Editor’s note: This story is part of a series about the Class of 2029 based on a survey conducted by The Chronicle in January and February 2026. You can read more about our methodology and limitations here or read all our survey coverage here.

The Chronicle surveyed the Class of 2029 about their academic habits over their first year at Duke, from their use of artificial intelligence and opinions on the new Trinity curriculum to their academic interests and perceptions of course rigor.

The Chronicle asked survey respondents to indicate the major they are most likely to declare, and nearly two-thirds — 63.20% — indicated they are likely to double major. We grouped intended majors by discipline and found that the natural sciences were the most popular discipline at 59.20%, followed by social sciences at 41.60%, engineering at 21.20% and arts and humanities at 11.60%. Another 2.80% of our respondents said they were likely to complete an interdepartmental major or design their own major through Program II.

We observed variation by academic discipline in survey responses regarding perceived course rigor and the clarity of AI guidelines.

AI: Use or abuse?

The Class of 2029 reported extensive use of generative AI in academic settings, which was defined as artificial intelligence capable of generating text, images or other media. Nearly three in four respondents said they used generative AI at least once a week to aid in completing their assignments.

Students’ AI use seems to be steadily increasing. Last year, three in five students — about 60.27% — reported using AI at least weekly to help with their assignments. The year prior, only about one in five respondents — 21.07% — said the same, though they were surveyed earlier in the academic year, which could have contributed to the disparity.

AI usage was slightly more frequent among respondents enrolled in the Trinity College of Arts & Sciences than it was for students in the Pratt School of Engineering — Duke’s two undergraduate schools. More Trinity students used AI for assignments at least once a week at 75.01% than Pratt students at 72.72%. But the trend flipped for daily users: One in five Pratt students reported using AI for assignments every day, whereas that figure dropped to 15.31% for Trinity students.

Problem sets were by far the most common assignment type where first-years used AI, with nearly three-fourths of our respondents reporting doing so. That’s up significantly from last year, where problem sets also led the pack, but only at 38.84%. Coding tasks and short written assignments like discussion posts tied for second, similar to last year’s results.

We asked our 10% of respondents who never use AI for assignments why they choose not to and offered eight possible explanations. Their top concern was that AI produces inaccurate information, followed by AI use being considered cheating, then that the technology could inhibit their learning experience.

This year, we asked respondents for the first time whether they thought their professors had clearly communicated their expectations about AI use for their classes. Duke has been reshaping its approach to the technology over the past year, most recently with a report from the AI at Duke steering committee advocating for a “strategic orientation” toward the University’s adoption. However, faculty and student opinions on AI in the classroom remain mixed, and instructors have been encouraged to create their own course policies surrounding use of the technology.

Despite Duke’s overarching AI approach remaining in flux, first-years generally thought professors were effective at communicating their expectations for academic AI use. The overwhelming majority of our respondents — 90.80% — said most of their professors had communicated their expectations for AI use in their classes, and over half said all their professors had done so.

Students intending to pursue engineering degrees were less likely to have clear AI guidelines from their professors than their peers in other fields, but the disparity was slight. By contrast, respondents studying the natural sciences or pursuing an interdepartmental major were more likely to have clear AI guidelines for their classes.

FOCUS or Constellations? Most say FOCUS

This year, Duke first-years had to choose between two academic programs, FOCUS and Constellations, for the first time. While the FOCUS program has existed at Duke for decades, the Constellations program was piloted this year as part of the new Trinity curriculum.

Students enrolled in the FOCUS program live in the same dorm and take two thematically tied seminar classes together in the fall semester. The Constellations program, on the other hand, runs throughout the entire year with a set of three interconnected courses and does not include a living-learning element. All Trinity first-years must enroll in one of the two programs.

Our Trinity respondents were relatively evenly split between them, with 50.51% enrolled in Constellations and 49.49% in FOCUS.

Both FOCUS and Constellations have been touted by the University as community-building and academic enrichment opportunities. However, our survey revealed a harsh divergence in student satisfaction across the two programs.

FOCUS students gave much more positive feedback on their experience compared to their Constellation counterparts. The majority of respondents enrolled in FOCUS reported that the program had enriched their academic and social lives, at 76.29% and 81.45%, respectively. Far fewer Constellations respondents said the same — 40.40% and 49.49%, respectively.

If first-years could opt out of these programs, would they? Exactly half of our Trinity respondents said yes.

But again, opinions were largely split along program lines. Nearly three-fourths of Constellations students said they would have opted out, while about the same proportion of FOCUS students said the opposite.

This disparity could be caused in part by selection bias from differences in program structure. FOCUS is a self-selected program for which students must complete an additional application. Thus, students willing to undertake the effort to apply may be more inclined to participate in first-year seminars to begin with. Constellations, on the other hand, are required for all remaining Trinity students and may capture more students who are less interested in seminar-style courses or students for whom the courses fall out of anticipated degree pathways.

Our survey did not ask students about the factors that influenced their feedback, though, so we cannot evaluate this hypothesis with the data alone.

Course rigor: Studying hard or hardly studying?  

Last year, we asked first-year students how the difficulty of their Duke courses in STEM, humanities and social sciences compared to those of their high school. This year, we removed the reference to students’ high school experiences and instead asked respondents to rate the difficulty of their 200-level courses to get more comparative results.

A much higher proportion of our respondents said they hadn’t taken any 200-level humanities or social science courses yet at 66.40% and 73.20%, respectively, than 200-level STEM courses at 33.20%. These students were excluded from the remainder of our analysis of first-year students’ perceptions of course difficulty.

The remaining first-years tended to find humanities courses easier and STEM courses more difficult, and they were more neutral on social science courses. No one found 200-level humanities classes very difficult.

Perceptions of course difficulty in the humanities varied depending on the type of high school attended by students. In general, privately educated students found their college classes easier than their public-school peers, although the difference was starker for social science and humanities classes than for STEM courses.

Students from rural communities tended to find their courses more difficult than those from urban or suburban communities, regardless of academic discipline.

The Class of 2029 applied to Duke in its fifth year of permitting test-optional applications. Our test-optional respondents — who made up roughly a fourth of the sample — generally found their classes to be harder than their peers who submitted the SAT, ACT or both as part of the admissions process.