Marcia Hultman

Cabinet Secretary

Font Size: A A A

South Dakota e-Labor Bulletin

December 2023

Is South Dakota ready for AI, the next major technology trend?

In last month's Labor Bulletin article, we talked about AI and how it may or may not impact our labor market. This month we will continue exploring AI and what the labor force of the future might look like because of it.

Last month, we left off with the following statement and subsequent questions:

Some economists predict 40-45% of the jobs currently performed by humans could be replaced by AI robotics. Some economists predict higher.

  • Could these predictions be true?
  • Could AI bots really replace or displace many jobs currently held by humans? What happens then?
  • Where are we headed?

All are good questions, and as fast as AI has been appearing in daily life, fair questions and concerns. The truth of the matter few solid answers exist. There have been studies conducted and some of them merited review in the congressional report mentioned earlier in this article. So, let’s take a look at some of the scholars’ findings and subsequent predictions.

Assessing the Impact of New Technologies on the Labor Market

To begin answering this question, one study combined a subjective assessment with an objective source of information to measure the impact AI may have on job replacement. The objective source used was the skills requirement portion of the O*Net. The O*Net is a free online Occupational Information Network database containing hundreds of job definitions to help students, job seekers, businesses, and workforce development professionals understand today’s world of work in the United States. The O*Net was developed under the sponsorship of the U.S. Department of Labor, Employment and Training Administration, through a grant to the North Carolina Employment Security Commission during the 1990s. This study consisted of expert categorization of a subset of occupations (70 of 702 O*Net occupations) by participants in a machine learning conference at Oxford University. Each participant was asked to rate an occupation as automatable based on the answer to this question: “Can the tasks of this job be sufficiently specified, conditional on the availability of big data, to be performed by state-of-the-art computer-controlled equipment?” Scientists studied how capable AI would be of mimicking the required skills to perform certain occupations.

The binary answers to these questions were then modeled using the skills portion of O*NET. The best-fitting models were then used to calculate an automatable score for all 702 occupations, using the features of jobs that best predicted automation as assessed by the experts. They classified occupations as high-risk if the estimated probability of automation is 70% or higher and low-risk if it is under 30%. This exercise led to the conclusion 47% of U.S. jobs are at high risk of automation within the next two decades. They found many jobs in office and administrative support, transportation, and services are at risk, even though many of the skills necessary to be successful in these types of occupations are not typically considered routine. In the past, tasks repetitive and mundane in nature have been the jobs typically targeted for automation. However, this and subsequent studies seem to indicate AI, in contrast with previous new technologies like software and robots, is directed at high-skill tasks. This research suggests highly skilled workers may be displaced at a higher rate, given the current rate of adoption of AI.

However, this study did not explicitly model the relative costs of capital versus labor, nor did it consider AI may only partially automate a job. Furthermore, the study also did not consider the research and development costs of these potential applications. Thus, some economists believe the results captured in this study were not a measure of what is economically feasible, but rather an estimate of what is technologically feasible.

A later study refined the 47% statistic by using the aforementioned study results and applying them to a larger group of estimates from developed countries. The new study utilized the occupation results of the previous study and calculated the probability of automation based on the underlying characteristics of the worker and his or her job. This new study allowed job tasks within the same occupational category to vary and have independent effects on the probability of automation. This approach acknowledged two important things: occupations contain multiple tasks, and even within the same occupation, workers do not perform exactly the same functions at the same level of complexity. The results of this study showed jobs involving more complex tasks are less automatable. In particular those jobs involving tasks such as influencing, reading, writing, and computer programming are harder to automate.

Moreover, the study also concluded human capital—measured by education level, experience, and cognitive ability—lowers the risk of working in an occupation deemed replaceable. Their final estimate, which they caution likely overstates the actual likelihood of automation, predicts about 9% of workers in the United States and other developed nations will face a high risk of losing jobs to automation within the next decade or two. This study goes on to state it is likely to be an overestimate because they did not consider the slow pace of technological adoption, nor the economic incentives for companies to produce or adopt the technology. Most companies will not lean on technology or change their way of doing business without cause. It must prove to be a cheaper way of doing business and/or a more effective and efficient way of doing business before they will lay out the capital investment to make advancements in how their business functions.

Nedelkoska and Quintini, economists studying the effects of automation on jobs, followed the above methods closely—but allowed tasks to vary based on classification—instead of the more general set of occupational characteristics used in previous studies. They concluded 14% of jobs are at risk of automation across 32 countries. However, their study emphasized roughly half of the jobs across these same countries could be affected by automation, as some aspects of the job would be likely to be changed.

Other relevant research in this area looks at how the skills of occupations have changed in relationship to technology. After analyzing changes in the important job tasks across occupations it was concluded skills which complement technology (e.g., familiarity with equipment) are increasing, as is the demand for skills which do not currently compete with machines (e.g., interpersonal). On the other hand, skills which do compete with machines (e.g., manual and perception) are declining, in that workers were less likely to report these skills as important in 2014 relative to 2006 within the same occupation group. This means job tasks are gradually drifting away from work that competes with machines, and these are more likely to be jobs requiring less formal education.

After reviewing the latest trends in technology across various industries, most economists predict technology will primarily affect lower-wage jobs requiring modest levels of education and training. But there are also those economists who believe those employed in high-skilled, high-wage jobs will also face competition from machines increasingly able to perform cognitive tasks such as writing, data analysis, and problem-solving.

Economists have wrestled with the implications of technological change from the beginning of the Industrial Revolution to the present, and yet no theoretical consensus has emerged as to what effects these new and emerging technologies will have on the labor market. The recognition of advances in AI and related technologies has suggested perception, creativity, and social intelligence are important cross-cutting characteristics of tasks which may or may not become targeted by AI bots in the future. If indeed it is concluded machines can perform these tasks successfully and it is financially feasible, perhaps more occupations could be impacted by machines capable of learning skills which to date have been reserved for and by humans.

Will AI ‘take’ without giving, or will some jobs or aspects of them lead to new jobs? What types of jobs could be created or become in higher demand as AI continues to advance?

It is predicted occupations with low to medium digitalization skill requirements will become more digitally intensive, suggesting a shift toward the need for increased digital literacy. For example, rather than writing down customer meal orders, more waiters and waitresses may be required to use electronic pads. The order is sent electronically to the chef in the kitchen, including any special requests for preparation or substitutions. After having served the meal, the wait staff presents the customer with an electronic pad with their charges and payment options rather than a handwritten bill. Once the customer approves their charges and provides electronic payment, the transaction is complete.

The demand for workers with high levels of skill in digital technologies has, and likely will continue to, increase. The U.S. Department of Education, BLS, and economists all predict further growth in science, technology, engineering, and math (STEM) jobs. In recent years STEM occupations have experienced fast employment growth both nationally and in South Dakota. Workers with these in-demand skills will continue to find good pay and high demand for their talents.

What emerging jobs will be created to keep up with advancements in AI?

Within technology-producing firms, there may also emerge a demand for entirely new professions. Specifically, new work will require professionals capable of training AI systems to perform intelligent tasks, such as teaching natural language processors and language translators, teaching customer service chatbots to mimic and detect the subtleties and complexities of human communication, and teaching AI systems (e.g., Siri and Alexa) to show compassion and to understand humor and sarcasm. There will also be a need for professionals who maintain and sustain AI systems to ensure they are operating as intended and to address unintended consequences.

In fact, this need exists today; currently, less than one-third of companies are confident in the fairness and transparency of their AI systems, and less than half are confident in the safety of their systems. Also in demand will be workers who can bridge the gap between high-tech professionals and the technologies they create, and businesspeople and consumers to help explain and provide clarity about AI systems. Companies will need data experts, such as data protection officers, to protect privacy rights and related issues; in other words, companies will require professionals who can communicate technical details to non-technical professionals and consumers.

What jobs are at risk of technological displacement?

The Industrial Revolution was beneficial to workers and their standard of living. Machines meant they could do and produce more with less effort and in a shorter amount of time. Workers once employed in agriculture were able to move into other sectors of the economy as agricultural production and food processing became more mechanized and efficient. These workers allowed other portions of the economy to grow and flourish.

Yet, this history is not guaranteed to repeat itself, especially if newer technologies have fundamentally different characteristics. For example, the power of automation to perform complex cognitive tasks—via AI—distinguishes it from automated technologies of the Industrial Revolution. Likewise, digital technologies can almost instantly transmit data anywhere in the world from any place, which is not something pre-digital technologies could do. Possible implications involve a reduction of demand for human monitoring and control activities.

The direct effects of technology on labor demand have resulted in ambiguous results. Some studies, particularly those looking at industrial machines, suggest as AI technology becomes more commonplace it could lead to fewer jobs in this industry. On the other hand, studies looking at the adoption of information or AI technology tend to lead to greater labor demand. This means the industry will still hire workers, but these workers must be trained to perform different functions than before. This difference suggests either technology has different effects on labor in different sectors of the economy, or unmeasured sector-level factors are the driving force behind either the adoption of technology or the demand for labor exhibited by those sectors.

The full effect of technology (how productivity advances related to technological creation or innovation) and the adoption of this technology theoretically should generate gains which will flow back into the economy and create demand for labor as well as advance living standards. In the past, this theory was proven as being true, and technology raised both living standards and the demand for labor. However, many scholars are not convinced AI will have the same effect.

During the Industrial Revolution, many scholars point to a reduction in the demand for craft and skilled workers while others argue the mechanization actually increased demand for intangible and informal skills. However, most economists suggest workers with lower levels of education, or whose occupations require the performance of largely routine tasks, have historically and will continue to be the ones experiencing the greatest threat of job displacement. Perhaps this is because routine work is often repetitive and therefore easily taught to and able to be replicated by machines and computer applications. Therefore, it is anticipated these same types of jobs will be at highest risk over the next 10 to 20 years in terms of experiencing lower wage growth and perhaps eventual displacement.

However, with respect to AI, it is unclear what tasks new machines can perform and what pressure that could place on occupations and workers with higher levels of education. Concerns about AI’s technological “deskilling effect” could, even if it is a small number of occupations considered “skilled” by today’s standards, lead to displacement of workers on a large scale. A few researchers feel this new reality will happen and is likely to happen soon.

BLS cited a challenge for future research entails the ability to further refine the characteristics of workers and the jobs most likely to see robust technological change will require tracking to better understand the potential impact of AI. In the Assessing the Impact of New Technologies on the Labor Market congressional report BLS indicates in order to push the empirical and theoretical literature forward, scholars will need granular data on skills, tasks, and capital investments. The need for new types of data sources is especially apparent when attempting to forecast the effects of AI and other advanced automation technologies.

So, where are we headed and how will the workforce be impacted?

Nobody can say with a 100% certainty how significantly AI will impact our workforce in the future. One thing we can say with complete certainty, however; it will have a significant impact. How quickly it will impact our workforce is likely to be blurred for a while. But we can use the past as a metric to help us contemplate what may occur and at what speed. Regardless of where the theoretical literature leads in settling various issues such as technology-skill complementarity, technological developments, and market dynamics, one thing is certain. None of these dynamics have ever operated independently of political forces. Among other policies, mass public education, training, trade, immigration, and public support for research and development have and will continue to affect the development of new technologies and the capacity of entrepreneurs and workers to use them.

This will not change. In South Dakota we are seeing government leaders take the lead as we enter an AI-driven world. Dakota State University is one of only 10 post-secondary institutes in the United States that can boast of receiving cyber operations, defense, and research designation from the National Security Agency and Homeland Security. These training programs have been part of Governor Kristi Noem’s vision for South Dakota.

In an article entitled “Cybersecurity: South Dakota’s Next Big Industry,” Governor Noem wrote of steps already taken to grow South Dakota’s footprint in cyber security and steps she envisions going forward. Perhaps the best and most hopeful message we can leave our readers with is a reprint of Governor Noem’s article, in which she shared her vision of South Dakota’s future labor force and how we could lead the nation and perhaps the world into unchartered AI territory. By using common sense, foresight, and good old-fashioned South Dakota preparedness, she believes together we will produce a labor force prepared to tackle this new labor market head-on. She sees a strong state filled with South Dakota-trained cyber scientists and sleuths forging the way in the AI world. As you read or re-read the article while reflecting on the topics we’ve covered in these Labor Bulletin articles, we think Governor Noem’s vision will provide both reassurance and hope for a South Dakota with AI.

Cybersecurity: South Dakota’s Next Big Industry

By Governor Kristi Noem

September 22, 2023

When I first ran for governor, I promised the people of South Dakota that I would bring the next big industry to our state. We want to keep our kids and grandkids living and working in South Dakota, so we’re focused on bringing the jobs of the future to keep them here. Careers in technology and innovation are the future. And South Dakota has the opportunity to build a future for our kids and grandkids.

Cybersecurity is South Dakota’s next big industry. Our fastest growing sector is “Scientific, and Technical Services,” with 4,000 jobs added in the past five years – a growth of 25%.

South Dakota already has one of the top universities in the nation for cybersecurity and emerging technology. Dakota State University has received designations in cyber operations, cyber defense, and cyber research from the NSA and Homeland Security. There are only ten institutions in the entire country that can say that!

DSU was also the first school in America to create a Ph.D. program in cyber operations. And earlier this year, Dakota State University established an Educational Partnership Agreement with the U.S. Army Cyber Command. This partnership allows students to experience work and educational opportunities at the top classification level. In fact, DSU was the only school in America to send four students to compete for Team USA at the World Cyber Games.

South Dakota is in the middle of the country – and we’re landlocked, so foreign spy ships and subs can’t reach us. It makes a lot of sense for cybersecurity resources to be centered here.

Dakota State’s “MadLabs” is the heart of this. It is the key facility in Madison, and we’ve built partnerships with private industry to expand it to Sioux Falls. They are doing incredible work to drive innovation throughout both the South Dakota economy and the entire nation. They focus on cybersecurity, AI, machine learning, digital forensics, and so much more.

We are also training kids even before they get to college. The Governor’s Cyber Academy provides opportunities for high school students to get some of this training early. Students can attend the academy to get dual-credit training opportunities during the summer. They earn digital badges as they learn skills that will help them get in-demand jobs in the future.

Dakota State University is giving students the kind of career opportunities that they can’t get anywhere else. These opportunities not only encourage our kids to stay in South Dakota to earn their college degree, they also put our state on the map and attract students from across America to choose South Dakota for their education.

I ran for governor to make sure that the next generation would inherit an America that they could be proud of. Accomplishing that goal starts with education.

I recently had the opportunity to be a member of a panel on the future of U.S. regional innovation at the Global Emerging Technology Summit. During this discussion, I got to tell South Dakota’s story and how we are making real progress right here in the heartland. I am so proud that South Dakota schools understand the importance of educating our students in the fields of the future – and I can’t wait to continue our hard work.