emergency pedagogical – Programming instructors are scrambling to reshape assignments and assessments as GenAI becomes part of everyday coding tools—often without training or resources.
Programming education is entering a messy new phase, one that doesn’t look like a planned curriculum upgrade so much as an urgent, ongoing fix.
For many instructors. generative AI isn’t a distant concept anymore—it’s woven into what students already use for web search. writing. and coding.. Yet even after ChatGPT has been publicly available for years. a key question remains: how many programming teachers have actually redesigned the way students learn and get assessed. not just updated course policies?. Misryoum’s reporting on emerging research into faculty responses points to an unsettling gap—welcome openness among instructors. but slower. more uneven changes to assignments. assessments. and day-to-day teaching.
Misryoum describes this new mode as “emergency pedagogical design,” a label borrowed from “emergency remote teaching” during COVID-19.. The analogy fits: instructors are reacting in real time. without a playbook. and often without the ability to directly control the AI tools students rely on.. Unlike a product interface that designers can adjust, instructors can’t modify ChatGPT or Copilot themselves.. Their influence has to travel indirectly through policies, assignment structure, course infrastructure, and the feedback loop built into teaching.
The research behind this framing looked closely at what happens when instructors go beyond “allowed vs.. not allowed” rules.. Misryoum highlights how interviews with undergraduate computing instructors revealed a set of constraints that shape their choices.. The work tends to be reactive—courses were built before GenAI became mainstream, so updates are retrofits.. It’s also indirect: instructors can’t change the AI’s underlying behavior. only the context in which students use it.. Instead of relying on controlled evaluations. many instructors depend on ambient evidence—what students say in office hours. what staff hear repeatedly. and what patterns show up informally.. Finally. there’s a time pressure: waiting for research or best practices often isn’t realistic when student behavior is already changing.
A major theme cutting through these accounts is fragmentation.. Misryoum reports that many instructors personally support GenAI adoption in teaching, but departmental or colleague buy-in lags sharply.. That gap creates uneven experiences for students—one class may treat GenAI as permissible. another may restrict it. and students are left navigating what feels like a “wild west” of rules.. Even when policies exist. they can blur important distinctions. such as whether a tool is paid or free. or whether AI is used through a standalone chatbot versus embedded functionality inside a code editor.
That confusion isn’t just frustrating; it risks widening inequality.. Misryoum notes that unequal access to paid GenAI tools can translate into uneven learning outcomes.. Students with better resources can experiment more. iterate faster. or receive more assistance—while others fall back to partial solutions or slower workflows.. This matters especially in computing education, where practice and feedback are tightly linked to skill development.
The biggest practical challenge may be assessment misfit.. Many instructors report a pattern where students perform relatively well on take-home assignments yet struggle on proctored or closely monitored demonstrations.. Misryoum’s analysis suggests this mismatch isn’t simply about “cheating” versus “not cheating.” It reflects a measurement problem: if instructors can’t reliably observe how students used AI during the learning process. grades can start to reflect access to tools and prompt-wrangling more than the underlying skill they’re trying to teach.
Some instructors respond by shifting evaluation toward oral explanations—stand-up style check-ins where students talk through their choices—or toward written reasoning that requires students to explain how they arrived at solutions.. Misryoum sees the trade-off clearly: these methods may better surface understanding. but they can also create new burdens around staffing. consistency. and grading load.. What helps students demonstrate learning can simultaneously make assessment management harder for instructors.
Behind all of these adaptations is a resource crunch.. Misryoum reports that many faculty cite insufficient resources and time constraints as the barriers that prevent GenAI integration from turning into thoughtful. scalable redesign.. The strain is especially acute at minority-serving institutions, where teaching loads and resource gaps can be larger.. In contrast. the instructors who managed the most ambitious course overhauls often had advantages—lighter schedules. external funding. or the ability to hire substantial course support.
This is where the equity risk becomes more than theoretical.. Misryoum frames the core concern as scalability: if only well-resourced institutions can afford to redesign curricula and build supporting tools. GenAI doesn’t just change learning—it can intensify existing gaps.. Students at under-resourced schools may fall further behind. not due to lack of student motivation or instructor care. but because redesign work becomes too expensive in time and labor.. In effect, curriculum “adaptation” can become a privilege.
What would make emergency pedagogical design sustainable?. Misryoum points to the faculty priorities emerging from the same body of research: faculty training. evidence about GenAI’s actual impact on learning. and dedicated funding.. The more productive direction may be to treat GenAI integration as infrastructure—something universities and partners plan for collectively. rather than something each instructor improvises alone.
For programming instructors. the near-term challenge is to keep learning objectives intact while changing the routes students take to reach solutions.. Misryoum’s editorial takeaway is that the conversation can’t stop at policies.. The real work is redesigning assessments to measure understanding under new tool realities. and doing so with enough support that the next semester doesn’t become another scramble.