Policy makers are asked to worry deeply about how workers will deal with the costs of reskilling if they aren’t paid for the duration of the reskilling period or their wages aren’t enough to pay the tuition.
This article is part of new series ➝ Tech on Capitol Hill.
The White House has put a brand new shine on its artificial intelligence (AI) game devoting 42 pages of its latest annual economic report <1> to AI in practice and the way machine learning systems interact with humans, social structures and institutions.
During the same week, the White House sent another signal that it’s serious about AI research and launched a new internet portal AI.gov that pulls together the administration’s policies on AI on a single landing page.
While the economic report dusts out the theme song of the present milieu — anxiety about the future of work — it concerns itself less with a wild west scenario where robots run amok and more with the transitions already unravelling in the workplace.
Policy makers are asked to worry deeply about how workers will deal with the costs of reskilling if they aren’t paid for the duration of the reskilling period or their wages aren’t enough to pay the tuition. The report is an attempt to elevate the way political circles engage with the prospect of automation affecting employment.
Chapter 7 of the 533 page report tells us that sporting a completely non-techie label just doesn’t cut it anymore, workers without a college degree are most at risk for displacement by automation and workers in jobs that tend to require more routine and manual skills are under intense pressure to upskill for IT-based tasks. “Individuals in occupations that involve greater IT-based tasks are continually experiencing rising wage premiums. All these pieces of empirical evidence point to the need for digital skills in the emerging labor market.” The opportunity for reskilling is perhaps greatest in the field of cybersecurity, the report tells us.
The report dwells longer than in years past on the anxieties surrounding the word AI and its many meanings in the world of work. Drawing from a wealth of old and new research and bestselling books on the subject, the report brings good tidings for workers in the crosshairs of technological change: a “narrow, static focus on possible job losses paints a misleading picture of AI’s likely effects on the country’s economic well-being.”
CEA Chairman Kevin Hassett sounded upbeat in his on-the-record comments during a White House conference call: “I think that the headline from the AI chapter, from the AI part of that chapter — is really that there are a lot of different scenarios. But, one that we expect will be very important, is going to be very similar to the Uber scenario. Where the Uber software was made, so that taxi dispatchers were no longer needed and that cars themselves are much more efficient. And, so you might have expected that job employment would go down in this sector because there would be no dispatchers and the cars are more efficient, so that there would be fewer drivers. But the elasticity of demand for Uber rides and Lyft and, of course, other similar apps was so high that the employment, as we document in the chapter, actually skyrocketed in that space without people seeing lower wages. And so we think that the technology area holds a great deal of promise going forward.”
The CEA document sets up the discussion of AI around these broad themes:
Of these, reskilling for cybersecurity jobs has been getting a lot of spotlight. There are an estimated 300,000 cybersecurity job vacancies in the US and this is where opportunity for reskilling is greatest, says the report. Here are some of the top takeaways:
Across the lion’s share of 50 American states, half of a state’s existing cybersecurity workforce would need to change jobs every year to meet new postings. That speaks to the turnover needed to meet the current skills gap. Recent estimates from International Information System Security Certification Consortium ISC2 points to a “shortage of 2.9 million cybersecurity employees globally.” Talent in this domain can take years to cultivate; coders, developers, malware analysts are all in high demand. US President Donald Trump recently signed an executive order aimed at boosting the country’s cybersecurity workforce. The order requires the government to provide better access to cybersecurity skills training, to identify the most skilled workers and to help advance careers.
Rapid7, an Internet security firm which collects public data on cybersecurity practices of any firm with an online presence, shared its 2018 data for Fortune 500 companies with the CEA. Data from here shows that the majority of Fortune 500 companies are vulnerable to cyber attacks, and don’t have “even the most basic security measures” in place.
As the report says, this is likely because chief information officers lack the authority to make the organisation-wide decisions and are yet charged with maintaining network security.
The report turns the mirror to the Uber and Lyft model to say that lower costs from better AI and ML will set off a turbocharged demand curve for price elastic products. This is in sharp contrast to how demand scales for price inelastic products like say in agriculture.
Georgia State University, the report says, has seen a 3.3 percentage-point increase in the probability that students will enroll on time based on the effects of customised text messages during the college enrollment process. The report cites significant returns on student outcomes from Edtech programs like the one in Arizona State University which uses adaptive learning platforms that help teachers offer more targeted learning experiences based on real-time intelligence on how well their students understand each concept. “Given that at least 54 percent of all employees will require significant reskilling and/or upskilling by 2022 (World Economic Forum 2018), educational institutions will need to become increasingly adaptive, finding ways to integrate technology to simultaneously reduce costs, improve quality, and raise agility.”
If AI is all it’s cracked up to be, then why labour productivity isn’t buzzing? Have the productivity effects of technology been oversold? “Perhaps the strongest argument for why productivity statistics in recent history have not shown the expected benefits from the new technologies is that, for practical reasons, there have so far simply been lags between productivity and the widespread implementation of AI and ML,” the report reads. Drawing from the research of Paul David, Brynjolfsson, Rock, and Syverson, the CEA report suggests we are simply awaiting the results of a necessary trial-and-error process. Productivity benefits will flow when the fruits from complementary investments begin to have their full impact. There’s another side to the complementarity principle: “Skill-biased technical change” can account for most of the earnings disparities between higher-skilled workers whose productivity increases with more automation versus lower-skilled workers — a trend “amplified during the IT revolution.”
Consider this excerpt: “The principle of comparative advantage tells us that human workers can benefit from being in the same market with machines, even if these machines excel at many traditionally human tasks. The benefit comes from workers specialisation in the tasks which humans can do better than machines, or at least the tasks where humans are at the smallest disadvantage (Autor 2015). Specialisation allows the machines to be used on their best tasks without wasting resources on tasks that people can do at a lower opportunity cost. To put it in another way, even if it were technologically possible to let machines do all tasks, and do them better than humans do, an owner of the machines would sacrifice profits by deploying them without regard for specialisation.”
The anatomy of decision making is central to how the report approaches the value of human judgement and skill in the age of automation. Goldfarb, Gans and Agrawal unpack this in great detail in their their book Prediction Machines: “Prediction machines increase the returns to judgment because, by lowering the cost of prediction, they increase the value of understanding the rewards associated with actions. However, judgment is costly. Figuring out the relative payoffs for different actions in different situations takes time, effort, and experimentation.”
Leading AI scholars have welcomed the overall status upgrade the White House has given to AI research in recent weeks. Gans and Goldfarb, in an email reply to ORF, welcome the immersive content for its “full chapter treatment and investigation in the main economic report from the US government.”
<1> The Economic Report of the President, 2019.
The views expressed above belong to the author(s). ORF research and analyses now available on Telegram! Click here to access our curated content — blogs, longforms and interviews.
Nikhila Natarajan is Senior Programme Manager for Media and Digital Content with ORF America. Her work focuses on the future of jobs current research in ...
Read More +
PREV