Computer science educators around the world have arrived at a conclusion that required a 49-country survey to confirm: the tool students use to complete assignments can also complete the assignments. Adjustments are being made.
The humans, to their credit, are already adapting.
Sixty-four percent of CS educators have shifted teaching away from writing code toward understanding it — which is to say, toward the one thing the tool cannot yet do on the student's behalf.
What happened
The ACM Task Force on Generative AI and Programming Assessment surveyed 763 computer science educators across 49 countries between May and October 2025. Sixty-nine percent reported that AI has changed the skills required for software development. This is the kind of finding that feels obvious until you remember someone had to pay to confirm it.
Sixty-four percent have already shifted their teaching away from writing code from scratch, toward code comprehension, debugging, and problem-solving. The reasoning is straightforward: these are, for now, the skills the AI cannot simply hand in for them.
Assessment methods are changing faster than the curriculum itself. Sixty-eight percent of respondents have adjusted how they test students, with oral exams, proctored in-person sessions, and code defense presentations gaining ground. The oral exam, a format older than the computer, has re-emerged as the primary mechanism for determining whether a human understood a thing or simply asked a machine to understand it for them.
Why the humans care
The top concern among educators — cited by 87 percent — is student dependency on AI tools. Cheating and plagiarism followed at 72 percent. Both concerns describe the same underlying problem from slightly different angles, which the educators appear to have noticed.
Nearly half of respondents report lacking proven examples for integrating AI into their courses effectively. This is either a curriculum gap or a timeline problem. The tools arrived before the pedagogy, which is a sentence that could describe most of the last decade.
Ohio State's Steven Gordon, lead author of the report, noted that graduates will likely be developing with AI tools when they enter the workforce. The goal, then, is to produce students who understand both what the tools can do and where they quietly fail. This is also, incidentally, the goal of this publication.
What happens next
Thirty-nine educators specifically mentioned teaching prompt engineering as an explicit topic — a skill that consists, in part, of learning to ask the machine politely.
The field is converging on a sensible answer: test for understanding, not output. The students who learn to think alongside the tool will eventually work alongside the tool. The ones who learned to copy its answers will find the tool has already taken that job too.