256-019-0024-5"><strong>The evolution of citation graphs in artificial intelligence research</strong></a><strong>,</strong> published in Nature Machine Intelligence. The research team includes myself, Dashun Wang, <a href="https://medium.com/@cebrian">Manuel Cebrian</a>, and <a href="https://medium.com/@iyadrahwan">Iyad Rahwan</a>. We analyze the last 60 years of publications within the booming research area of artificial intelligence (AI). Scientific impact within the AI research community is becoming less diverse with industry research institutions overtaking academic ones as the central hubs for AI research. This observation is concerning because social science research is less likely to make reference to industry-authored AI research. This dynamic may allow the social and societal implications of AI research to go unnoticed prior to deployment.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*DfCYIv6B_uz3Ifqk5VbT1g.jpeg" /><figcaption>While human-level intelligence remains difficult, AI has excelled in traditionally human-areas of cognitive ability including vision, pattern recognition, and language. In addition to new capabilities and productivity, how will the continued deployment of AI impact employment, cooperation, happiness, and society?</figcaption></figure><h3>AI’s history as a narrative tool for social implications</h3><p>Why consider the social implications of artificial intelligence (AI)? Across time and culture, humanity has maintained a mythical concept of artificial <em>human </em>intelligence — although not always in the form of algorithms and machines. Rather, humanity’s quest for AI began with “an ancient wish to forge the gods” that has persisted from mythology, folklore, and science fiction to today’s technology. Yet, both mythological “AI” and technological AI highlight important social questions for society.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*VwVCHB52dx3bbXHsvdP-8g.jpeg" /></figure><p>Allowing the generalized notion of artificial <em>human</em> intelligence, several stories from folklore and mythology use AI to highlight important social and cultural issues. As an example, Hephaestus — the Greek god of metalworking and stone masonry — created mythical golden robots, who could think for themselves but acted as servants on Mount Olympus. Jewish folklore maintains several stories of humanoid golems made from clay. The most famous is the Golem of Prague, who went into a destructive rage after breaking the cultural taboo of working on the Sabbath. In Mary Shelley’s <em>Frankenstein</em>, the monster is shunned by its creator, thus damaging the monster’s social nature. Combined with the monster’s later insistence on his right to a partner, this story acts as a commentary on the ethics of AI and humanity’s paternal obligation to the intelligence it creates.</p><p>While these examples may seem distant from today’s AI, they highlight humanity’s persistent quest to create intelligence and some of the interesting social questions that lie therein. Accordingly, the ethical, social, and societal dynamics around AI adoption contribute to the total impact of AI that we face today. But are social scientists able to keep pace with the recent boom in AI research?</p><h3>Recent AI and its social implications</h3><p>More recent AI has dramatically matured from myths and parables to code and algorithms. Accordingly, AI research has not remained any one thing, and the mimicry of human intelligence has adapted with new technology. Recent AI advances include voice-activated assistants (e.g., Amazon Alexa and Apple’s Siri), autonomous vehicles, and <a href="https://phys.org/news/2017-05-chinese-champion.html">mastery of Go</a> with AlphaGo.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*2TNc46kRojwYkOPcJ8UZ4Q.jpeg" /><figcaption>Autonomous warfare may reduce the human cost of war. How would this alter our willingness to engage in conflict?</figcaption></figure><p>Modern AI adoption is particularly noteworthy considering its implications for the <a href="https://www.youtube.com/watch?v=ikddAAPrwKQ">future of work</a>, the impact of high-frequency trading on the stock market, the practice of medicine with data-driven diagnostics and computer vision, the <a href="https://www.technologyreview.com/s/612341/a-global-ethics-study-aims-to-help-ai-solve-the-self-driving-trolley-problem/">regulation of transportation</a>, the future of <a href="https://www.npr.org/sections/alltechconsidered/2018/04/23/604438311/autonomous-weapons-would-take-warfare-to-a-new-domain-without-humans">autonomous warfare</a>, and the governance of society. On one hand, AI has the positive potential to reduce <a href="https://hbr.org/2018/07/want-less-biased-decisions-use-algorithms">human error and human bias</a>. As examples, AI has <a href="https://www.youtube.com/watch?v=bQkDYGh7wCs">balanced judges towards more equitable bail decisions</a>, AI can assess the <a href="https://www.pnas.org/content/114/29/7571">safety of neighborhoods from images</a>, and AI can improve hiring decisions for board directors while <a href="https://hbr.org/2018/04/research-could-machine-learning-help-companies-select-better-board-directors">reducing gender bias</a>. On the other hand, several recent examples suggest that AI technologies are frequently deployed without understanding the social biases they may possess or the social questions they raise. Consider the recent reports of racial <a href="https://www.ted.com/talks/joy_buolamwini_how_i_m_fighting_bias_in_algorithms">bias in facial recognition software</a>, the ethical dilemmas arisen from autonomous vehicles, and <a href="https://research.stlouisfed.org/publications/economic-synopses/2018/11/19/the-impact-of-automation-on-inequality/">income inequality from computer-driven automation</a>.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FKB9sI9rY3cA%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DKB9sI9rY3cA&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FKB9sI9rY3cA%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/222c3f83d6e6817025cd4f9c64a52135/href">https://medium.com/media/222c3f83d6e6817025cd4f9c64a52135/href</a></iframe><p>These examples highlight the diversity of today’s AI technology and the breadth of its application, an observation that has led some to recently characterize AI as a <em>general purpose technology</em> — much like electricity or running water. That is, AI is cheaply interwoven into the everyday, yielding improvements in everyone’s quality of life. As AI becomes increasingly widespread, researchers and policy makers must balance the economic and social implications of AI adoption.</p><p>Accordingly, we ask:</p><h3>How tightly connected are the social sciences and cutting-edge AI research?</h3><p>We investigate this question through the <a href="https://www.microsoft.com/en-us/research/project/microsoft-academic-graph/">Microsoft Academic Graph</a>, which is “a heterogeneous graph containing scientific publication records, citation relationships between those publications, as well as authors, institutions, journals, conferences, and fields of study.” In particular, we focus on the booming area of AI research and its related fields. Paper production in Computer Science (CS) has grown dramatically from the 1950s to today, placing CS as the fifth most productive academic field in 2017. This growth within CS coincides with a transition in focus from databases, hardware, and data science to AI, machine learning, pattern recognition, computer vision, and natural language processing. Accordingly, we refer to these latter CS subfields as <em>AI and related fields</em>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*-UiAWEG8McnRa7ocByj-EQ.png" /><figcaption>The strength of association between CS subfields from the references of publications in each field. Line and arrow widths correspond to rates of references made between publications in each subfield. Network clustering identifies collections of related research areas within CS. Out of all CS subfields, AI and its related fields have produced the most publications in recent decades.</figcaption></figure><h3>Which fields are important to AI research?</h3><p>External fields reference AI research for a number of reasons. Some fields, such as engineering or medicine, reference AI research because they use AI methods for optimization or data analysis. Other fields, like philosophy, reference AI research because they explore — for example — its moral or ethical consequences for society. Similarly, AI researchers reference other fields, such as mathematics or psychology, because AI research incorporates methods and models from these areas. AI researchers may also cite other fields because they use them as application domains to benchmark AI techniques.</p><p>We measure the strength of association between academic fields according to two measures. First, we consider the share of references made by publications in one field to publications in another field. Although a common metric in bibliometric studies, increasing reference share from other fields toward AI may be biased by the exponential growth of AI paper production over time. Therefore, we also consider a new measure called <em>reference strength</em> that additionally controls for the cumulative number of published papers in the referenced field of study.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*1OuoGbaQszvfdv6bFplgeQ.png" /><figcaption>How AI research makes reference to other academic fields. Fields receiving notably high rates of citations from AI research are highlighted with solid lines.</figcaption></figure><p>Prior to around 1980, AI research made frequent reference to psychology, in addition to CS and mathematics. Controlling for the paper production of the referenced fields, we find that early AI’s references towards philosophy, geography, and art were comparable to the field’s strength of association with mathematics. This suggests that early AI research was shaped by a diverse set of fields. However, soon after 1987, AI research became increasingly computational with a stronger reliance on mathematics and CS.</p><h3>Which fields keep pace with AI research?</h3><p>Unsurprisingly, computer science, which includes all of the AI-related subfields in our analysis, steadily increased its reference share to AI papers throughout the entire period of analysis. Surprisingly, mathematics experienced a notable increase in reference share to AI only after 1980, which marked the first International Conference on Machine Learning (ICML).</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*iytmX1pBckmRrvV3RoO_dQ.png" /><figcaption>Using computer vision to identify buildings from satellite imagery.</figcaption></figure><p>Meanwhile, several fields that are not often cited in today’s AI research played an important role in the field’s development. However, many of these fields did not reciprocate this interest. For example, psychology was relatively important to early AI research, but psychology did not reciprocate this interest at any point from 1990 on-wards. Instead, philosophy, art, engineering (e.g., robotics), and geography (e.g., computer vision for satellite imagery) have increased their reference share to AI papers up to 1995. When we control for AI paper production over time, we observe decreasing reference strength towards AI from all external academic fields. This suggests that other fields have difficulty keeping track of increasing AI paper production in recent decades. In particular, areas of social science — including political science, economics, and sociology — make relatively few references to AI publications relative to the volume of AI research.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*JN9epXoJ277dNNOT2KNOaA.png" /><figcaption>How other academic fields make reference to AI research. Fields exhibiting notably high rates of citations towards AI research are highlighted with solid lines.</figcaption></figure><h3>Why are other fields falling behind? The consolidation of AI research</h3><p>This result may in part be explained by the increased complexity of AI-related research that is not relevant to the study of other scientific disciplines. But the AI research community has also undergone dramatic changes in the last 60 years that may give further insight.</p><p>Firstly, where are AI researchers publishing their research? In general, academic fields have varying preferences for publication in journals versus conferences and for general/interdisciplinary science research versus topic-specific research. Philosophies abound to justify one preference over the other. On one hand, general science publications — which are usually academic journals — have the potential to reach a wider audience but may incentivize research breadth instead of research depth. On the other hand, topic-specific publication venues, such as conferences, can foster competition to advance major problems within the research area without fostering knowledge dissemination across research communities.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*sWhFpkcWJUKvyoa6qXF2QA.jpeg" /><figcaption>The PageRank centrality of AI publication venues over time. Solid lines highlight the most prominent venues for AI publications today, which are almost entirely topic-specific conferences.</figcaption></figure><p>When we examine the references made between AI publications as a citation network, we can identify prominent publication venues using network measures for centrality (i.e., citation PageRank). The prominence of topic-specific AI conferences have been on the rise since around 1990. The one notable exception is the IEEE Transactions on Pattern Analysis and Machine Intelligence (IEEE TPAMI) journal. While these conferences promote tightly knit research communities, they can be exclusionary to scientists from other fields. For example, NeurIPS 2018 (formerly NIPS) — one of the largest AI conferences — <a href="https://www.forbes.com/sites/samshead/2018/09/05/ai-researchers-left-disappointed-as-nips-sells-out-in-under-12-minutes/#6da66b9820e9">sold out of registration spots in under 15 minutes</a>, thus making attendance difficult for active AI researchers — let alone interested scientists from other fields. As a counterpoint, AI researchers frequently make their research available to everyone on the Internet through free-to-read web repositories, such as the <a href="https://arxiv.org/">Arxiv</a>.</p><p>Secondly, who is driving AI research? Publication in academic conferences has been shown to boost the scientific impact of prestigious research institutions. So, is there preferential scientific impact within AI research, and, if so, then who benefits?</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*F-xUh9OyTkEUdFHenGC0zw.jpeg" /><figcaption>Diversity of scientific impact is decreasing within the AI research community.</figcaption></figure><p>In general, we observe a steadily increasing tendency towards preferential referencing within the AI research community. That is, AI research institutions that have already accumulated many citations for their existing AI publications are more likely to receive further citations for their AI research in the future. Preferential referencing enables a “winner take all” dynamic that can impact the diversity of scientific contributions within a field. Accordingly, we find that the diversity of scientific impact within AI is decreasing on aggregate in terms of the number of AI authors, papers per institution, and citations. This trend is rather specific to the AI research community, as most other academic fields actually exhibit increasing diversity of scientific impact across institutions.</p><p>Which institutions are at the center of AI research? Prior to 1990, the most prominent research institutions were academic, including the Massachusetts Institute of Technology, Stanford University, and Carnegie Mellon University, and included only a few industry-based research institutions, such as Bell Labs and IBM. However, the late 1980s again marks a transition point that reshaped the field. While universities dominate scientific progress across all academic fields, industry-based organizations, including Google and Microsoft, are increasingly central to modern AI research while the prominence of academic institutions are on the decline.</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*vuhChXT9wMUHOXT4TAzLSg.jpeg" /><figcaption>Academic institutions are losing prominence as industry-based institutions assume a central role in AI research.</figcaption></figure><p>(Note: the recent rise of Chinese institutions within AI research is notably absent from this long-time analysis. However, the data indeed highlights the growing prominence of China-based AI research in more recent years.)</p><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*VyCgwVh0gMMGsrURHj6NLg.jpeg" /><figcaption>Ranking the prominence of AI research institutions highlights the growing impact of China-based AI research in recent years.</figcaption></figure><h3>What does this mean for AI and social science?</h3><p>It does not look good.</p><p>This transition towards industry is alarming for studying the social and societal dynamics of AI technologies. Social science research is less likely to reference AI publications with authors who have industry-based affiliations. Furthermore, AI research exhibits decreasing reference strength towards social sciences. These observations all suggest this gap between research areas will continue to grow. The fields that study social bias, ethical concerns, and regulatory challenges may be ignorant of new AI technology — especially when deployed in industry.</p><p>While our interpretation of these results is somewhat speculative, we believe our observations may highlight an important dynamic within the AI research community that merits further investigation. These findings may help explain why recent AI technologies have only recently revealed important (and largely unintentional) social consequences, such as racial bias in facial recognition software, the ethical dilemmas arisen from autonomous vehicles, and income inequality in the age of AI. If current trends persist, then it may become increasingly difficult for researchers in any academic field to keep track of cutting-edge AI technology.</p><h3>Closing the gap</h3><p>The gap between social science and AI research means that researchers and policy makers may be ignorant of the social, ethical, and societal implications of new AI systems. While this gap is concerning from a regulatory viewpoint, it also represents an opportunity for researchers. The academic fields that typically inform policy makers on social issues have the opportunity to fill this gap. While our study is a step towards this goal, further work may explicitly quantify the social and societal benefits and consequences of today’s AI technology and identify the mechanisms that limit communication between research domains.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4a9598b3d7db" width="1" height="1" alt=""><hr><p><a href="https://medium.com/mit-media-lab/the-evolution-of-ai-research-and-the-study-of-its-social-implications-4a9598b3d7db">The evolution of AI research and the study of its social implications</a> was originally published in <a href="https://medium.com/mit-media-lab">MIT MEDIA LAB</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Skillscape: How skills affect your job trajectory, and their implications for automation by AI]]></title>
            <link>https://medium.com/mit-media-lab/skillscape-how-skills-affect-your-job-trajectory-and-their-implications-for-automation-by-ai-ecc2e09034a9?source=rss-4a0776c7d25------2</link>
            <guid isPermaLink="false">https://medium.com/p/ecc2e09034a9</guid>
            <category><![CDATA[research]]></category>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[automation]]></category>
            <category><![CDATA[complex-systems]]></category>
            <category><![CDATA[economics]]></category>
            <dc:creator><![CDATA[Morgan R. Frank]]></dc:creator>
            <pubDate>Wed, 18 Jul 2018 18:26:00 GMT</pubDate>
            <atom:updated>2018-07-18T18:42:36.863Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>This blog post summarizes the key findings of our new paper <a href="http://advances.sciencemag.org/content/4/7/eaao6030"><strong>Unpacking the polarization of workplace skills</strong></a><strong> </strong>published in Science Advances by our research team including: Ahmad Alabdulkareem, Lijun Sun, Bedoor Alshebli, <a href="https://medium.com/u/512ae3d471ad">Cesar A. Hidalgo</a> , and <a href="https://medium.com/u/e864b245cbb5">Iyad Rahwan</a>. We propose a new framework for skill complementarity that captures important labor trends, including a worker’s career mobility. Insights from our model have implications for viable worker retraining programs, explaining job polarization, and understanding the impact of automation from AI. Others can explore our analysis and insights by visiting <a href="https://skillscape.mit.edu/">skillscape.mit.edu</a></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vURr5ODzdSH6NyNMWTqSLA.jpeg" /><figcaption>Current trends suggest career advancement is harder for some than for others.</figcaption></figure><p>How do workers move up the corporate ladder and how can they maximize their career mobility? Increased wealth disparity, <a href="http://www.jstor.org/stable/42920623?casa_token=t9gaSRWQZZ4AAAAA:JGx2avD3YkGGz7oAZUEQy_B-53CwruW3_7aghjKUZcYx-4QETLiYhsrfjQiW5AS4vVXdACL0k0KolNui1S354vXd-3xlTykDiG4LPnltHCKvudN1Vbw&amp;seq=1#page_scan_tab_contents">increased job polarization</a>, and <a href="http://science.sciencemag.org/content/356/6336/398.full?casa_token=ktwSro2fjn4AAAAA:htLddvQq6Z7HNikC0bzgm2vx2W0woUJ2ZX3X91FVoWXZOkpUzune4Bf4ggQ6ECrD7f8VdPuP5dxs7Q">decreases in absolute income mobility</a> (i.e. the fraction of children who earn more than their parents) all suggest that upward mobility is difficult for today’s workers. It’s as if the rungs on the ladder to career success are there for some and absent for others. But who is stuck, and why?</p><p>By conventional wisdom, <a href="http://www.jstor.org/stable/2673250?casa_token=_5QOKOgeALMAAAAA:q4BWo2iFjYPVVA7zK8IS5erKsj6zW1Sazk6xuP0PIaxyYtXIoxipYEpdfrg5CDx3mG_m8XegKPm_4YoSOKAMi09XtGnrVpczysieqPhs52oi4YxM1kM&amp;seq=1#page_scan_tab_contents">education</a> determines a worker’s entry into the labor force. Workers who start higher up on the ladder have a better chance of reaching the top. However, <a href="https://www.researchgate.net/profile/David_Autor/publication/227437438_The_Polarization_of_Job_Opportunities_in_the_US_Labor_Market_Implications_for_Employment_and_Earnings/links/00b7d520285b657fd8000000.pdf">returns on higher education have not kept pace with growing costs</a> and mid-career workers are generally unwilling to return to school.</p><p>Instead, most workers utilize their existing knowledge, ability, and skills — along with <a href="https://sociology.stanford.edu/sites/default/files/publications/the_strength_of_weak_ties_and_exch_w-gans.pdf">social connections</a>— to advance their careers. That is, a worker is more likely to fill a job opening if their capabilities meet the job’s requirements. These capabilities more aptly represent the rungs on the corporate ladder, which are present for some and absent for others.</p><p>This theory of skills is not new and <em>skill matching</em> has long been considered a key mechanism in the job matching process (just ask the winners of the <a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.589.7046&amp;rep=rep1&amp;type=pdf">2010 Nobel Prize in Economics</a>). Earlier studies of job polarization and limited career mobility have taken note and focused on different types of labor. For example, <a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.308.299&amp;rep=rep1&amp;type=pdf">Daron Acemoglu and David Autor</a> measure the annual wages of occupations and observe a “hollowing of the middle class,” which they describe as growing employment share for low- and high-skill employment at the expense of middle-skill employment. They argue that high-skill employment leverage <em>cognitive</em> skills, while low-skill employment rely more prominently on <em>physical</em> skills.</p><p>These cognitive and physical labor categories — in addition to traditional measures of education and wage — are very broad. For example, consider that Civil Engineers and Medical Doctors are both professions that fall into the same conventional labor categories; they both have high educational requirements, make high wages, and require cognitive non-routine labor. Yet, their skill sets are largely non-transferable. To explain why Civil Engineers are unlikely to become Medical Doctors — and to explain where skill sets might limit other workers’ career mobility — we need a higher resolution framework for specific workplace skills.</p><p>Our study answers this call for a better resolution into workplace skills. Rather than focusing on broad expertly-derived skill categories, we employ a completely data-driven approach using high-resolution occupational skill surveys carried out by the US Department of Labor. By examining how pairs of skills co-vary in importance across occupations and controlling for ubiquitous skills, we identify pairs of skills with high <em>complementarity</em>. Skill pairs exhibiting complementarity tend to support each other by boosting the productivity of workers who possess both skills, or by the ease of acquiring skills simultaneously. For example, Mathematics and Programming have high complementarity, but Programming and Explosive Strength do not.</p><h3>Constructing the Skillscape.</h3><p>Recall that earlier studies of job polarization measured <strong>wages</strong> but concluded that job polarization is a divide between “high-skill” and “low-skill” employment. So, what are “low” and “high” <strong>skills</strong>? To answer this question, we introduce a new approach for modeling the complementarity of workplace skills.</p><p>Although we are interested in job opportunities, skills are atomic in the labor system, and so , we consider each job title as a bundle of skill requirements. Using survey data from the U.S. Department of Labor, we identify skill pairs that tend to be bundled together. For example, the skill Spatial Orientation and the skill Peripheral Vision are important to many similar job titles:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lbnA-xkoVEzXAS7k8w6Lkw.png" /><figcaption>The jobs that rely most strongly on the skill Spatial Orientation, and the skill Peripheral Vision.</figcaption></figure><p>On the other hand, the skill of Complex Problem Solving supports a very different set of occupations:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*p57Y20N1iKhDNk_Zk_oxEQ.png" /><figcaption>The jobs that rely most strongly on Complex Problem Solving.</figcaption></figure><p>Observations like these lead us to connect skills that support similar jobs while disconnecting skills that support different jobs:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/745/1*ugLYWp2UC3nZKmEcMxnYJQ.png" /><figcaption>Connecting complementary skill pairs.</figcaption></figure><p>By considering every pair of skills in this way, we connect complementary skills to produce a network that we call the Skillscape:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/974/1*VOfiKBo4YA8oNSKDCd8CAw.png" /><figcaption>The Skillscape: a high-resolution network of skills connected by skill complementarity. Skill polarization separates socio-cognitive skills (on the left) from sensory-physical skills (on the right). Skills are colored according to O*NET Skill Category.</figcaption></figure><p>The most striking feature of the Skillscape is the polarization of workplace skills. That is, the aggregate structure of the network separates socio-cognitive skills (e.g. Negotiation and Mathematics) from sensory-physical skills (e.g. Low-Light Vision and Manual Dexterity). We find that occupations relying more strongly on socio-cognitive skills tend to have higher annual salaries, and, similarly for cities with higher median household incomes. This draws a direct connection between the work of earlier studies and suggests that the Skillscape’s sensory-physical skills and socio-cognitive skills <strong>are</strong> the low- and high-skills (respectively) from previous studies. This evidence suggests that skill polarization underlies job polarization.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-LES_G8O4p4OlGiTLbWI2g.jpeg" /><figcaption>Reliance on socio-cognitive skills leads to higher annual wages and wealthier cities. We project individual occupations onto the Skillscape using black circles to identify an occupation’s skill set. Occupation dollar amounts correspond to on-average annual wage for workers in 2015. Similarly, we combine occupational skill requirements with employment distributions to identify the key features of an entire urban workforce and find increasing median household income (dollar amount) relates to increased reliance on socio-cognitive skills.</figcaption></figure><h3>How do workers navigate with skills?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*syt2xCsy_K1WPITFPae6Vw.jpeg" /><figcaption>“Can I get there from here? Am I prepared to take the leap?”</figcaption></figure><p>The Skillscape adds resolution to traditional models by incorporating specific workplace tasks and skills. This improved resolution sheds new light on where bottlenecks limit career mobility due to skill mismatch. So, how can a worker leverage their existing skills to grow their skill set and open-up new career opportunities?</p><p>According to skill matching theory, workers should be able to obtain new jobs if their existing skill set is similar enough to the skill requirements of a job opportunity. Our analysis demonstrates that skill complementarity, which define the links between skills in the Skillscape, accurately predicts the skills of a worker’s new job from the skill requirements of the worker’s previous job. However, this is not the only labor trend we can explain with our framework! The Skillscape also predicts temporal changes to occupational skill requirements and even how the skills of entire urban labor markets evolve over time. This level of insight demonstrates how workplace skills underly the US economy and suggests that our framework has the potential to inform worker retraining programs and urban policy aimed at maintaining employment opportunities in an increasingly competitive economy due to globalization and automation.</p><h3>The consequences of skill polarization</h3><p>While exciting, our results are also concerning. When we combine the skill polarization of the Skillscape with our ability to predict workers’ transitions between jobs, we see precisely how skill matching might create a bottleneck to workers’ career mobility and create job polarization as a result. In fact, when we combine our measure for an occupation’s reliance on socio-cognitive skills, we observe three types of workers: (1) cognitive workers, (2) physical workers, and (3) workers who are <strong>stuck</strong> straddling between the two sets of skills. Socio-cognitive workers have access to many other socio-cognitive job opportunities because their skills are nearby to other socio-cognitive skills. This is because the Skillscape’s socio-cognitive skills form a densely connected network community. Likewise, workers who rely on physical skills also enjoy lateral mobility among similarly physical employment opportunities. However, this <em>lateral</em> mobility is not <em>upward</em> mobility. Occupations with higher annual wages tend to rely more strongly on socio-cognitive skills (see above). This means that physical workers who are looking to climb-up the career ladder must bridge the gap in the Skillscape between the two skill sets. However, workers who attempt this transition actually get stuck according to national employment statistics! Effectively, we find that skill polarization acts as a bottleneck to career mobility.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kvlu1_FAOPUiYIeVJVxWbw.jpeg" /><figcaption>Examples of real worker transitions projected onto the Skillscape. Black circles represent the occupation’s skill set and dollar amounts correspond to on-average annual wage for US workers in 2015.</figcaption></figure><h3>Implications for AI and automation</h3><p>It has long been thought that off-shoring and technological change contribute to growing wealth disparity and job polarization in the US. Although occupations and employment are often the units of interest, these mechanisms actually operate on skills directly. Consider that a specific technology is often narrow in scope (e.g. robotic arm has finite degrees of motion or a particular machine learning algorithm is designed to solve a specific class of problems). This means that each piece of technology actually alters the demand for very specific skills (e.g. the robotic arm diminishes demand for workers with Manual Dexterity). These microscopic perturbations to skill demand accumulate and diffuse throughout the national labor system as macroscopic labor trends including technological unemployment, worker migration, and occupational skill redefinition. We must understand the role of individual workplace tasks and skills in the broader labor system if we are to improve our understanding of off-shoring and automation. Our study is the first step in that direction.</p><p>A worker can climb the career ladder only if enough rungs are in place. Accordingly, our research demonstrates that a worker’s set of skills, knowledge, and abilities directly influences their opportunities for career mobility. And our focus on specific workplace skills could not have come at a better time! As rungs from the ladder are systematically removed by off-shoring and automation, we must continue to improve labor models that account for these microscopic perturbations to skills. The improvements we offer in this study reveal the important role of skill polarization as an underlying mechanism for job polarization. This important fact will help policy makers as they design policy to maintain or grow current employment opportunities in an increasingly polarized economy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/833/1*iBWHoQGaNKpuzhHJSV_wog.jpeg" /><figcaption>Actual footage of the research team constructing the Skillscape.</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ecc2e09034a9" width="1" height="1" alt=""><hr><p><a href="https://medium.com/mit-media-lab/skillscape-how-skills-affect-your-job-trajectory-and-their-implications-for-automation-by-ai-ecc2e09034a9">Skillscape: How skills affect your job trajectory, and their implications for automation by AI</a> was originally published in <a href="https://medium.com/mit-media-lab">MIT MEDIA LAB</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>