Overview:

What is this dataset?

The PATHWISE dataset includes various workforce and education metrics for AI and Cyber talent in the United States. The dataset includes these metrics for all U.S. states and core-based statistical areas (CBSAs). You can use ETO’s PATHWISE to explore these metrics for each region.

Which ETO products use it?

This dataset powers ETO's PATHWISE tool.

What are its sources?

PATHWISE is built by consolidating data from the following sources:

• Lightcast job postings data to describe workforce demand
• Lightcast worker profiles data to describe employment
• Integrated Postsecondary Education Data System(IPEDS) data from the U.S. Department of Education's National Center for Education Statistics (NCES) to describe education metrics.

What are its main limitations?

• Education levels do not include non-degree award programs. PATHWISE’s current iteration focuses on associate’s, bachelor’s, master’s and doctoral degrees.
• Education metrics lag behind the workforce metrics. PATHWISE is built using education data from 2024, and workforce data from January 2025 to April 2026. We aim to reduce the time lag between these sources as new data will be added regularly.
• PATHWISE inherits additional limitations from its source datasets, including:
  • Workforce demand only indicates the number of job postings. We do not have any information on whether a given posting was filled or not.
  • Workforce data largely relies on online activity. The underlying data for all workforce variables are primarily based on online job postings and profiles, which may exclude postings and/or profiles that don’t have a digital presence.

What are the terms of use?

Because this dataset incorporates licensed data from commercial providers, it is not publicly available. However, you can view most of the data in it using PATHWISE.

How do I cite it?

Because the dataset is not publicly available, you should cite PATHWISE or this documentation page instead. To cite PATHWISE, please use "CSET Emerging Technology Observatory PATHWISE", including the link.

Structure and content:

The basic unit of the PATHWISE dataset is the geographic region, which is either a state (including the District of Columbia) or a Core-Based Statistical Area (CBSA). CBSAs, as defined by the U.S. Office of Management and Budget, consist of counties (or county-equivalents) that contain a core area with a substantial population nucleus along with adjacent communities having significant economic and social integration with the core area. For each region, we calculate the workforce and education metrics using our methodology. Read More >>

Note: Variables names are slightly modified in this documentation for clarity

Methodology:

The following methods were performed in the given order, to obtain the final dataset for PATHWISE.

Identifying Emerging Technology Demand:

We developed machine learning classifiers to identify AI and cyber job postings to measure AI and cyber workforce demand. Existing methodologies (including prior CSET workforce research ) captured a broad set of AI and cyber related jobs, but often missed cutting-edge technical roles that didn’t fit into established occupations or skills buckets. To address this limitation, we trained two classifiers and ran them on Lightcast job postings data. One classifier identified AI relevant postings according to a CSET definition, and the other identified cyber relevant postings according to the NICE Workforce Framework.

More information about each classifier’s definition are as follows:

CSET defines AI development jobs as roles that require specialized knowledge, skills, and abilities (KSAs) and that directly contribute to the technical development of AI systems––for example, systems built around machine learning models, including natural language processing (NLP), computer vision, and large language models (LLMs). We specifically focus on AI development jobs to identify the workforce driving the design, training, and deployment of AI systems.

CSET defines cyber jobs by adopting the NICE Workforce Framework, which divides the cyber workforce into 5 work role categories. These include technical cybersecurity roles as well as analytic and governance roles like privacy compliance and cybersecurity policy and planning. We use this definition as it is a standardized framework used across government agencies.

For PATHWISE, we grouped each of the classifiers’ results by U.S. States and CBSAs to identify demand for all regions.

Identifying Emerging Technology Employment:

We used the result of the classifiers to measure AI and cyber employment. In line with previous workforce research, we assume that the share of AI and cyber job postings within each occupation is similar to the share of current AI and cyber employment in that occupation. Accordingly, we calculated occupation-level shares of AI and cyber postings across the U.S., and multiplied them by the total number of employees within that occupation in each U.S. State and CBSA. We then summed the number of AI and cyber employees across all occupations to obtain total AI and cyber employment for each region.

Identifying AI and Cyber Educational Programs:

We analyzed the post-secondary education requirements cited in identified emerging technology talent job postings. In particular, we focused on the Classification of Instructional Programs (CIP) codes, a standardized taxonomy defined by the NCES to identify instructional program specialties within educational institutions. We used the 6 digit CIP code as they are the most detailed level for instructional programs.

We tested the following three approaches to identify the most relevant CIP codes for AI and cyber talent:

• The total number of AI and cyber job postings associated with each CIP code
• The share of AI and cyber job postings within each CIP code
• The change in the number of AI and cyber job postings between each CIP code when ranked by their total frequencies.

All three methods resulted in the same set of top CIP codes for AI and cyber job postings. Based on significant drops in their frequencies, we include the top 6 CIPs for AI and top 5 CIPs for cyber talent.

Identifying federal government demand:

We identify federal government job postings within the AI and Cyber job postings subsets by tagging postings whose source is usajobs.gov.

Identifying top 5 Occupations:

We aggregate the AI and Cyber postings dataset by the specialized occupation group in Lightcast’s occupation taxonomy, as it offers more granularity for technical roles than the federal SOC (Standard Occupational Classification). We then identify the top 5 occupations by frequency. The same process is repeated for the AI and Cyber profiles.

Identifying number of graduates:

We count all graduates whose major degree corresponded to our AI or Cyber CIP codes and group them by degree level. We then map these results to the physical location of each educational institution, excluding educational programs or universities that operate fully online.

Final Consolidation:

We merge our job posting and profiles dataset with the number of AI and Cyber graduates for our geographic regions.

Maintenance:

How are the data updated?

The data are updated intermittently, depending on the data releases from Lightcast and NCES IPEDS.

Credits:

• Data collection and analysis: Jacob Feldgoise, Sonali Subbu Rathinam
• Engineering: Jacob Feldgoise, Sonali Subbu Rathinam
• Review: Jacob Feldgoise, Katherine Quinn
• Documentation: Sonali Subbu Rathinam

Early versions of PATHWISE were supported in part by NobleReach Foundation.

Major change log

2025-10-30	Initial release
2026-06-16	Data update with classifier results and addition of associate degrees