The impact of artificial intelligence on the future of work is not yet clear. However, the Rewiring the Job Market series on Pulse of the Heartland explores what is known, giving students, workers and policymakers the information they need to navigate the changing work landscape.

In a previous Pulse of the Heartland piece, Heartland Forward explored how artificial intelligence may reshape job opportunities for those without a bachelor’s degree, highlighting the complex tradeoffs between wages, AI resiliency and job growth. That analysis made one thing clear: avoiding AI altogether is not a practical long-term strategy. Instead, resilience increasingly comes from understanding how AI changes work and how workers can adapt alongside it.

Now, new research from Heartland Forward extends that same framework to jobs that generally require a bachelor’s degree, while incorporating new data that differentiates between the augmentation potential and automation risk for each job. It is still largely unknown how AI will reshape work, as a recent Google report indicates only 3% of companies “have truly transformed with AI.” However, it is important to distill what is known to guide students, workers and policymakers navigating an increasingly AI-powered economy.

A U-Shaped Reality for Degree-Based Jobs

Multiple recent studies have now estimated AI exposure for each job in the US economy. One of the earliest and most notable studies released in 2023 sought to determine AI’s impact on the job market. It did so by understanding what portion of a job’s tasks could not be made more efficient by using large language models (LLMs). That data could then be used to determine the resilience of that job to AI automation—in other words, the jobs with a higher share of tasks that could be automated by AI were deemed less AI resilient. 

The first Rewiring the Job Market piece explored AI resiliency for well-paying jobs that don’t require a bachelor’s degree and are expected to increase in demand over the next 10 years—determining which career paths will require collaboration with AI to stay relevant, as well as those whose share of tasks will be largely AI resilient. 

As a follow-up, this piece explores that same measure of AI resiliency for jobs that require a bachelor’s degree and are expected to increase in demand in the coming decade. In the visualization below, there is a distinctive U-shaped pattern between median annual pay and AI resiliency¹, demonstrating that the lowest- and highest-paying bachelor’s degree jobs tend to be the most resilient to AI automation, while many middle-income roles fall into a more vulnerable zone.

On one end of the curve are roles with relatively lower pay but high human interaction, such as certain education and social service positions—teachers or school counselors, for example. On the other end are highly compensated, deeply specialized roles, particularly in medicine. In between sit many early-career professional jobs and mid-level management positions that rely heavily on tasks AI can already augment or automate, including writing, data analysis and routine decision-making.

This pattern matters deeply for the heartland, where universities and employers are producing thousands of graduates each year for precisely these middle-income professional roles. Jobs such as sales managers, public relations specialists and some early-career tech roles show relatively low AI resiliency. These positions often involve synthesizing information, generating content and managing workflows, all areas where generative AI is advancing rapidly.

However, the way that AI resiliency is measured in this instance does not take into account whether tasks are being fully automated (conducted by AI without human assistance or guidance) or augmented (tasks are accomplished by AI and a human working together). Therefore, it is difficult to fully know what the impact of AI will be on these career paths given this measure of AI resiliency alone. 

Adding a Second Lens: Automation vs. Augmentation

Understanding the need for greater nuance, researchers at Microsoft addressed the distinction between automation and augmentation by separately studying the work activities that individuals augment with Copilot, Microsoft’s LLM and digital assistant tool, and the work activities Copilot implicitly automates when responding to prompts. 

The results were separate measures of automation and augmentation potential for each job, as well as a combination of these two measures that estimates overall applicability of AI to each job – that is, the degree to which AI tools can be successfully utilized to complete work through automation or augmentation. The combined measure is analogous to the AI resiliency metric explored above, but one that is able to draw distinct conclusions given the different data explored. 

To understand how overall applicability of AI to a given job differs from automation potential specifically, the visualization below allows users to toggle between the Microsoft researchers’ combined measure (Overall AI Resiliency Index) and the automation-specific measure (Automation Resiliency Index). These measures do not have a straightforward interpretation, but higher values indicate lower applicability of, or greater resiliency to, AI. 

When viewed using the overall index, the visualization differs notably from the first. The familiar U-shaped pattern largely disappears, driven by wide variation in AI exposure among lower-paying bachelor’s degree jobs. At the same time, many middle manager roles appear relatively resilient under this broader measure. Still, several themes persist: highly specialized medical jobs continue to offer the strongest combination of high wages and resiliency, while early-career writing roles and many non-managerial tech jobs remain among the least resilient.

The clearest insights emerge when switching to the Automation Resiliency Index. Writing-intensive roles see substantially lower resiliency scores, suggesting AI poses greater automation risk than augmentation potential in these occupations. By contrast, many tech roles, such as software developers and computer systems analysts, show higher resiliency under this measure, indicating AI is more likely to enhance productivity than replace workers. Not all tech jobs follow this pattern, however, as data scientists see little change between indices, reflecting continued exposure.

Together, these two views underscore an important takeaway: AI risk is not just about how much exposure a job has, but what kind of exposure it faces. For bachelor’s degree holders across the heartland, long-term career stability will increasingly depend on whether AI is positioned as a substitute or as a tool—and on how well workers are prepared to use it.

While this data paints a vivid picture, a few key points cannot be ignored. First, the Microsoft report only studied Copilot interactions, which may not be representative of AI applications or usage across different AI models. Copilot may be stronger or weaker at certain capacities compared to Google Gemini or ChatGPT, for example. Second, there is no definitive way to measure the applicability of AI to work activities. The Microsoft researchers have taken one approach, but it is still yet to be determined how AI will integrate into the economy and impact jobs. In short, there is a difference between theoretical AI applicability and applicability in practice. 

High Pay, High Resiliency Comes From Specialization

While none of the metrics explored in this piece are perfect measures of AI resiliency, we should not ignore the fact that some very high-paying jobs have high estimated resiliency across all three metrics. These are roles like nurse anesthetists, surgeons and other advanced medical professionals. However, each of these jobs requires years of education, training and credentialing.

AI may assist these professionals, but it seemingly cannot replace the human judgment, accountability and hands-on expertise they provide. These roles highlight a broader truth: the most secure and lucrative careers in an AI-driven economy are often those built on deep specialization and long-term investment in skills. For the heartland, this underscores the importance of strengthening education pipelines in health care and other specialized fields that already anchor many regional economies.

What This Means for the Heartland

Taken together, these findings reinforce a consistent message across Heartland Forward’s research: the future of work is not divided neatly between “safe” and “unsafe” jobs. Instead, it rewards adaptability, specialization and AI fluency.

The heartland is well-positioned for this transition. The region is home to world-class universities, strong health care systems and employers eager to invest in productivity-enhancing technologies. With the right alignment between education and workforce needs, heartland states can turn AI disruption into a competitive advantage.

As with jobs that do not require a bachelor’s degree, the goal of this research is not to rank careers, but to empower people with better information. In an AI-driven economy, informed choices are the foundation of resilience, and the heartland is leading the way in preparing its communities to embrace an AI-driven future.

---

¹Employment forecasts and pay data were obtained from Lightcast.