In the 2017–18 academic year, 35% of US high schools offered computer science. By 2023–24 that had reached 60%. But the national average hides how differently states prioritize it: 100% of high schools in Arkansas and Maryland offer CS, against 31% in Montana. Participation tells a flatter story — across 41 states with data, 5.1% of high school students took a CS course in 2020–21 and 6.4% did in 2023–24. South Carolina reaches 26%; Florida, Arizona, and Idaho sit at 2%.

Who gets offered the course

• School size is the sharpest divide: 91.2% of large schools offer foundational CS, against 76.4% of medium and just 43.1% of small ones.
• Income matters too. Schools where fewer than 25% of students qualify for free or reduced-price lunch offer CS at 65.0%; where more than 75% qualify, the rate falls to 50.0%.
• Geography cuts both ways — suburban schools lead at 70.1%, with urban (58.2%) and rural (56.1%) close together at the bottom.
• By race and ethnicity, Asian students have the highest access at 91.6%, followed by Native Hawaiian (83.3%), students of two or more races (83.0%), white (82.5%), Hispanic/Latino (80.4%), and Black students (79.7%). Native American students are last at 66.3%.

Who actually enrolls

Comparing CS enrollment against each group's share of the student population produces a ratio where 1.00 means proportional. Asian students sit at 2.60 — more than two and a half times their share. Black students (1.13), Native American/Alaskan students (1.00), and white students (1.00) are at or above parity nationally, though data gaps in nine states warrant caution. Hispanic/Latino students (0.69), Native Hawaiian/Pacific Islander students (0.75), and students of two or more races (0.80) remain underrepresented. So do girls (0.65), English language learners (0.64), students with IEPs (0.67), and economically disadvantaged students (0.72). Students with 504 plans are the one subgroup that is overrepresented, at 1.33.

Advanced coursework grew twelvefold

AP computer science exams went from 19,390 in 2007 to 243,180 in 2023. AP CS Principles, designed to attract a broader class of students, covers some AI content areas even though AP CS A does not. But the growth has not evened out participation: in 2023, 91,216 exams were taken by white students and 69,695 by Asian students, against 43,083 Hispanic/Latino, 16,351 Black, 11,238 multiracial, 801 Native American, and 321 Native Hawaiian/Pacific Islander. Asian students, white boys, and multiracial students are overrepresented among AP CS test-takers; every other group is underrepresented.

81% of CS teachers believe that using AI and learning about AI should be part of a foundational CS learning experience. Fewer than half feel equipped to teach it — 46% in high school, 44% in middle school, and just 34% in elementary school. The gap widens as students get younger, which is the opposite of what an equitable pipeline would need.

They teach it anyway

• Over two-thirds of middle and high school CS teachers say they cover AI specifically, despite the lack of an explicit definition in CS standards; 65% of elementary teachers do the same.
• Of the 2,245 teachers who did spend class time on AI, most spent fewer than five hours per course. Elementary teachers spent the least — 70% spent only one to two hours.
• In a separate 2024 survey of 364 CS teachers, 88% identified a need for more resources for AI-related professional development. Asked what specifically, they named AI literacy: how AI works, how to use it, and its ethical impacts.
• Teachers named their greatest benefits as improved productivity, differentiating student learning, better academic support, and preparing students for the future. Their greatest concerns were misuse tied to academic integrity, AI limiting student learning or engagement, overreliance, misinformation and replicated bias, and student privacy.

The standards have not caught up

The CSTA K–12 standards were last published in 2017 and contain only two standards, both at the advanced high school level, that specifically require AI knowledge. State-adopted K–12 CS standards average 97% coverage of the same subconcepts as the CSTA standards, indicating strong national coherence — but coherence around a framework that predates the generative AI era. Of the 44 states that have adopted K–12 CS standards, 33 have AI-specific standards, generally minimal and focused on high school grades. Four states have adopted more substantial AI-specific standards spanning K–12: Colorado (2024), Florida (2024), Ohio (2022), and Virginia (2024). Arkansas has defined standards for a high school AI and machine learning course.

Federal guidance so far has been about AI in education rather than AI education — the distinction the chapter opens with. The Department of Education's Office of Educational Technology released a series of reports in 2023 and 2024, the most recent in October 2024 offering guidance on safe and effective implementation of AI in K–12 schools. As of January 2025, 26 states have issued guidance on AI in education.

In 2023 US institutions produced 87,435 new computing bachelor's graduates, 52,107 master's, 20,725 associate degrees, and 2,540 PhDs. Bachelor's degrees in computing have grown 22% over the past decade; master's degrees grew 26% between 2022 and 2023 alone, and 83% over the decade. AI-specific programs, tracked under the CIP code created for them in 2016, are much smaller but moving faster: 935 AI master's degrees and 104 AI bachelor's degrees were awarded in 2023. The number of US institutions offering an AI-specific bachelor's degree nearly doubled between 2022 and 2023 to 19; 45 institutions offered an AI master's.

Carnegie Mellon and everyone else

Until recently Carnegie Mellon was one of the only universities offering dedicated AI programs, and it still graduates more AI majors than anyone: 32 bachelor's, 178 master's, and 28 PhDs in 2023, having doubled its bachelor's output. The next AI master's producers are the University of Pennsylvania (98), the University of North Texas (76), Northeastern (55), and San Jose State (52). At the PhD level the entire national list is Carnegie Mellon (28), Capitol Technology University (4), and the University of Pittsburgh (1). Penn State graduated its first AI class in 2022.

Who is in the classroom

• Women earned 22% of computing bachelor's degrees, 23% of associate degrees, 24% of PhDs, and 32% of master's degrees in 2023 — even though women graduate from college at higher rates than men overall.
• Black students earned 8% of computing bachelor's and master's degrees and 7% of PhDs. Hispanic students earned 13% of bachelor's, 8% of master's, and 4% of PhDs. White students earned 46% of bachelor's and 52% of PhDs; Asian students 23% of bachelor's, 28% of master's, and 17% of PhDs.
• Graduate computing programs in the US run on international students. In 2023 nonresidents accounted for 67% of master's graduates and 60% of PhD graduates. International CS master's students more than doubled between 2022 and 2023, from 15,811 to 34,850.
• Two countries dominate that flow. Among international CS master's students in US universities in 2022, India sent 72,020 and China 13,190; among PhD students, China sent 5,130 and India 2,760.

Globally, the US leads and Turkey is the outlier

Using OECD data on ICT graduates — informatics, communication technologies, and computer science — the United States produces more graduates than any other country at every level, and more than twice as many at the associate, master's, and PhD levels as the next country. In 2022 it awarded 116,401 ICT bachelor's degrees against Brazil's 61,760 and Mexico's 32,738; 55,706 master's against the UK's 21,688; and 2,759 PhDs against the UK's 1,156 and Germany's 1,008. On gender parity the ordering inverts. Women average roughly a quarter of ICT graduates at the associate, bachelor's, and PhD levels and closer to a third at master's level — in the US, 24% of bachelor's, 35% of master's, and 26% of PhDs. Turkey is the exception, where women make up at least half of ICT graduates at all four levels.